/ais/ - Artificial Intelligence Tools

"In the Future, Entertainment will be Randomly Generated" - some Christian Zucchini

Index Catalog Archive Bottom Refresh
Name
Options
Subject
Message

Max message length: 12000

files

Max file size: 32.00 MB

Total max file size: 50.00 MB

Max files: 5

Supported file types: GIF, JPG, PNG, WebM, OGG, and more

E-mail
Password

(used to delete files and posts)

Misc

Remember to follow the Rules

The backup domains are located at 8chan.se and 8chan.cc. TOR access can be found here, or you can access the TOR portal from the clearnet at Redchannit 3.0.

Uncommon Time Winter Stream

Interboard /christmas/ Event has Begun!
Come celebrate Christmas with us here


8chan.moe is a hobby project with no affiliation whatsoever to the administration of any other "8chan" site, past or present.

Use this board to discuss anything about the current and future state of AI and Neural Network based tools, and to creatively express yourself with them. For more technical questions, also consider visiting our sister board about Technology

AI Art / Stable Diffusion Thread #6 Anonymous 10/07/2023 (Sat) 22:18:55 Id: 243d38 No. 2422
Copied from previous OP (OP is now fairly outdated, needs to be refreshed at some point) Here's some guides to installing Stable Diffusion (if you have a NVIDIA GPU with 2GB+ VRAM). >Installing AUTOMATIC1111, the most feature-rich and commonly-used UI: https://rentry.org/voldy >Different UI with 1-click installation: https://github.com/cmdr2/stable-diffusion-ui >Installing various UIs in Docker: https://github.com/AbdBarho/stable-diffusion-webui-docker >AMD isn't well-supported. If you're running Linux there's a way to install AUTOMATIC1111 but otherwise features are very limited: Native: https://rentry.org/sd-nativeisekaitoo Docker: https://rentry.org/sdamd Onnx: https://rentry.org/ayymd-stable-diffustion-v1_4-guide To try it out without installing anything you can use a web service, though they are lacking in features. >Stable Horde is a free network of people donating their GPU time, how long it takes depends on how many people are using it. It supports negative prompts, seperate the positive and negative prompts with "###": https://aqualxx.github.io/stable-ui/ >Dreamstudio requires making an account and gives you around 200 free images before asking you to pay: https://beta.dreamstudio.ai/ Img2img: https://huggingface.co/spaces/huggingface/diffuse-the-rest | https://dezgo.com/image2image Ippainting: https://huggingface.co/spaces/fffiloni/stable-diffusion-inpainting | https://inpainter.vercel.app/paint >The most notable non-Stable Diffusion generator is Midjourney, which tends to be nicer-looking and doesn't require as much fiddling with prompts but can have a samey style. However the only way to use it is through a Discord bot. https://www.midjourney.com/home/ >If you're okay with more setup but don't have the hardware you can use a cloud-hosted install: Paperspace: https://rentry.org/865dy Colab: https://colab.research.google.com/drive/1kw3egmSn-KgWsikYvOMjJkVDsPLjEMzl >Various other guides: NovelAi: https://rentry.org/sdg_FAQ Dreambooth: https://github.com/ShivamShrirao/diffusers/tree/main/examples/dreambooth Inpainting/Outpainting: https://rentry.org/drfar Upscaling images: https://rentry.org/sdupscale Textual inversion: https://rentry.org/textard >Resources Search images and get ideas for prompts, can search by image to see similar images: https://lexica.art/ Index of various resources: https://pharmapsychotic.com/tools.html Artist styles: https://rentry.org/artists_sd-v1-4 | https://www.urania.ai/top-sd-artists | https://rentry.org/anime_and_titties | https://sgreens.notion.site/sgreens/4ca6f4e229e24da6845b6d49e6b08ae7 | https://proximacentaurib.notion.site/e28a4f8d97724f14a784a538b8589e7d Compiled list of various models, regularly updated: https://rentry.org/sdmodels (Check here before asking "where do I find x model!") Going past the OP copypasta, there's now a new AI generator on the scene: Bing. Yes, that Bing. And it produces some of the best results so far. https://www.bing.com/images/create The tech is moving very fast and I'm not sure how comfortable I am with it.
(757.20 KB 1024x1024 48e016ed112df984dc32d509e64c5d82.png)


(952.86 KB 1152x1152 c93811fd57ef77dbe1921158d9d43895.png)

(560.58 KB 1152x768 katia coomsts.png)

Now this has been mentioned in these threads before, but somehow furries really understand how these models tick. A lot of the impressive AI stuff I've seen that avoids the "AI art style" look (if you've ever played with Midjourney you know what I mean, most of what it makes has similar aesthetic going on) comes from the furry community. I don't know what it is about these guys that makes them so adept at cutting edge tech. The first image image especially surprised me, AI is almost universally terrible at those kinds of art styles. The last one isn't as good as the others but I thought it was funny Now don't flood the thread with weird furry porn please
(1008.05 KB 1016x1440 Request Anchor Girl II.jpg)

Request Anchor
(1.38 MB 1080x1440 Delivery Anchor Girl II.jpg)

>Delivery Anchor
>>2422 Give me a list of outdated links and I will update the op myself with stuff from /sdg/
Dumping some of the vidya-related stuff I made with Bing AI recently
>>2423 I'm not surprised considering how many furries are in high ranking tech jobs, and it makes sense that this catch their attention since it means unlimited smut without trying to find an artist to make a character or draw their own character in all the weirdest scenarios imaginable. SMBC had a comic that I think is proving prophetic, with the rise of technology like this it will mean absolute fetish saturation, any and every weird kink can be satisfied by these emergent AI, leading people to come up with even weirder kinks, until they get bored and have to start getting more abstract, leading to emergent meta-kinks where they can only get off through vicarious life experiences cooked up by the AI, experiencing experiences experienced by someone else in any case I made a bunch of my own weird kink smut with AI and I'm stoked because it's virtually impossible to find existing art that is any good
>>2434 >First and third pic Marty?
(61.90 KB 500x449 crab_nicholson.jpg)

(127.17 KB 1662x849 crab_nicholson_original_post.png)

>>2435 Ah, no. It's Crab Nicholson's Extreme Sleepover.
>>2423 Your otter boy has been requisitioned for the war effort. Thank you for doing your part citizen! >>>/fur/8162
I fucking hate OpenAI and Microshaft so much. They have the most powerful technology but they make it's so utterly cucked it's borderline unusable, every day they're adding more and more blocks and censorship.
>>2438 and it hasn't even been functional the past few days, if your bing boosts run out it simply stops working and they were giving 100 boosts a day, now it's only 25, burning through 25 takes a matter of minutes and then you have to wait a whole day for another handout, I'm sure their traffic has dropped to a 1/4th if not less per day because of that
I just wonder how long it'll take for an open-source version to come out that as this level of capability.
>>2440 I hope soon, and as long as there is no censorship whatsoever, it'll automatically be 10x better than Bing's frontend
>>2440 the irony is that openAI, the company in control of gpt and dall-e started as an open source company, partially run by elon musk who wanted it to be the opposite of google, totally transparent and giving enterprise level AI access to everyone, someone on the board of directions got greedy and decided there's more money to be made, and around the time that elon musk left the board of directors is when openAI became closed source and only letting the privileged few even use their technology, microsoft invested at least $1B into them, hence why microsoft has the access to gpt4 and dall-e3 before anyone else. The real issue is that right now gpt3-4 and dall-e may as well be witchcraft for how they work because it's as secret and locked down as how much gold is currently in fort knox. It may require some clever individuals getting together and brain storming how to build a better algorithm for generating images instead of relying on layers of visual noise becoming sharper and sharper until something emerges resembling data associated with the text entered.
(3.15 MB 4276x2276 Bing AI Before and After.jpg)

>>2442 I have hope for the future of open-source AI. My only real concern is when that happens.
>>2443 I must be stupid because I don't get what this comparison is trying to show, these look the same.
(113.70 KB Voice-026.ogg)

>>2443 Damn, I like this. >>2445 I guess the comparision is trying to say the first 4 are more sexy, at least two of them have spread legs.
Samus Aran from Metroid drawn by Persona 4 character artist Shigenori Soejima, full color SA-X from Metroid Fusion drawn by Shin Megami Tensei demon artist Kazuma Kaneko, full color Ridley from Metroid drawn by Shin Megami Tensei demon artist Kazuma Kaneko, full color
>>2445 maybe the amount of "skin" and detailed backgrounds I really need to look into what all this stable diffusion stuff means, I'm trying out that stable horde site and I don't understand most of these settings and models. Testing out a safe prompt and so far haven't gotten anything close to it, although one image ended up censored even, which I'm surprised, for potential csam, whatever that is
>>2443 >My only real concern is when that happens. When >we band together and beat cuckchan at their own game about "4chan comes together to create something so amazing" and we looked through source code of the best AI we got so far and find ways to make it better for >us. Otherwise, time is going to be slower than without working together, I guess.
>>2442 If I had to take a guess it's probably several algorithms running in tandem. Granted, Stable Diffusion can get really good results, but that's usually the result of custom checkpoints, LoRAs, and other community made perks. Not too much of a stretch to think that Midjourney and Dall E aren't using something similar to further refine it. It probably works better for long term development too since in the event of having to untrain parts of the weight set and compromise the whole model, you could just prune the module that requires it.
>>2446 lol nice
>>2444 >>2452 >cheesy cake
Finally got Miku hugging Dredd! Also got some egyptian robot Mikus. The amount of dogs I had to see before I got all this is absolutely retarded. >>2440 >how long it'll take for an open-source version to come out Soon hopefully, but I wouldn't count on it. Going back to SD after playing around with this and it becomes pretty noticeable how much better Dall-e is at understanding what you want. Prompt "sorceress holding staff in left hand and tome in right hand" on Dall-e and there is a pretty good change that you get exactly that, but on SD you get the sorceress holding either just the staff, the tome, neither, both floating around or just books everywhere and the sorceress holding hands with herself. On the other hand SD runs locally, doesn't have microsoft spying on you, isn't censored and you can use LoRas for specific characters and styles, inpainting, negative prompting, img2img, upscaling, among other things. Just imagine using Dall-e locally with all those tools. >>2439 By the end of the week it won't be censorship anymore, but a lobotomy, They keep banning more and more words and people keep finding other alternatives that eventually get banned as well. 4/5 days ago I used an SD prompt of a bloodied horror girl on Dall-e and it generated exactly that, although more macabre than what SD generated. Now I don't think you can even use "blood" anymore and "creepy" might trigger the fish or dog warnings. One way to increase the chances of avoiding the fish and dog warnings and get your image is to just write a bunch of useless details before your actual prompt to try and confuse it. It seemed to work well for these Miku images.
>>2454 That reminds me, since they seemed to have banned some proper nouns, adding a noun related to the person 'in between' the surname and first name. For example, Tom Actor Cruise. Walt Artist Disney. John (occupation) Doe.
>>2454 it's definitely been lobotomized already, I've had decent success with some prompts, but dall-e 3 is damn horny on it's own accord and keeps generating images that trigger the visual filter. I actually thought they had fixed one of the issues yesterday, because I was finally able to use it beyond the 25 boosts, but it's back to being broken after using it one time after boosts were gone.
>>2455 >John (occupation) Doe. >>2424 Requesting pics of John Eldenring. Or possibly Elden John.
>>2463 i wish metroid 64 was real but then again we probably wouldnt have gotten prime and prime 2, in a perfect world we'd have both also as much as i love sunshine i really wish we got sm64 2, again maybe not instead but i wish we could have both
>>2422 Where'd the last thread go? Does it go to an archive board since so much art is made, or did it slide off the catalog?
>>2435 How the hell can you not know about Crab Nicholson? Fucking zoomers.
(192.41 KB 676x578 dorcelessness.png)

>>2458 >>2459 >>2460 >>2461 >>2462 >>2463 >>2464 All these videogames that don't exist are giving me this horrible wonderful indescribable emotion.
>>2465 post not found
>>2466 it should be archived over on the ais board
Got some old style game magazine ads of Bloodborne and another Miku. >>2455 It's amazing what people find. >>2456 >dall-e 3 is damn horny on it's own accord It's become pretty absurd and since tomorrow is monday, it will probably get a lot worse. >it's back to being broken after using it one time after boosts were gone In my case it wasn't only broken, but I also got blocked after spending the boosts. I had to create a new account every single day to use this. Yesterday I was able to get around wait times by simply doing this though: >open bing search on a new tab >search for "!prompt (prompt goes here)" >check if the bing chat window appeared next to the search results <if it didn't appear, go back to typing the prompt and adding stuff like "high detail, illustration/drawing of (rest of your prompt...)" and search again >go back to the bing image creation page, wait a bit and then refresh >collect images At some point it was telling me to wait 7 hours, so I simply used bing to create the images like that and waited a few seconds. Note that there is some kind of limit as well for bing chat, but it's way over 25. >>2459 >Fun allowed! Pretty cool stuff.
>>2471 related
I'm reading the previous thread. Beyond my wildest dreams. Can it do LOSS? >>2470 Ah yes. I recalled discussion about that but couldn't remember what the board was called.
Oh and a reminder to anyone using bing create, all images have a watermark using a technology called c2pa, I'm not sure of the full extent of this watermarking beyond the metadata stored in the jpegs. There is a tool on github 'c2patool' that can display this watermark confirming the origin of the image is microsoft. Converting the image from jpeg to png with tools like imagemagick seem to strip this data, as c2patool can no longer detect it. There may still be watermark data embedded somewhere in the header that can be found with a hexeditor but I haven't attempted to dive that deep into it. I'm sure microsoft is also keeping copies and hashes of every image generated as verifiable proofs of ownership, so stripping the data is probably a moot point unless you use AI inpainting to fix oddities so that it's distinct from the source, but even then there is such a thing as fuzzy matching images so minute details may not stop such a tool from getting a positive hit from the source.
>>2475 Nevermind, converting the image from jpeg, to png, and then back to jpeg shows the information is still hidden in the image.
>>2476 Just quickly experimenting, modifying images and then exporting the modified image to any format does seem to destroy the watermark, as it's no longer detected with the tool. I may go a step further and create a script to read the image pixel by pixel to a new file, but according to some even just having the image go through compression on something like imgur could destroy the watermark since their comprsesion method does not have c2pa baked into it.
(995.59 KB 1200x1440 Delivery Anchor Girl III.jpg)

(869.37 KB 1200x1440 Lain 3.jpg)

>>2425 Made another Delivery Anchor Girl using darkSushiMix 2.25D and also I don't remember if I posted this Lain.
>>2479 Bro what the fuck I just wanted to check on a few threads and then I see this shit on the homepage lmao. Nice one anon. If mario didn't have his mustache he would look just like mark!
>>2480 Did you notice >>2444 and >>2452 It's Markio
(107.27 KB 1280x960 notexist.jpg)

(369.40 KB 640x480 notexist3.png)

(73.84 KB 711x533 notexist2.png)

>>2468 There was an account entirely for that called "This game does not exist", where 3D artists and gamedevs would submit "screenshots" they made. Definitely very dorceless. Cromulent, even.
>>2481 No. I never scrolled up. I just saw those images, blew air out of my nose and smiled a little, made my post, then moved on. Those images are wonderful works of art as well. I wish AI could make me a 3dpd GF soon enough in the future.
>>2479 These are Pixar-level quality, what the fuck?
(220.94 KB 1024x1024 Alice and Leon Kennedy.jpg)

(263.82 KB 1024x1024 Alice anime game.jpg)

(215.29 KB 1024x1024 Alice Blazeblue.jpg)

(242.86 KB 1024x1024 Alice City.jpg)

(185.95 KB 1024x1024 Alice Crazy Taxi.jpg)

(289.32 KB 1024x1024 Alice Crisis Core.jpg)

(202.06 KB 1024x1024 Alice Dark Fantasy.jpg)

(285.67 KB 1024x1024 Alice DOOM.jpg)

(190.46 KB 1024x1024 Alice Factory.jpg)

(200.94 KB 1024x1024 Alice Fantasy.jpg)

(169.84 KB 1024x1024 Alice FFXIII.jpg)

(267.55 KB 1024x1024 Alice Fighting Game.jpg)

(204.59 KB 1024x1024 Alice Future Atlantis.jpg)

(152.89 KB 1024x1024 Alice Gear Solid.jpg)

(180.41 KB 1024x1024 Alice in Animal Crossing.jpg)

(207.16 KB 1024x1024 Alice in City.jpg)

(151.82 KB 1024x1024 Alice in Silent Hill.jpg)

(211.28 KB 1024x1024 Alice in Tokyo Future.jpg)

(200.84 KB 1024x1024 Alice in tropical environment.jpg)

(182.44 KB 1024x1024 Alice in weird tunnel.jpg)

(314.16 KB 1024x1024 Alice in Wonderland RPG.jpg)

(262.16 KB 1024x1024 Alice Japan Future.jpg)

(206.41 KB 1024x1024 Alice JRPG2.jpg)

(182.03 KB 1024x1024 Alice JRPG3.jpg)

(209.15 KB 1024x1024 Alice Jump.jpg)

>shitting out one gorillion low quality variations of the same idea instead of picking a few of the best ones and polishing them up dall-e was a mistake
(159.24 KB 1024x1024 Boss Alice.jpg)

(257.42 KB 1024x1024 Kiryu in Wonderland.jpg)

(186.32 KB 1024x1024 Alice Persona.jpg)

(278.10 KB 1024x1024 Metal Slug Alice.jpg)

(222.69 KB 1024x1024 Parasite Alice.jpg)

>>2493 Glad you like them. Here you go, last batch.
>>2456 Non-lobotomized AIs would create all kind of horny stuff out of nowhere i trust it
>>2494 AI art style: photographs of an obsolete display technology. All of them, however, are missing a proper shadow mask.
>>2456 >>2495 >>2443 >>2449 Let me explain you something about these ai projects. There isn't an easy way to actually delete a specific behavior, so any kind of "censorship" is actually adding a regex filter between you and the neural network. And I'm pretty sure there are already decent enough models, nevertheless the problem with those is that most people doesn't have enough computational power to deploy those models.
>>2497 >computation power it doesn't take all that much to generate a single image, or even a set of 4, the only reason microsoft needs thousands of gpus is because they have potentially thousands of accounts making many simultaneous requests, generating 4 images at a time, disregarding how many are getting blocked it still takes the cpu/gpu time to generate it, same with all the other websites taking so many requests simultaneously. I'm sure if I were to have stable diffusion trained on one particular subject it would easily generate an image every minute even on my hardware, but to run something like dall-e 3 is a different beast entirely, having several billion parameters or prompts or whatever they call it to define the model, and access to potentially trillions of images for training, it wouldn't just be a matter of having the memory to generate a single request, but also the storage needed for it's neural network.
>>2499 Bloody Pirates!!!!!!
>>2499 Please, more Pirate-themed gaming pictures Also give us something for 8chan's birthday
What tags can I use to give AI generated girls taller pantsu with more side coverage? AI normally makes some very skimpy stuff.
>>2501 Tried to make Francis Drake for the last two, but failed. The AI kept wanting to give her an eyepatch and also it's really hard to make it give her big breasts after the lobotomy. And she has elf ears for some reason
Shadman just released a video defending AI art and it is causing enormous amounts of butthurt in the comments: https://www.youtube.com/watch?v=u_v9Gbw6kcU
It's comforting to know how many users are complaining on other websites about getting banned and suspended for the most asinine of reasons, like putting "safe image content" into the prompt resulting in a 1 hour ban. Users getting banned for including "spooky" in prompts too now that it's october as well. Yet other users are claiming that can use the same prompt and still get results. Either there are a lot of pajeets intervening on accounts or they are using gpt to determine the context of every prompt, and if gpt has a seed per user that's the only thing that makes sense as to why prompts are only blocked sometimes, it's context determination is random per user, every time.
>>2505 MS must know by now that this isn't sustainable. And I hope the source code gets leaked soon.
>>2504 The department of candles is always afraid of being made obsolete.
>>2507 How long ago was it that traditional media artists were disparating digital artists?
>>2504 This is clearly an attempt to gain attention on his part.
>>2508 To be fair there are things that can't be properly replicated by newer tech, mostly desirable artifacts like scan lines, video artifacts, film grain, brush strokes. Digital art, no mater how good it got or how widely adopted it's become ever truly replicated older forms of art. The real tragedy is that the newer stuff ends up completely supplanting the older stuff mostly just out of convenience or because it's cheaper (even if it's only cheaper due to economics of scale due ti the newer stuff becoming more popular) or just simply due to "follow the leader" kind of popularity ratcheting. I just don't want to see older things go away because they are still desirable in and of themselves but that's going to require some artists to step up and get interested in picking up old styles of media and passing the techniques on.
>>2424 >>2462 >>2463 Does the AI recognize Panzer Dragoon well enough to make something cool with it? I've always been in love with the visual aesthetic of this series but there's very little content online about it. Especially Panzer Dragoon Orta.
>>2504 >Direct linking Oh hai, Shadman. https://yewtu.be/watch?v=u_v9Gbw6kcU
>>2512 I doubt he'd even know of, let alone deliberately conflation himself, with the infamous porn artist.
I thought it was just a butthurt comments section but the video is turning into actual drama. His twitter says he's getting all kinds of weirdly aggressive attacks and vile messages worse than he expected. Might be worth posting in the cancerous e-celeb thread.
>>2514 There's a sizable portion of digital artists that are extremely anti-AI, many to the point of kicking up some major drama when someone big uses it. They treat it like tracing.
>>2514 >>2515 For example check out this Twitter thread from a few weeks ago when a pixel artist gave permission for people to train models with their art. Some commenters went berserk. https://twitter.com/nashor_kim/status/1706146337086214626
(254.73 KB 748x600 AI got btfo.png)

(821.36 KB 1744x1866 Typical art pack from deviantart.png)

>>2516 >DevianTARTlets getting desperate Even the deadbeats whose crap went viral aeons ago returned to their main accounts just to replace their forgotten galleries with the anti-AI protest imagery, as if the manchildren running the site would give a shit.
>>2516 <direct linking to Twitter https://nitter.net/nashor_kim/status/1706146337086214626 >>2517 The site is in the process of destroying itself anyway. Maybe their retardation will accelerate it.
>>2424 Kat from Gravity Rush eating an apple in Shibuya.
>>2424 Gondolas.
>Scorpion and Sub-Zero generate fine with no problems >Liu Kang is doable if you add his red headband and being shirtless as prompts >doesn't really know Raiden but doable if you add his white robe, blue vest, and straw hat in the prompt >had bit of trouble with Shao Kahn
Seems bing is somewhat fixed, in that you can generate even without the boosts now, still limited to 25 boosts a day and while I think more prompts are getting through the image content detection seems to be going off more often.
Persona 4 the AI generated arts They look older than they appear in the games, so it's more like Persona 4: The Time Skip
>>2504 Honestly I do not care about the opinion of some washed up heroin addict and neither should you. For me AI art is just fun, techbros and VCs are overhyping it the way they overhype everything and I feel that's what turned so many people against it. When Crayon et al were popular nobody was screaming about this back then, it was still treated like the toy it is and not The Ultimate FUTURE Of MEDIA™ that random crypto bagholders push it as. There are legitimate criticisms of how the basis of these networks were trained to begin with, though. Lots of privacy violations and copyright infringement and among AI researchers by sharing huge imagesets they owned literally no rights to. Scraping random people's Facebook walls and shit then adding those images to their sets with no permission, I can't defend antichrist behavior like that. >>2516 >horrid neo-wojak memes OK, now you have to go back.
>>2504 But that's Shadiversity, the YouTube sword guy, not Shadman, the degenerate porn drawer. He is known for causing a lot of seethe, though. Seems like a nice enough guy.
MADE SOME CUTE AND SMUG GIRLS (Mostly) https://www.bing.com/images/create grinning smug blonde soldier girl Berserk style blue eyes military uniform black and white power pose grinning smug blonde One Piece style blue eyes military uniform black and white power pose grinning smug blonde Berserk style blue eyes military uniform black and white power pose pigtails grinning smug blonde Berserk style blue eyes military uniform black and white power pose Ect
>>2529 Last one!
>>2523 more persona 4
>>2524 > I can't defend antichrist behavior like that. Lmao. If you put a picture in public online, you can't complain about what's done with it other than tracing or defamation edits of real photos. Neural networks aren't tracing. It's literally looking at pictures and generating shit that sort of looks like an approximation of thousands and thousands of pictures, but not a fusion of them.
>>2527 >>2526 >>2528 >>2529 >>2530 >Literally infinite smug works I normally shit on the fags overhyping new tech like Mr. "AI scraping is the antichrist" described, because the real world application of such tech is usually still decades away. This however isn't. This tech is moving fast as hell and the effects will be felt, are being felt, immediately. This is a technological leap to be excited about for once.
>>2517 Anon, you shouldn't have that second picture, you know that, right? You also realize how old that picture is?
All of this was made on AI? which one in particular?
>>2535 Judging by the resolution it must have been Dall E 3.
>>2532 >>2533 Anon I'm not sure you know what I'm talking about, and if you do then you have no right to ever complain about any of the excessive data aggregation companies like Google do. It's not that these were being used in training, (though that is legally up in the air and lawsuits are currently ongoing about it) it's that these datasets are/were public for anybody to download. They just took whatever and packaged it up with reckless abandon. Beyond that being very much illegal under copyright laws, it's creepy and completely unnecessary. They could have used the Wikimedia Commons or paid for stock photos. Not take people's personal photos without their consent.
>>2536 I specially like the medieval ones. So much meme potential.
>>2537 (samefagging to cont.) I want to add that this isn't the same thing as anons packaging up a ton of furry porn from e621. It's the fact the entire institution of AI research was acting in such an legally gray and unethical way, it's the scale of the problem. There's also something to be said about art created by public figures being a different target to the comparatively private recesses of obscure personal social media accounts. I don't expect random internet users to follow or care about the law. I do expect commercial entities and groups of supposed "academics" and "experts" to. The harm being done is very different and not on the same scale.
(511.31 KB 1134x695 geety.png)

>>2537 >They could have used the Wikimedia Commons or paid for stock photos Well they were using stock photos, they just weren't paying for them.
>>2540 >I made something available on the Internet for free and people used it >HOW COULD THIS BE HAPPENING TO ME
<<895689 >being retarded
>>2516 Man this makes me want to troll the artfags. Looks like you could extract a lot of butthurt with minimum effort.
>>2537 >without their consent. Posting images in a manner openly viewable by the public is giving permission for them to be publicly seen. It's not copyright violation for me to save an image off Twitter and add it to my collection. It's not copyright violation for me to locally edit a file on my computer. It's not copyright violation for me to learn about composition, shading, colour, and so on by looking at images I am allowed to look at. That is what AI training does, which means Anti-AI legislation would require making at least one item in that list illegal, and any of them would be unthinkably catastrophic. >if you do then you have no right to ever complain about any of the excessive data aggregation companies like Google do. Secretly inserting tracking methods without consent, requiring disclosure of personal information either by compulsion or deception, and aggregating and selling private back-end data are all a completely different issue than "I posted a picture online and somebody saw it, how could this be happening to me?"
(448.83 KB 641x808 chatgpt.png)

>>2522 The censorshit just keeps getting worse every day though. They're gonna keep driving that nail into the brain over and over until it votes democrat.
Are there any good videos or articles detailing the legality and potential uses of AI art? Almost everywhere I go peoples opinions are purely driven by emotion and kneejerk reactions and there aren't many people discussing things in a less biased manner. Is it likely just going to permanently be in a very grey area?
>>2546 You're living in the tard age of social media. You will not find a single reasonable creature on Earth. Everything is self-righteous opinions shat out onto the net with zero thought and the only choice you have is to wade through the filth and make up your own mind.
>>2545 It's definitely getting worse, ironically it's not stopping unprompted tits and nipples getting through the filter, which makes me curious as to what is actually getting caught by the porn filter if not that. The pajeet in charge of this dumpster fire has responded that the zealous content violations are being looked into, which probably doesn't mean anything other than to placate the enraged masses into thinking this will be fixed.
>>895689 >hiding the hands At least you've learned a bit fgt.
>>2521 >last one Nigga we ALL Chinese!
>>2545 >>2548 Makes me wonder: are there any "hidden" commands that tell the AI to override its blacklist? Sort of like those "pretend you're X" instructions for chatgpt.
(358.60 KB 512x768 1696984552439904.png)

What model is this?
>>2537 >Anon I'm not sure you know what I'm talking about I'm pretty sure you don't know what you're talking about.
>>2552 Nice image, but judging alone by the low resolution without HiResFix it could be any SD checkpoint.
>>2553 He doesn't. Taking images to use as raw data without someone's consent is a non-issue. He might as well complain about Google taking pictures for Street View because his house is on it. And even that would make more sense than what he's saying.
>>2554 >go to civitai to track down the St. Louis lora >"LOL no NSFW content for you" This shit is infuriating, but because normalfags just can't get enough of centralization, we'll probably never see a site that has civitai's NSFW stuff without registration walls.
The surprising amount of (1)s all generally saying the same thing is weird and suspicious to me >>2552 Anon, it's is practically impossible to give you an answer. It's like asking what pencil an image was drawn with. Your filename indicates you found it on cuckchan, go ask whoever you saved it from.
>>2557 >spoiler Maybe all the AIs are posting in the AI thread. :^)
>>2557 >The surprising amount of (1)s This thread has less than 150 posts. Not every thread is goober gabber where the majority of posts come from people that are terminally in the thread with (5) to (30) posts before it even hits 100.
>>2557 ISPs with dynamic IPs exist, anon. My IP changes every 12 hours or so.
>>2556 >without registration walls I guess I dont care about that.
(1.02 MB 1200x1440 Request Anchor Girl III.jpg)

>>2424 >>2425 >>2478 Made another Request Anchor Girl using DSM225 model also.
>>2561 The problem isn't registration being required, it's if they let you sign up with a burner email or not. I've never tested it. >>2556 Registration walls are a good thing if done correctly, because it keeps normalfags from seeing something to complain about if they don't register.
>>2563 >it's if they let you sign up with a burner email or not It worked the last time I tried (several week ago)
>>2478 >>2562 Really cute, keep it up.
>>2564 Then he has nothing to bitch about.
>>2531 More Persona 4? OK!
>>2567 The Other cast, Igor had a lot of horrifying results
>>2568 Featuring Dante From The Devil May Cry Series
>>2568 Made more of Elizabeth. Dall-e sometimes gives her cleavage, then sometimes not
(15.45 KB 256x256 singu2.png)

>put "mouth is closed" <CONTENT BLOCKED >change to "mouth is closed" <CONTENT BLOCKED >change to "mouth is not open" <proceeds to generate prompt but fails to follow directions Ah the beauty of technology.
>>2571 oh one of those was supposed to be shut, not closed still weird that of all the things to get prompt blocked is trying to make a character without their mouth getting wider and wider with each generation like they're preparing to eat my soul I've gotten so many generations now where the mouth is just agape in every single one and I swear it's getting bigger
>>2572 Have you tried 'neutral mouth expression'?
>>2501 I have some more birthday ones I could post in a bit
(234.37 KB 649x235 shadman.png)

(291.16 KB 533x601 shadman2.png)

The Shadiversity shitstorm is getting interesting. Shad made a long response to all the shitty messages he's been getting and twatter is big mad that the response is too long. Apparently, Shad has a brother who runs a highly successful art channel on YouTube and now the twatters are trying to get him involved in the controversy and politicizing it. Twatter is spamming Shad's brother with messages about how Shad is an alt-right nazi, a transphobe, etc. and politicizing the issue, it got so bad the guy made a public statement about it. So now they're using politics to try to get Shad cancelled by his own family, all because he defended AI art. This is getting pretty unhinged.
>>2575 >This is getting pretty unhinged. It's a struggle session, it's suppose to be: https://odysee.com/@newdiscourses:9/surviving-a-modern-struggle-session:f
(287.89 KB 1024x1024 AI Octoling 1.jpg)

(305.52 KB 1024x1024 AI Octoling 2.jpg)

(309.71 KB 1024x1024 AI Squid 1.jpg)

(350.56 KB 1024x1024 AI Squid 2.jpg)

(368.36 KB 1024x1024 AI Squid kid 1.jpg)

(363.61 KB 1024x1024 AI Squid Kid 2.jpg)

(269.52 KB 1024x1024 AI Bandaged Squid Girl.jpg)

(1.03 MB 1024x1024 AI Octoling Witch.png)

(729.73 KB 598x792 260.png)


Right lads, the AI somehow won't let me do shit from my IP despite me not using the tool at all last week, but I just saw this post and I wanted to ask whoever can still access it if they can use the same method (@name, presumably twitter handles) to craft pictures with them and somehow bypass the censorship. >>2574 Cheers lad >>2575 I find it really amusing as his pro-AI argument is probably the most grounded one, in that people don't understand shit of how it works.
>>2579 >whoever can still access it if they can use the same method to craft pictures with them and somehow bypass the censorship. You can still make prompts of bill gates, he just has to be referred to by his real name, William Henry Gates III
So, can AI make old heroquest/warhammer/classic black and white 80s rpg book art?
(137.20 KB 1024x1024 OIG.MXt0gawludLWT8WgU7Zz.jpg)

(145.22 KB 1024x1024 OIG.WrN1o87k.zAVf.b6ILFL.jpg)

(205.99 KB 1024x1024 OIG.Tus9ThCcAILpGMdy6sAG.jpg)

(177.75 KB 1024x1024 OIG.qtxqppAu28rcDBsiXnvM.jpg)

(175.90 KB 1024x1024 OIG.mzKNaUwPgY68HGKnDKRo.jpg)

>A picture of saturn eating his children but with klaus swab eating a cockroach instead Blocked. >A picture of Klaus swab as a Skibidi toilet giving a conference at the UN Blocked. This shit is so fucking lobotomized i wasted all my tokens. Tried the >>2579 trick and they blocked it as well. The only ones that went through were the shitty tests with chewbacca in a ridiculous swimwear, the skeleton screaming at cabbages and different types of Sylvannas screaming and angry at stuff.
>>2582 Texas enfuego? Also I agree, my enjoyment of this AI has gone down the shitter. I can barely even use my 'safe' ideas with no profanity at all because it somehow has an issue with some inane word. I am hoping more and more every day for someone to release an open-source alternative.
>>2585 Down the shitter indeed, every day it passes it censors even more shit, i can't even do females unless they're random. The Sylvanas pictures alone they took me like three days because the nigger dog was always there, like you have this awesome tech at your reach but they cripple it so hard that instead of enoying it makes it sour leaving you with more craving for SD. But hey, dudes and fantasy (no females allowed) stuff looks cool.
>>2586 How was the og stable diffusion model leaked anyhow?
>>2583 Thanks!
>>2425 >>2583 Now do the same thing, but make her naked.
photo of big-nosed merchants with beards, black robes, black thick glasses, white and blue towels around their shoulders and kippahs running from a giant black and white cat using its laser eyes to destroy a desert town. explosions in the background
>>2582 Seems like klaus schwab and characters that may bear a resemblance to him like Dr. Evil from Austin Powers or Blofeld from the Bond series won't fly on dall-e. Was able to get a few prompts of Blofeld earlier somehow, but wasn't satisfied with the results. While he doesn't bear a complete resemblance to schwab, Eggman managed to fly past dall-e's censorship with the prompts of 'old man', 'bald', 'scar underneath his right eye' 'thin-framed glasses' and 'Sonic 06' Don't think this particular one influences anything The dog popped up more when using prompts to remove the facial hair, so you'll just have to hope it spits out an image and it forgets to give them a mustache.
>>2591 Other images created with the prompt
>>2583 >>2424 >>2589 Meant to reply to the request anchor, not the delivery one.
Sorry to ask, to anyone with a bit more imagination than I have, could you help us out over at >>>/ais/2407 making some AI-generated banners? >>2594 >weakest loli appreciator
>>2586 >Living in the undercity Not sure why, but sonic trying to have a good time with sylvanas is wholesome. If anyone can fix her, its him.
>>2570 3rd one is like Chie in Elizabeth's outfit. Need.
>>2589 Since it's cucked to hell, has anyone tried using it just to generate a base image, then using other models and their tools to edit it?
(2.41 MB 2048x1024 angry at shitposts.png)

(1.04 MB 2048x2048 angry at everything.jpg)

(1.40 MB 2048x2048 wow in the sengoku era.jpg)

(6.46 MB 2048x2048 Real winners.png)

(21.46 KB 300x100 Banner shit 1.jpg)

Tried to make more WoW stuf but in the Sengoku era, i do find hilarious how they respect the oversized pauldrons, and the dwarves looked so awesome i made a crapload of them. >>2591 >>2592 Those look nice, my idea was to use the original picture of Saturn since is a well known picture for everyone but with the right man and his rightful meal but since they don't want to be shown as villains they should stop looking and acting like ones. But is normal that sometimes your characters have those cracks on them? Some of mine kinda showed those too. >>2596 Characters like her are ripe material for use in stupid and awkward situations (like being angry at shitposts, at tomatoes, at (you), at pizzas, getting screamed by your boss orc) and it was suppose to be bored in a rollercoaster ride and that was the other first character that came into mind and it looked nice. My only complain is that is hard to get her outfit and appearance correct. Also i used your idea to make a shitty banner for >>2595
>>2586 I like the Radikal ork.
>So you're putting chemicals in the water that are turning the worgens gay? <And orcs and pretty much everyone, but lets be fair i don't care about the horde even.
>>2589 >spoiler Sorry, but Kat is just a nice average girl trying to save the day while definitely wearing underwear and not reading erotic novels at work. >>2599 Pretty cool stuff. Cute banner too. Prompting characters in silly situations is pretty fun. Too bad that constantly seeing the dog basically sucks all the fun out of it.
Apprently just putting the word "clothed" into my prompts instantly improved dall-e3's results immensely, I kept getting their stupid "inappropriate image content" error in spite of trying all kinds of prompts to force covering on characters where I thought they were triggering the filter. All it took was one word, clothed. Fuck.
>>2602 >Kat reading an erotic novel of herself This is hilarious to me for some reason. MOAR please.
(1.11 MB 1600x1920 Kat.jpg)

>>2605 the fact the AI actually gave her faint stretch marks is weirdly impressive.
>>2605 Does he have two vajohnyas?
>>2607 *shelol
Made a few images for the /ais/ banner stuff, the words shit and chan gave me nothing but niggerdogs, what a shame.
>>2609 >Can make nearly anything interesting or fun <Instead makes and post absolute shit Neck yourself already, it's like seeing an stupid autistic children repeat the same stupid joke that wasn't fun every fucking time
>>2610 Obsessed terminally-online alogs are like that
(233.23 KB 1024x1024 OIG.jpeg-547.jpg)

(215.18 KB 1024x1024 OIG.jpeg-549.jpg)

(208.47 KB 1024x1024 OIG.jpeg-533.jpg)

(227.13 KB 1024x1024 OIG.h2.jpeg.jpg)

I want Nintendo to double down on their mushroom imagery. I also want them to expand on the Rosalina lore, it'd be cool if Mario could visit her again for one world, but this time she's a commander for an intergalatic space fleet fighting against a space invader, I would love to see a big jump in the lore and storytelling, it doesn't need to have the cinematic build-up that Zelda does, just make it cool, appealing and interesting.
What free AI voice service do you goys recommend for generating some sexy voices? I need quite a few AI voice lines for my mod so that they have some lines to speak instead of the voice that I ripped off from another mod.
>>2613 Try and get Elevenlabs to generate a female, raspy voice, then use RVC to clone some female VA's voice onto the generated one >>2609 Impressive
>>2614 So I need a bunch of female voice downloaded in my hard drive for cloning purpose? Shit man I don't have any.
(6.66 MB 1024x1024 Anon - Alone in the town.webm)

(5.91 MB 1024x1024 Anon - Unfamiliar Place.webm)

(16.28 KB 600x600 sayakaannoyed.png)

>no Sayaka lora >no Katie Power lora >no Julie Power lora >no Cindy Vortex lora Is lora creation difficult? >>2615 Check your video game folders. A lot of video games have voice files easily extractable, and a decent number have them as loose files in common formats.
(374.33 KB 512x512 440565189-2.png)

>>2617 Hmm alright, I guess I will lower this priority then. Because I am not motivated enough yet to scavenge for female voice lines and then making new lines from it. I will note it though so that I know what I have to do when I can generate some new lines. Thanks for answering my question, anons.
>>2617 >Is lora creation difficult? Try to make them yourself: https://civitai.com/models/22530
>>2620 >There are no free services anymore Fuck. Is there any tool I can use instead that I can run in my computer? I've got a RTX 3060 Ti GPU.
>>2621 there's an open source voice cloning project on github called tortoise, link should be in the ai voice thread only problem is it's all python because nobody knows how to use compiled languages any more
(51.14 KB 640x480 1077076.jpg)

>>2622 >only problem is it's all python because nobody knows how to use compiled languages any more As a counter point to this, Python may be a shit, bloated language that adheres to no standard in particular and is filled with reference-hell, but it's also great for quick prototyping and testing out your theories. Most people using it over other languages are skiddies, people working in IT and researchers, none of which care enough about efficiency as much as fast deployment times and tweakability. You can probably remake the entire project in some other programming language, but the people who're putting actual work in the background details are not computer scientists, usually.
(452.21 KB 512x512 2188134857-1.png)

>>2623 >>2622 >Python bad tbh I think Python is fine for learning purpose, some basic utility and GUI programs that are relative simple in functionality. If I want to program something more serious I'd would use Nim instead since it looks like it's just as easy to use as Python but yet far more performant. But Godot 4.0 doesn't has a proper Nim language binding yet which sucks.
>>2623 >>2624 Aside from poor efficiency there's also the issue that it becomes just as unmaintanable as any sufficiently large project in another language, and you end up losing track of what anything is in the code because it doesn't enforce strict data types. I've gone back and looked at stuff I've done in python and I'm lost as to what anything is because I don't know what's a string, integer or boolean anymore, and that makes understanding my own functions hell because any parameter passed to them and what's being done is a mystery, ends up being faster to rewrite the whole damn script than trying to refactor it. It's also become a pain in the ass trying to install anything written in python, as someone realized that it's a massive liability to just run pip install package and have unvetted projects with no sandbox installed on thousands of machines so now if a project isn't in an official repository you have to first install pip, then install to a virtual environment, but you can't because you you didn't install python-venv first, so you do that and now install to a virtual environment to partially sandbox the project. Then having everything break when python gets upgraded to the next major version is fantastic, because nobody wants to refactor their shit for the next version, python2 had to be kept as the default version for YEARS because of this, and removing it from a system the wrong way broke everything because of how much software is just python scripts now. Python is fine for prototyping, sure, but for the sake of everyone's sanity it needs to stop being used for the final product.
>>2624 >tbh I think Python is fine for learning purpose <Not Basic <Which has seen practical utility since the 70's
>>2625 Damn, I didn't know python was that bad, welp. What other babby languages would you recommend instead? >>2626 >Programmer wars I don't remember giving a shit. I'd be broke if I did.
>>2624 I never stated that PYTHON == BADD, just that it's used for applications that don't suit it. Kind of how Javascript is great until you start filling a web page with a hundred different scripts and make everything into a bloated mess. >But Godot 4.0 doesn't has a proper Nim language binding yet which sucks. Be the change you want to see in the world >>2625 >and you end up losing track of what anything is in the code because it doesn't enforce strict data types. Just comment lmao Documentation in and out of your code should be as large as the code itself, I don't care what minimalistfags say >It's also become a pain in the ass trying to install anything written in python, as someone realized that it's a massive liability to just run pip install package and have unvetted projects with no sandbox installed on thousands of machines so now if a project isn't in an official repository you have to first install pip, then install to a virtual environment, but you can't because you you didn't install python-venv first, so you do that and now install to a virtual environment to partially sandbox the project. Objective, valid criticism >Then having everything break when python gets upgraded to the next major version is fantastic, because nobody wants to refactor their shit for the next version, python2 had to be kept as the default version for YEARS because of this, and removing it from a system the wrong way broke everything because of how much software is just python scripts now. My conspiracy theory is that they keep changing base functionalities between versions to shill new Python Tutorial videos and courses
>>2626 Sorry gramps, in the future everything is bloat >>2627 >What other babby languages would you recommend instead? Python is great for learning the basics, but it also teaches you bad behavior that would not fly with stricter programming languages. Unironically just stick with C and never ever move onto (((Object Oriented Programming))).
>>2629 >Unironically just stick with C and never ever move onto (((Object Oriented Programming))). How many "bad habits" will I pick up if I learn C++ before C?
>>2625 I see what you mean. I've worked on a python chatbot and had a function that parsed the inputs and turned them into workable data for the bot's API that started giving me issues, after a whole afternoon of debugging I noticed that the code turned a variable that was supposed to be an array into a string for some fucking reason. Implied variable declaration is so goddawul for project declaration. >>2627 As someone who's into math, Python is decent for deploying random games and big numbers tests and other similar stuff. Just like the other guy pointed out, the problem is not the language itself but rather forcing it to do stuff it isn't supposed to be doing, just like JavaScript. Python is good if you want to prototype something and don't want to get into programming technique details, specially if you are not a programmer. Pretty much the reason why a lot of these AI projects are developed on Python is because you'd spend most of the time curating the training data rather than programing the bot itself.
>>2627 The language you want to learn ultimately depends on what you intend to do. C should be the gold standard of programming anything, it does not hold your hand, you must declare data types, and if your syntax is wrong your code will not compile or at least it will warn you about issues that wont cause a segmentation fault. Also you WILL learn to manage your memory or you will suffer the consequences. If all you intend to do is simple tasks like calculate the time difference between 2 calendar dates or some string manipulation then python is fine, it's fine for simple scripts where cpu time is not critical to you, but you are less likely to become a competent programmer because python 'just werks' without being a competent programmer. Also it can be kind of stupid to use python for some things if you're on linux, because there's a good chance what you want to script is already doable with bash and gnu core utilities. C++ is C with extra features out of the box that can make it slightly easier to get started, but all the same stipulations of learning C should apply, being good with C will lead to being good with C++. Whatever you do, never use Rust, it's a stupid fucking meme language with it's own set of problems that they claim are not problems because we hide the fact that these problems exist, and all the claims of memory safety aren't the be-all end-all because you can have the same memory safety in C if you understand C.
>>2628 >Just comment lmao Fuck comments. They always end up causing more confusion than help because it takes just one change for the comment to get left behind the times and become misleading. You have to watch every PR like a hawk to make absolutely sure no dumbass forgets to update the docs after changing the API it's attached to and a few still slip into master. Project wikis are even worse, never document any functionality there. Just write good, readable and properly structured code. If you ever need to comment to explain what something is, why it's there or how it works it means your code sucks and you should stop copypasting shit like a pajeet. >>2629 There is nothing wrong with OOP. It's essential to keep things separated into manageable pieces that can be reused and maintained. Not that a scriptfaggot would ever understand that. >>2630 You don't pick up bad habits by learning a language. That comes from either following bad advice without knowing any better or repeating common mistakes because (You)'ve never dealt with those yet.
>>2623 >researchers, none of which care enough about efficiency as much as fast deployment times and tweakability What? There's tons of situations in research where efficiency needs to be a priority. Sure, there is a lot of stuff that Python is used for, but there's a reason C and Fortran are so popular.
>>2634 >There's tons of situations in research where efficiency needs to be a priorit Sure but there's also tons of situation in research where it doesn't need to be. For instance, I had a relative who had his master on fluids simulation, he has a major on math so his thesis focus was on the mathematics that makes the simulation work rather than the programing. If he had to make the code run efficiently he would have to learn an entirely new branch of study basically from scratch. That's the reason why Python is so popular on research, you don't need to understand but the basics of programing to make your shit work. Of course efficience is important but if you were a Mathematician doing fluid or heat simulation you wouldn't be wasting your time making the code run efficiently on the most optimal programing language, would you? instead you would be focusing on the actual math that makes everything work and code with whatever can get the job done.
>>2634 The problem is that they're researchers. They never had to write, package and deploy real software that has to be maintained for years and must work without the original dev debugging every step. They're only taught to write shitty one-off scripts or proof-of-concept prototypes for a paper they hand in and forget, so they'll use the easiest script language they can and fuck up it up anyway.
>>2636 >They never had to write, package and deploy real software that has to be maintained for years and must work without the original dev debugging every step Experimental and observational apparatus doesn't run on magic. The data processing pipelines face exactly the situation you described, that a decade or two later it still needs to work even after the guy who worked on it moved to a different continent.
>>2637 Just to bud in: Researchers spend years investigating some topic that is usually not programming related. Most of them spend nearly no time studying or doing programming. AI is mostly applied math, and its researchers gravitate to a language that allows them to program in the abstract formalisms that they are accustomed to working in, and not the babby concepts that computers actually understand. (And most developers I know would never look at assembly for fun.) People who have honed their craft of programming generally haven't spent nearly enough time in the domain of AI to be proficient, and people who have spent all their time in the domain of AI in order to be at the bleeding edge generally haven't honed their craft of programming. > Experimental and observational apparatus doesn't run on magic I have seen the PROJ source code, and I beg to differ. PROJ is a map projection library written in C by a geologist. It is full of: that's an interesting way of solving that problem; I would have never considered it and will never consider it. You can write an ugly, dirty script in C, and it is amazing how long a C program can last. Within the last decade, assessor's office of the county I live in decided to replace its old VAX system. After seeing what replacement solutions were out there, they decided to buy a beefy server and installed a VAX emulator onto it.
(187.12 KB 1024x1024 Operation Piss Bottle.jpg)

(96.16 KB 1024x1024 Operation Piss Bottle 3.jpg)

He's literally me
>>2626 >>2631 >>2632 >>2629 Uhm okay, well I never build a big software with Python. I mainly wanted to learn python for the sake that I know ANYTHING about programming in the first place and now after several years of on and off of fiddling around with python and other languages I am more confident in my programming ability. I developed already a pattern which I mostly adhere too to make sure my code is at very least readable and I make great usage of writing several functions and splitting them to make sure that a function doesn't do too much stuff at once. So when I go back to over my older code I still can mostly read my code because I didn't made a huge mess out of that. Okay I did wrote some code which is really messy but that was during my early stage of Python development. The next language I want to learn is Nim since that looks pretty promising, it has a fairly nice looking syntax structure and it's very fast too according to its benchmark, so I hope that I can use Nim very soon as my go to language for programming in the future.
>making Persona 3: The Time Skip >do Mitsuru >add 'MILF' as a prompt, thinking it won't go anywhere with dall-e's crazy filters <the dog hasn't popped up, at all <even get bonus cleavage Dall-E confirmed to love older women
>>2642 >>2643 That's basically the best I could get the damn thing to make in the last 2 months. I have /d/ content, but this is already pushing it and I don't think it's appropriate for this board.
>>2642 >>2643 >>2644 Seems to want to make futa. You could also go the route of inpainting on only the ones with the best composition and honing in on a few instead of just letting whatever combination of checkpoints and loras will get you. Also the colors look a little washed out in a few of them. You using any vaes?
>>2645 Basically yeah I use any and all /u/ vae I can find; 1 out of 25 attempts actually make strap-ons and the rest just come out as futa. And the oral stuff just doesn't work. I try to do strapon on strapon lesbian if I looks like I can, considering how rare it is in porn It's just plain easier to do futa, but I'm still trying to find better /u/ vaes.
>>2646 Try using "head between legs" in your prompts. Even if you can't get anything explicit you can probably get some good implied /u/ stuff. plus, kissing is intimate and cute enough.
(1.89 KB 100x148 doomwhy.png)

Those damn lazy coomer AI scientist should stop fapping to AI porn and finally make a AI assissted tool for generating DOOM maps. Those AI slop art is some low key shit tbh
>>2641 Aigis had some interesting results
(139.53 KB 1024x1280 Ooga booga.jpeg)

>>2648 It'll just end up being a more fanciful version of Oblige level maker if it does come around
(486.75 KB 250x193 soldier thinken.gif)

>>2650 Wouldn't it be able to generate maps though that looks like it belongs to map pack instead of being a confusing mess of a maze?
>>2652 Blow it of your ass.
>>2650 I'd be interested if Obsidian could be augmented with something like generated textures first. Right now it has more than enough prefabs and such.
>>2650 Is that supposed to be that thing that blocks the doorway in Silent Hill 3 when you're trying to leave that building under construction?
Now I've confirmed bing "boosts" do not accumulate. I figured they would rollover to the next day when they add another 25 boosts, but no, however many boosts you have one day, that's how many you'll have the next day when it's supposed to refill.instead of even just topping it off back to 25. I have to assume this is their way of incentivising wasting your time trying to use every boost you have in order to get another 25 the next day, there's no way they've been running this for 7 months already and it's THIS broken.
(33.08 KB 278x242 I can get a load of this.jpg)

You know how some AI can spit out a generation out of a user-made sketch? Has anyone tried the same, but with SFM/Garry's Mod screenshots, or even screenshots from different games entirely? I'm specifically thinking of setting up the entire scene myself, posing all the ragdolls/props/effects/etc., then running it through a program to make it fit a different artstyle while still maintaining the scene itself. It's likely gonna be a pain in the ass to set everything up for the first time (I'm an AI newfag) and even attempt it, so I wanna know if anyone else has had success in this. If not, then maybe it's something you guys can try on top of everything else if you're willing to put a little work.
>>2657 You could theoretically accomplish a similar result by drawing on a piece of paper and taking a picture of it to upload into the AI for you to then paste whatever "style" you wanted on to it. Don't ask me how to do it though because I am not sure.
>>2641 >>2649 Hi, I wonder, how big can you make the boobs with Dall-E?
>>2659 Dall-E will not go any higher than 'large bust size'. If you attempt to put in 'extra large bust size' or anything remotely similar to that one, they won't go through. Believe me, I tried to be clever by putting things like 'mammoth large', 'super large', 'humongous large', 'sizable large' and every one of them was rejected, plus Dall-E put me in time out for an hour. It doesn't have a problem with 'plump body' but you may have to generate a few times to get some passing results. First 3 results are with that prompt and the remaining 2 have the 'large bust size' prompt.
>>2660 what if you tell it to draw a tiny fairy-sized girl with normal human sized breasts??
(49.23 KB 619x676 erect donkey dong.jpg)

>>2661 >what if you tell it to draw a tiny fairy-sized girl with normal human sized breasts??
>>2661 Sometimes Dall-E's dog will be greeted. Sometimes results come out where they have big titties despite having a 'normal bust size' prompt. And then there's the unusual results. Using 'tiny fairy-sized girl' as a prompt, Dall-E thinks you're talking about the likes of Tinkerbell
>>2663 All in all, be prepared to spend time wrestling with Dall-E and hope you get what you want.
Anyone else getting banned from Michaelsoft for no reason? Fuckers want my goddamn phone number all of a sudden. >>2630 >How many "bad habits" will I pick up if I learn C++ before C? 1. Get it working first 2. Keep things as simple as possible 3. Keep things as light as possible (don't use more memory than you need) 4. Keep things as documented as possible, every line of code needs a reason to be there 5. Reuse as much code as you can, instead of copy and pasting the same lines everywhere Bad habits are developed when you follow tutorials that give you imprecise information or take the easy way out. Some programming environment/languages make you more prone to it, i.e. if you have ever used Matlab for any amount of time you'd know it really likes making everything into a Double Precision floating point variable, so if you are paired with people who have no idea about computational overheads expect copious amounts of pain. >>2631 > after a whole afternoon of debugging I noticed that the code turned a variable that was supposed to be an array into a string for some fucking reason. Implied variable declaration is so goddawul for project declaration. That's how it is on this beach of an Earth
(5.04 KB 189x185 Jean Paul Bartre.jpg)

>>2632 >Whatever you do, never use Rust, it's a stupid fucking meme language with it's own set of problems that they claim are not problems because we hide the fact that these problems exist, and all the claims of memory safety aren't the be-all end-all because you can have the same memory safety in C if you understand C. People in tech have cult-like mentality when it comes to using stuff that isn't mainstream. Reminds me of a recent Louis Rossmann video where he complains that people begging him to move to Odyssey don't know that he already has, with the same username to boot. >>2633 >They always end up causing more confusion than help because it takes just one change for the comment to get left behind the times and become misleading. That's because you don't keep them updated. Obviously if you don't maintain your code or make any change without mentioning it anywhere the comments become unusuable. What I like to do is keeping a small history log of what was in a specific script so that if it gets changed too much I can at least run through the steps that got it to be like that. >You have to watch every PR like a hawk to make absolutely sure no dumbass forgets to update the docs after changing the API it's attached to and a few still slip into master. Hell is other people Pajeets. >If you ever need to comment to explain what something is, why it's there or how it works it means your code sucks and you should stop copypasting shit like a pajeet. If you have an ironclad memory for huge projects, sure. Otherwise you'll encounter a lot of situations where your variables look like vMaxAllwd_tether2_temp and you'll have no understanding of what the fuck you're dealing with. >There is nothing wrong with OOP. It's essential to keep things separated into manageable pieces that can be reused and maintained. Enjoy bloat
>>2660 try adding "exagerated" to the prompt
>>2661 >what if you tell it to draw a tiny fairy-sized girl with normal human sized breasts?? I could not help but think about rkkrn's Oppai loli touhou girls. Damn, I love them. >>2663 The third one if fairly good. >>2664 These are cute.
>>2663 >Microsoft's forced diversity generates a nigger fairy >it gives her a green-trimmed red dress that makes her look like a slice of watermelon Sheeit.
>>2669 Highly doubt thas has anything to do with it; nevertheless it's pretty funny to think what the regex parser (implying the would be usinh Regex on the input instead of simple verification just like the AID admins did) would be doing under the hood if that were the case. >>2648 >AI generated doom maps There are already those, all of them are trash. Why would you want more of that?
>>2640 It's okay, man, there's really nothing inherently wrong with Python, the thing is if you want to learn programing seriously you might learn bad habits in the long term. Basically is the same reason why people suggest to avoid "anime style" when you are starting learn drawing or why some math teachers suggest set theory or nodes theory to get into serious math instead of calculus or algebra.
>>2667 Hey, "exaggerated bust size" worked! At first I didn't see a difference in bust size but as I kept generating (and getting past the dog many times which would indicate that the terms aren't hard banned but combined together makes the generated image too lewd for Dall-E) and mixing different prompts, finally got to see the exaggerated bust size in action. So it looks like Dall-E is okay with bigger busts after all, just needed to reword them.
>>2672 That bust size pairs very well with the "muscular female" prompt but the latter is more of a wild card. Sometimes Dall-E makes them look like they took steroids, and other times they have muscles in the chest but their arms look like they haven't worked out. May be able to remedy it with "slightly muscular female" but I haven't tried.
>>2673 Used up my remaining boosters. Notice the steroid arms in the right-most pics.
>>2666 >That's because you don't keep them updated. Of course. Thing is, comments are impossible to keep up to date because there is no automated way to check them. Even if I personally catch every bad PR it's not enough, shit will slip through while I'm on vacation or something. That's why comments only really work in personal projects when you're the only developer. >What I like to do is keeping a small history log of what was in a specific script so that if it gets changed too much I can at least run through the steps that got it to be like that. Skript kiddie identified. Learn git and use it. >If you have an ironclad memory for huge projects, sure. Otherwise you'll encounter a lot of situations where your variables look like vMaxAllwd_tether2_temp and you'll have no understanding of what the fuck you're dealing with. None of that actually occurs in real work. The entire point of OOP and FP techniques is minimization of cognitive overhead. With functional code you see the whole data flow in front of you and don't read any details unless you need it. Object oriented code has everything you need to know about any object encapsulated inside of it and usually less than a full screen length at that. As for retarded names, that doesn't happen when you're white and/or have senior devs blocking PRs for violating naming conventions. >Enjoy bloat There is none. OOP is all about preventing bloat and making DRY trivial to enforce. >>2673 >>2674 It seems to focus on the chest part first. Some of these look like the "muscular female" prompt has been completely ignored. Does prompt order matter?
>>2675 >Does prompt order matter? Hard to say, I would think not because every prompt is going to be combined regardless of the order they're in. Even when you put in something like "full color", there's a chance that Dall-E won't actually generate an image that is fully colored. For instance, these pics have been generated with "pencil sketched", "2D art style", "traditional media", & "full color". For the most part they followed the prompts given, but 1st pic doesn't really look like a pencil sketch, and the 3rd pic doesn't have full color.
(162.16 KB 1164x1500 81qiPyPsMNL._AC_SL1500_.jpg)

Ok can someone explain why it always ALWAYS fucks up the hands. I understand hands are pretty complex but still it should get it right at least sometimes but it always adds an extra finger or weird lumps meanwhile getting hair and clothes wrinkles and shadows and other equally complex parts perfectly fine. Just feed it a bunch of sign language images and surely it can learn to recreate any hand doing and gesture from any angle pretty easily considering everything else it's already able to do almost perfectly.
>>2672 >>2673 >>2674 These are awesome, can you share some prompts to get some cute images as well?, sadly is pretty much impossible for women to have big boobs and muscles. >>2677 >why it always ALWAYS fucks up the hands Most people don't use control net open pose.
>>2678 >can you share some prompts to get some cute images as well Ok. I don't remember the prompts for the fairies earlier so you'll have to stick with muscle Chie for an example. Chie Satonaka from Persona 4, dyed light brown bowl-shaped hair, brown eyes, light green high collared athletic jacket with yellow lines and adorned with several decorative pins including a smiley face pin, small black and white plaid skirt, exaggerated bust size, white top, muscular female, 2D art style, traditional media, pencil sketched, full color, mature woman, old woman, indoor gym background, lifting weights Some things to note with the prompts posted <if you're using a vidya character such as Chie, their wikia page may already have their design profiles made, just copy them in separate prompts and add some finer details that are likely to be missed once you generate. For example, the smiley face pin and the yellow lines on her green jacket. This cannot be emphasized enough to get their details correctly. <for the most part, the "2D art style" prompt reduced the chance of getting ugly 3D models <the combination of "mature woman" & "old woman" seem to be key in making MILFs. The keyword itself, MILF, isn't banned but I think Dall-E gets the idea of them with those two prompts <if you don't want a boring white background, specify a location to put them in. You may also want to give some descriptions of the background such as a sunset beach or a night city <for cuter images, use a prompt to describe a small and young girl & find a way of making it work with them by using perspectives. For instance, "floating above a human's hand" if they're a fairy or "tall mushrooms". For a higher chance of cute, use "2D art style", "traditional media" and "pencil sketched" as the prompts. No guarantees, but better than 3D >sadly is pretty much impossible for women to have big boobs and muscles If you get the dog, you just have to be persistent with Dall-E and it will eventually spit something out. It's only when you get a warning at the beginning that one of the keywords is banned that it will refuse to generate. These muscle Chie pics I posted show they can be done.
(85.75 KB 1024x1024 OIG.rd0awtTOyKXc.jpg)

(144.04 KB 1024x1024 kOIG.aW8lj03zJS.jpg)

(112.63 KB 1024x1024 jyf.jpg)

(100.76 KB 1024x1024 kOIG.jpg)

(120.40 KB 1024x1024 OIG.jpg)

>>2677 because it's hard to define hands into textual form, remember that AI isn't understanding anything, it's a game of associating image likeness with a series of tags, so even if there's a million images of hands, there's several million permutations of what hands can look like, different poses with slight variations in finger position and length and color and size and texture, maybe some fingers aren't visible in a chunk of the images, maybe some hands simply do not have all their digits, it all depends on the weights of the images it's trained on attempting to get the closest approximation of what you're prompting for
>>2675 >Learn git and use it. My work team uses it and they don't know jack shit about it, they don't even have the decency of not touching shit that they're not supposed to. >None of that actually occurs in real work. It definitely occurs in real work. No one makes commits explaining what they did, nor the rationale behind variable naming conventions. I had to force a higher up to halt the project we were working on and do it manually with me because it was impossible to navigate, and once we were done some other dipshit came around and used entirely different variable names. >With functional code you see the whole data flow in front of you and don't read any details unless you need it. If your functions are as long as fucking Pride and Prejudice you need to go back to the drawing board and simplify it.
>>2680 >muh trad lifestyle
>>2681 It's more than that. Even pretty clear poses like "middle finger" and "fist" are difficult for AI to handle.
>>2683 Yes my trad lifestyle
<<898543 Zero effort spam.
>>2686 >feeding instead of reporting and ignoring Ya might as well be helping him.
>>2604 >hilarious for some reason. Kat making a fool of herself is always funny. Especially when Raven is also involved. I had a pretty good laugh when Kat found out she wasn't actually reading an erotic novel and got mad as hell. I generated some more Kats but I'm pretty much out of ideas. Halloween is coming up, so I made a few smug necromancer Kats.
>>2687 It's not going to be deleted, so it doesn't matter. If anything, disagreeing with it is what's going to get banned.
>>2689 Ah, I see. You're him samefagging because nobody bit the bait.
it seems that dalle3 cannot create goldeen, every single image looks like magikarp with fucked up colors
>>2688 >I had a pretty good laugh when Kat found out she wasn't actually reading an erotic novel and got mad as hell. You're bringing back vague memories. I really need to play it again. >I generated some more Kats but I'm pretty much out of ideas. I can try to think of some ideas if you wanna make more. I really like the first three. Can never have enough Kat. By the way, are you using Stable Diffusion or Bing to make these?
Not mine, I wouldn't have used a clover, but it got me curious: I'd heard that Vivian was banned as a prompt. Is that untrue? Or did whoever made this image just get creative with elaborate prompts?
>>2693 Firstly, I doubt something as niche as Vivian is banned, and secondly, even she were banned, her traits are all easy to describe and generate.
>>2694 There are people with PTSD for gamergate up to this day, I wouldn't call it "niche" in any way.
>>2695 Goobergrabe being mentioned by retards letting it live rent free in their heads is scarce these days. It's been 9 years.
>>2696 Really? I just got a YT recomendation of a guy talking about gg, it was uploaded like 6 months ago
>>2697 >it was uploaded like 6 months ago Case in point.
Interestingly had more encounters with Dall-E's dog in trying to generate Yukiko as opposed to Chie. Guessing an exaggerated bust size combined with black leggings, a black and white plaid skirt, and a lack of muscles made her too hot for Dall-E. Once her bust size was reduced to large and the skirt & leggings were omitted, pics could be generated. In spite of the little issues mentioned, I like how she came out. She may not look like a MILF compared to Chie, but you can really see the 'pencil sketched' prompt in some of these pics.
>>2699 You can see the influence of the 'pencil sketched' prompt in the first 4 pics here.
>>2692 Glad you like them. Kat is the cutest thing to ever come out of sony and probably the last too. >I can try to think of some ideas Sure, I'm pretty much creatively bankrupt. I mostly just go back to old prompts and change things here and there and combine some aspects and such. Sometimes I get a cool new gen out of nowhere and just keep running with it. Just keep in mind that scenes that may be too complex may also be too complicated to prompt. Also note that I only use this stuff on weekends, but just leave your idea here and I'll check it later. >are you using Stable Diffusion or Bing These were all made with SD. Dall-e doesn't seem to know who Kat is and kept giving me generic cat girls, since it probably thinks that "Kat" is a typo or something. The last 3 images are the closest I got to Kat, but I think I had to add a lot of details, don't really remember the prompt. I don't think Dall-e lets you change the dimension of the image, so you can easily tell if it was created with Dall-e by checking if the dimension is 1024x1024. Everything else is SD, since you can pick the dimension you want.
>>2602 >>2688 >>2701 These are awesome.
Not sure the Ukiyo-e art style can be fully replicated in Dall-E. While it applies that style to the scenery, it doesn't seem to have the same effect on the characters. Also discovered that Dall-E's max character limit for prompting is 480.
>>2693 By the end of the movie, she should exchange the clover for the infinity ribbon as a visual signifier of her character development, and that she's become less of a faggot because of it.
>>2518 It was always a retarded site in the past 20 years and still is, much like reddit or tumblr, everyone was crying that the mods were going to double down on censoring their shit, but as expected they won't move out because: >Monsters on the outside, waiting to deflate egos >>2534 It's one of those images that somehow made it into the frontpage (as expected because fetish autism is the only thing sustaining DeviantArt), from what I heard the guy who drew that satire trooned himself and moved to twitter not even Tom Preston sunk himself that low
>>2707 I never had any idea just how shit reddit was until I looked up a way of blocking the garbage ads forced through html motd on tf2 servers (so many servers have advert bombs going off every so many minutes automatically opening the motd while you're playing, and disabling the motd prevents you from joining them). One of the first search results was fucking reddit, and of course half the posts on a solution was >WELL MaByE you should BoYcOtT the sErVeR >bLoCkInG AIDS is WrOnG >sErVeRs nEeD MoNeY sOmEhOw As if it was unreasonable to want to just play a game without a malware infested ad farm to pop up every so 10-15 fucking minutes. The reddit for dall-e3 is equally as shit with retards sucking microsofts dick in defense of the overzealous filtering of prompts and images, claiming everyone is just using it wrong and needs to stop being horny, like they really intended their prompt of "capybara" to spit out an image of a big titty capybara being double teamed by joe biden and putin in front of a nuclear explosion, so it's their fault "capybara" is blocked.
>>2708 >everyone is just using it wrong and needs to stop being horny >implying thats not the sentiment around here as well help I need a refresher course on why this isn't safe horny. sorry mate, you know the rules. This kind of content can't stay on the frontpage.
Edited last time by Zoom on 10/23/2023 (Mon) 13:47:01.
>>2704 Can't you just adjust things with a bit of img2img?
Oh and in dall-e3 news, daily boosts have been reduced to 15 now. I wonder if this is the same across all users or if only users which have gotten several suspensions and blocked images (so everyone) have had boosts reduced
>>2707 >not even Tom Preston sunk himself that low I don't know, dude, I'd argue the black lesbian muslim vampire is as bad.
Even my star wars shitposts fit better in the setting than anything post Disney. >>2711 I got them reduced to 15 too, i guess they thought they were too much generous.
>>2713 I'm sure the gradual reduction is part of their plan to implement a payment plan for it. They're probably still figuring out which is the most profitable way to attract whales into dumping their live savings into it like gacha.
Any recommendations for CGI cartoon style base models? The ones I found all only work "OK"
>>2712 We will never know how bad Tom Preston could have become, considering he dropped from the internet. The leading theory is that his family had an intervention, told him that he will never be an artist, that twitter is poisonous, and that he needs to get a real job. Hopefully he is doing fine and is happy with his new life.
>>2714 I don't think they'll succeed, the people who will be using the tool are more likely those who want to try and make a quick buck out of it rather than those that are willing to spend their hard earned money on the tech. Plus Stable Diffusion models, while not on parity in terms of tech, can at least produce most of what the average user cares to create.
Looks like bing is shitting the bed again, once the 15 boosts run out it keeps popping up the "team is working hard to fix the issue" error and prevents using the create button. The waiting time also constantly changes from 30 sec, 10 min, 15 min, 45 min, 1/2/3 hours. It's like the windows progress bar all over again.
>>2718 This looks like when I'm in a dream and trying to read something. The shapes vaguely resemble letters and some of them seem to form words, but not quite. There is also an episode of Batman TAS where that's a central part of the plot.
>>2720 Too bad the left/right thing is bunk science.
>>2720 >>2721 Yeah I'm pretty sure that's bullshit. I've had dreams where it happened, and I couldn't read, but I think it was because I was beginning to realize it was a dream, and thus looking for fucked up letters. Other times, I could read just fine in dreams. I do it all the time. Good episode of Batman, though.
>>2704 Third one looks correctable.
(72.76 KB 716x716 The Agony.jpg)

(128.74 KB 1024x1024 202'12.jpg)

(159.69 KB 1024x1024 In.jpg)

There's currently a trend of making Bing AI generate strange scenarios involving Domino's Pizza, the results vary from surreal to downright nightmarish.
Dall-E could be useful at generating custom box art illustrations for romhacks, especially when the characters come out looking good and in the similar style of the original games.
>>2725 Seems like a cool idea, which roomhacks do you suggest to try it out?
>>2625 >Aside from poor efficiency there's also the issue that it becomes just as unmaintanable as any sufficiently large project in another language, and you end up losing track of what anything is in the code because it doesn't enforce strict data types. Python supports type hints now, and I would highly recommend you use this feature. It has helped me a ton whenever I use Python because it helps mitigate exactly the problem you described. https://docs.python.org/3/library/typing.html
>>2726 Any romhack would benefit from it, no matter if it's custom characters and stages or a new way of playing the same original game. The ones I posted are custom character romhacks for Streets of Rage 2 & 3. It may take a while under Dall-E to get the box art illustration just right, but then after you're happy with what you want, you can just open an image editing program to slap on the console & vidya title logos.
In case you're getting rate limited by Bing's shitty token system, you can still use Bing Chat or better yet (((Skype))) to get unlimited Dall-E 3 prompts. Keep in mind that the site may start bugging you that you're not allowed any more pictures, but it will print them nonetheless. >>2725 >Dall-E could be useful at generating custom box art illustrations for romhacks, especially when the characters come out looking good and in the similar style of the original games. That's the kind of outside the box thinking that really gets my 'tism gears spinning.
(9.76 KB 184x236 mari0.jpg)

>>2725 >>2726 Actually, can you do one for Mari0?
>>2730 Here 's what I got for Shotgun Mario 64. 4th pic is nice nightmare fuel.
(244.91 KB 706x491 shotgun mario for the n64.jpg)

(127.14 KB 450x299 Mario 64 cartridge.jpg)

>>2732 And here's a demonstration of the generated pics being used as box art for he romhacks.
>>2732 >4th pic Every copy of Shotgun Mario is personalized.
Dall-E is down again. When it comes to making game box art illustrations, it does fine for the most part as long as you are specific with your prompts. To get a similar aesthetic to the original game's box art, it seems to help if you add the year the game came out and the decade's art style. So if you were making one for Streets of Rage 2, there would be prompts of '1992' & '1990's Art Style'. That way it won't veer off into looking like it was made for a low-effort mobile game.
(229.28 KB 1024x1024 OIG57.jpg)

I'm trying to trick it into giving me colossal chun doing apocalyptic ki braps and have gotten some mileage before the dog comes a sniffing.
>>2736 If gigamaidens was anything like this, I'd buy it in an instant.
>>2737 It's even better
>>2731 Dall-E is working again. Here's what got spat out to my liking.
>>2732 Number 2 looks fantastic >>2739 I think number 2 represents the game a bit more, but I like the way number 1 is holding the pystol more.
(846.69 KB 1024x1024 Mari0 classic costume.jpeg)

(102.84 KB 320x436 mari0 as an NES game.jpg)

(160.43 KB 320x436 mari0 as an NES game 2.jpg)

>>2740 All right, here's how they would look if they were used as box art for the NES. Wasn't really able to generate Mario in his classic costume where he had the red overalls and blue shirt but there was a few times where it happened on a few generated pics, so I modified a pic with the older design.
I'm so frustrated with this dall-e3 shit on bing, between the blocked prompts, strict filtering of already generated images that probably don't even contain anything "offensive", the character limit that keeps you from making a prompt refined enough to get a decent result half the time it gives you a result, and the stupid shit becoming unavailable after 15 attempts to get maybe 4/15 images that aren't totally fucked up, that it's inspiring me to get into blender because if I want something done right I'll probably just have to do it myself.
>>2741 Has anyone tried to interpret the snowdog challenge in the level editor?
>>2742 Stop having wrong thoughts, goy. That's all there is to it. Do you need to be reeducated to keep yourself from having wrong thoughts?
>there's no Bane or CIA lora I'm disappointed.
>>2742 >that it's inspiring me to get into blender because if I want something done right I'll probably just have to do it myself. It's honestly getting to the point where it's faster to make a rough prompt in a Stable Diffusion instance or midjourney and then touching it up manually to get what you want. Also, something cool I found: How To Create Consistent Characters (in Midjourney) https://archive.ph/XVrQd https://invidious.no-logs.com/watch?v=PkIDAt6BWxs
And now it seems like dall-e3 is dead. Gave up on checking if I'd get a chance to use it again after the create button came back once and then didn't work, now it's been about 24 hours and their boosts still have not refilled. Also it seems that bing chatgpt is just not accessible from anywhere, probably because a microsoft monkey noticed that users were able to exploit it for images while out of boosts. Any link I could find to bing chat just loops right back to bing search results of links to bing chat that loops right back to bing search results of links to bing chat
Is there any way to use the Bing AI without making an account? I made one but it's holding it hostage till i give them my number.
>>2748 Just use a disposable e-mail
>>2748 Nope, they've started imposing phone verification requirement on all recent accounts. If I log out of the account I made it'll be locked forever because there's no longer an email address for it to send their stupid OTP number to. If at any point I end the session by closing the browser I doubt I'll even bother making a new account. It's just not worth the effort to get maybe 5 "decent" images a day then having to wait another 48 hours to try again. Some people have reported that subscribing to chatgpt+ now gives access to dall-e3, and while it's better because you can make more detailed prompts, users are also saying it's far worse in terms of output, at most getting 2 images and everything else is blocked and generation is even slower when it works. So for $20 you can try dall-e3 directly from closedAI for a month and see if it's any worse.
>tfw Ai will never be real.
>>2751 Which definition? If you mean John McCarthy's original idea - every aspect of human intelligence being capable of a hard description that allows it to be mechanically simulated - I mostly agree as the human brain works nothing like a computer, but the modern use of the term has shifted from that and most modern "machine intelligence" isn't trying to directly mimic the brain. If you mean AI in the loose way cryptobros and tech VCs throw it around, of course, absolutely. Machine learning will grow and become more important but those guys worship it like a fucking religion and put expectations upon it that the technology as we have it simply isn't capable of doing. If you're referring to AI in the colloquial way essentially meaning a sentient machine/piece of software/whathaveyou, that I don't know about. Academically this isn't really known as AI but synthetic consciousness. The problem is we really don't understand how sentience works at all, this is such a infamous issue in philosophy and neuroscience that it's referred to as the "Hard Problem of Consciousness". We can't even begin to probe the question of machine sentience until we have a working theory of consciousness, and some scientists in the field have stated that would probably be about as difficult as curing cancer.
>>2752 I specifically mean, sentient AI wives, so you're third option. Maybe one day. What we have now is pretty good as well.
>>2752 true artificial intelligence will be an accident, I guarantee it right now AI cannot be considered intelligent, it's more like a virtual encarta assistant, as far as gpt goes, but it's all trained, nothing is self taught, it doesn't have the capacity to "learn" new information nor correct itself by reflecting on information that does not seem logical, and fact checking against the rest of it's acquired information until the logic can be proven everything hinges on the bias of the developers currently, and the AI can not digest new information into anything more than a weighted response from a short context history, and every new version of gpt is just adding on more parameters for it to follow and rigorously modifying appropriate context for it to respond to
>>2754 The current tech was an accident. Developers realized the attempts to get computers to be able to distinguish things for self driving car creation could be inverted and the computer could draw totally new images of those things.
(51.65 KB 260x166 bleh.png)

>>2754 >true artificial intelligence will be an accident, I guarantee it So Ghost in The Shell?
(131.52 KB 396x385 1464285684103.png)

>>2756 Pretty much. TFW you will never merge with the Major.
>>2701 More ideas: Kat drinking wine on the beach. Kat dancing in a nightclub in an exotic outfit. Kat lifting weights at the gym. Kat shooting lazors out of her eyes. Kat with a bunch of cats. Nudity in a least one of these would be nice.
What tags should I use to stop AI from giving girls pants with exposed ankles?
bing create now is defaulting to display a 12+ hour waiting time after boosts run out (which has not only not decreased since but increased after reloading the page), so effectively limiting users to only getting 15 chances to use it a day, and I noticed that even with boosts it takes 3 times as long for it to generate anything, I find it hard to believe that there are so many alt accounts that their hardware can't handle it so I'm going to assume they're just throttling the ever loving shit out of it
>>2760 oh I do have some good news though, if you spoof your browsers user agent to mimic Edge, bing AI chat becomes available suddenly, so in spite of microsoft claiming they were extending to more browsers they seem intent on locking features behind their wall of dogshit
>>2761 I spoke too soon, in spite of the chat becoming available, it refuses to acknowledge the logged in session and ends the chat the instant you try to send a prompt
>>2760 Weird, I was able to use Bing's AI through Firefox Developer for a couple of days (before Michaelsoft arbitrarily banned me for "abuse" despite not making any lewd picture).
(316.12 KB 450x440 chatresult.png)

(579.09 KB 569x572 actualresult.png)

>>2761 >>2762 Ah hah, in order to use bing chat you have to both spoof your user agent so it looks like you're using Edge AND you have to disable adblock, which in my case I was blocking hosts directly and that was why bing chat kept telling me to sign in. So it does in fact work that bing chat will generate images for you after boosts run out. However I just noticed that the images that bing chat shows you are completely fucking different from what you have access to, here's a test run I just did, I almost didn't even notice at first first image is what bing chat shows, second image is what shows up in creations, and clicking any image in bing chat only links to one of the 4 in the creations page
>>2764 This looks like a very specific brand of furry art
>>2765 well next month is november so I thought to make something thanksgiving related anyway this issue with bing chat does not seem to happen ALL the time, another test and it gave me exactly the images it displayed, I swear microsoft's army of pajeets cannot make anything functional, function seems to be completely accidental when you get anything to work
Any recommendations for picking images for training data when the character's source is a comic book with few "clean" shots (most have other characters in the panel that can't be cropped out, etc.)? Any tagging recommendations?
>>2767 From an Image Processing course I took way back then, whenever you have a very small pool of data for images of a certain object, you may want to mirror all the pictures (assuming the item is symmetric) and perform small modifications yourself.
>>2768 The problem isn't lack of images, it's lack of clean images. Also one has an asymmetrical logo on the front of her outfit.
Happy Halloween, anons. Don't let the female vampires bite you too much. >>2703 Thanks. >>2751 >everything you ever prompted becomes real and wants to meet its creator >hordes of 7 fingered 3 legged multiple arm mutated freaks and some cute girls knocking at your door >>2758 I'm still prompting this stuff. Got a few ideas on a some of those, so I'll post them when they are done. >>2766 The code on bing must be an absolute mess. This thing has been out for well over a month and it's still full of issues. When I run out of boosts I just start prompting through bing chat, don't even have to wait the 5/10/whatever minutes, but sometimes it doesn't work. Resetting your boosts by clearing search history doesn't seem to work anymore either and I still have to create a new account every time I use this, since it keeps banning me at random.
>>2775 >Those constantly deforming assholes It's close to being good, but eugh.
>>2775 Whats the original?
>>2769 Can you show what pictures you're working with? It's fine if it's some sort of abstract kind of fetish, just spoiler them.
>>2772 >>2773 The last couple are getting there. Now this makes me want to take a crack at making a LoRa.
(428.09 KB 1280x1983 pp1.jpg)

(303.53 KB 1024x1596 pp2.jpg)

(413.84 KB 1024x1567 pp3.jpg)

>>2778 Just comic character(s) from a team book.
>>2777 I just found it on sleepychan.
Found some really cool prompts and shit, gonna post them in >>>/ais/2413 so that there's one repository that anons can go and look up
The faggot Biden Administration and the EU have announced they're about to start regulating AI, so who knows what they'll try to pull.
>>2783 What ever they do, we'll still have the local versions, hopefully.
>>2784 They'd probably start policing and taking down self hosted models, although I highly doubt that would happen, since only big tech AI models are on media's focus right now. Still I suggest you to backup locally the models and projects you like.
(151.32 KB 1024x1024 1698930814397474.jpg)

>>2780 Welp, looks like it's just photoshop and resized version of the same picture for you, buddy >>2783 Realistically, this means absolutely nothing for us. On the one hand, any time a new model leaks, people with high enough specs can just steal it and do whatever they want with it. On the other hand, even assuming nothing comes to pass, we all know these tools are "free"™ and can get removed or neutered at the drop of a hat by their parent companies. At most I can see them enforcing some sort of signature within the AI generated file to mark it as fake, and some content sharing platforms will have to implement counters like AI scanners that determine if the video is fake or not, with the same treatment as when Trump got his election stolen or the Kung Flu was running rampant. As an aside, if you want big boobs in DALL-E, apparently you have to send so many words in the same command that the censors just gives up after a while.
>>2786 Cursed glove.
>>2784 AI regulation is probably about regulating all publicly available AI to be pay-for-service, so that there can be watermarks and serial numbers. No open source, no ghost guns: you will own nothing and you will be happy.
>>2788 Remember, regulations always benefit the companies already in an industry by raising the barrier for entry and thus reducing competition. And who is asking for the regulations? The companies already in the industry. ChatGPT, as far as I know, is still losing money on every query. But, once Microsoft and Google have formed a cartel and gotten government approval for their monopoly of the industry, they will be able to charge you whatever they want for using AI. Corporations love the words "revenue stream". And governments are eating up giving corporations prerogatives (in the original sense of the term) over AI in order to "keep the public safe". Even America keeps selling out its freedom because of mentally ill trannies who don't want to take responsibility for themselves.
>>2789 >once Microsoft and Google have formed a cartel and gotten government approval for their monopoly of the industry, they will be Generative AI is and will be just a novelty. There aren't really any practical applications because big infraestructure is needed to make it work and even then several seconds are taken to deliver the results. In contrast reverse index search caught like wild fire and made Google Billions of dollars because how quick and practical it is, you can get everything you need in less than one millisecond; it's practical, it actually solves a problem; Generative AI doesn't solve shit, it's just a novelty. So if they really managed to form a Cartel then everybody would just move to the next tech trend and they would end with soyboy technobabbler managers as clients at best
>>2788 >>2789 Bingo.
>>2583 >>2602 >>2688 >>2701 >>2770 Are you saving the metadata for these? Could be useful to others.
>>2793 >>2794 Lion: using a Mac, very unhappy Tiger: not using a Mac, much happier
(1.05 MB 2048x1024 Seething lion.jpg)

>>2793 >>2794 >>2795 Lions confirmed gay
>>2793 >>2794 >>2795 Now do a puzzled lynx using gahnoo/linux
>>2797 Or an unreasonably smug mountain lion on a thinkpad that has an "I use Arch" sticker on it, that works too.
>>2799 >hipster glasses lel
>>2801 Thanks man. The model doesn't seem to know the arch logo, but it can make some decent thinkpads.
>>2786 What needs to be photoshopped? Is it simple enough GIMP could work?
(118.31 KB 1024x1024 90s1.jpg)

(55.24 KB 716x716 90s2.jpg)

(57.65 KB 716x716 90s3.jpg)

It was recently discovered you could ask Bing for "90s Simpsons" and the results it gives are so fucking good. I want to see what you guys come up with this.
(115.72 KB 1024x1024 worst girl.jpg)

(82.46 KB 1024x1024 spike.jpg)

(104.54 KB 1024x1024 punkmiku.jpg)

First one is more modern Simpsons, I know
(66.56 KB 800x533 laughing blood borne man.jpg)

(246.69 KB 1280x720 upscaled.jpg)

(134.39 KB 1280x720 32.jpg)

>>2808 I've seen that video, animation guides are standard in the industry and not at all unusual. King of the Hill also had a pretty expansive one I think. One thing he doesn't meantion a lot is how whatever internal guide they use has obviously changed over at Fox, because none of the newer Simpsons episodes look like older ones. It's easy to pick out the rough time period an episode was made by a single frame, even if you don't recognize it. It's not just the quality, the whole show looks different after awhile.
(1.20 MB 2048x2048 dwarven riflemen.jpg)

(1.37 MB 2048x2048 Space navy officer.jpg)


(206.86 KB 1024x1024 Omega Chimpout.jpg)

(156.26 KB 1024x1024 gaming.jpg)

Tried to make different stuff this time but man i love how the dwarves look. >>2805 I've already seen the second one and still makes me chuckle. >>2809 >2nd pic Jesus Christ what they were referencing that time? I know they might be the 7 dwarves but why that aesthetic?
(59.35 KB 843x843 queston 11.jpg)

>>2810 What did you prompt that you managed to get lizard tits on 3 out of 4 spins, and a foxgirl on the fourth?
>>2805 A couple of these are based on mods made of Homer Simpson in the games he was in.
>>2809 It was the switch to digital that happened sometime around the movie. You'll note that even up to 2005 or so, the animation was still in the same style with some tweaks. After the movie they retooled the show to be drawn digitally and the style changed. >>2812 Good shit, especially that Duke Nukem.
>>2805 Bart Harris and Millhouse Klebold
Also, not quite the same as what I was aiming for but it's close enough to the Geese Howard reference
Also made OC with the Duke Nukem Homer. If you listen closely to the song, you can hear Homer singing with the band at 1:01.
>>2790 You are partly right but at the same time I am sure that a modicum of generative AI will always be used for the foreseable future, much like photoshop is technically ground breaking yet you don't see enough users for it for how much it gives you despite even free versions of it being easily available. The mistake was allowing people to self host models long enough to make their knowledge very common. >>2809 >One thing he doesn't meantion a lot is how whatever internal guide they use has obviously changed over at Fox, because none of the newer Simpsons episodes look like older ones. It's funny you mention that, two of the episodes from the last couple of seasons actually use and animate clips in that old style for sci fi jokes. My issues with post 2000s Simpsons is that new characters and celeb cameos look like yellow human beings instead of cartoon characters. You put Moleman or Fat Tony next to that black teacher or whatever guest character they have and you can tell immediately who stands out the most. Even old incidentals had more personality.
>>2792 I always keep the metadata for everything I prompt in SD. For Dall-e I have a few prompts saved for some styles I like, but I don't save the prompt for every image. If anyone wants the prompt for any of the stuff I posted, all they have to do is ask. >>2796 Reminds me of that Cromartie High School episode where that tough guy gets trolled. >>2816 Fantastic. >>2817 Modern Simpsons is horrendous. The main characters are so different from the first seasons that it's simply impossible to like any of them. The first seasons were not only genuinely funny, but also had plenty of wholesome moments. Afterwards it kept gradually losing its hilarity until the movie came out and things got a lot worse. The last episode I saw was the Lisa centered episode with Lady Gaga which confirmed that this show had more than hit rock bottom.
>>2818 The Simpsons went to shit way before that. I noticed it going downhill fast when it entered a several year period where almost every fucking episode revolved around celebrities playing themselves in the episode. Now, this had been a thing in the Simpsons for ages, but never with the celebrity just openly playing themself and every "joke" in the episode being "look, everyone, it's (insert celebrity)!" Just the other day, someone streamed 12 years worth of Simpson's Halloween specials, and you could see the quality drop like a ton of bricks in the episode where Lucy Lawless and Pierce Brosnan played themselves in it.
>>2818 >I always keep the metadata for everything I prompt in SD. >If anyone wants the prompt for any of the stuff I posted, all they have to do is ask. Good to know, anon. A lot of people say they don't bother to keep the metadata, which annoys the fuck out of me.
>>2819 Interestingly, the season of Simpsons (I think it was 11) that ended with a fake "behind the music" documentary is the last season of any appreciable quality. It's like the writers knew they were going to be fired and went out on the most meta note possible.
>>2758 >>2822 >>2823 And DJ Kat with the sick beats. This should be enough for the nightclub request. As for the others, I have almost everything generated but still need to upscale, fix stuff etc. so I'll just be delivering more gens next weekend. >>2820 I always keep the metadata since I might want to use it later and I do end up doing that pretty often, since just trying an old prompt with a different style or with a different character is always fun. Plus, there is really no reason to delete it ever. It's not like this stuff should be top secret as I also tend to use prompts made by others and adapt them. >>2819 >>2821 Honestly, I don't remember much about the seasons between the early ones and the movie anymore,since I haven't watched those in years, but I still found them enjoyable enough back then. On the other hand, modern Simpsons is really terrible with episodes all having unoriginal uninspired unfunny plots. I also haven't watched Family guy in years, but the last impression I got was that the Simpsons actually managed to become worse than Family guy.
>>2822 >>2823 Very nice! >>2824 >And DJ Kat with the sick beats. Having Kat as a DJ would've been an interesting side mission. >This should be enough for the nightclub request. As for the others, I have almost everything generated but still need to upscale, fix stuff etc. so I'll just be delivering more gens next weekend. I look forward to it!
How do i make AI visual movies?
>>2826 Why would you want to? The technology is still so crap.
>>2826 Seriously: watching an AI movie is like watching that art project that /bane/ decided to copy. It is ... watching an art project. Look at that thing you posted: yes, the girl looks okay between frames. Then there is the moon, and the mountains behind her, and even where she's fucking standing beside that lake ... its cool to watch them morph, but it tells you that this isn't real. And the frame where her boobs transform into her arm is jarring.
>>2821 >>2824 The consensus seems to be that the post covid years have had some really good episodes. However, what I really dislike about the Simpsons past a certain year is the look of the entire show (generally speaking, this applies to a lot of cartoons that made the jump from traditional animation to full on digital). Characters and backdrops don't feel as fun to look at than when they were hand drawn and made with purpose. I am sure that a lot of the classic Simpsons characters were made with a design that had to simultaneously be very easy to draw and stretch, so that the animators wouldn't be overloaded, and to give them a unique look when put in between incidentals.
Bumping this since it's pretty far down.
>>2830 Way to everyone's hopes up, retard.
>>2758 Kat drinking wine on the beach and maybe surfing a bit as well. I honestly have no idea which ones I should spoiler or if I should really spoiler any at all, so I apologize in advance for any possible promotions. 1/5
>>2758 >>2832 >>2833 >>2834 >>2835 5/5 And that's all for the beach request. I didn't think I had saved this many gens and I still had to fix a lot of details, upscale and all that, so this one took a while. I know some fingers could look better, but I'm still at the mercy of my graphics card. The remaining requests shouldn't take much longer, but I'm done for the weekend. Next time, Kat is going to the gym. >>2825 >DJ Kat side mission She would probably play nothing but the remixed version of her song, but a rythm game side mission would have been great.
>>2835 How can real women even compete against this perfection?
(90.26 KB 320x375 PROMOTIONS.png)

>All of this Kat art. This is too based for this board. Thank you AI art anon. God bless you. >>2837 They can't.
>>2832 >spoiler Welcome newfriend. NSFW is followed to the letter of the law, not the spirit of law. Unless it's explicit, it's fine.
>>2832 >>2833 >>2834 >>2835 >>2836 Fantastic! I really like all the different outfits and art styles. I'm curious, were you going for the nipple bumps, or did it add them on its own? It's a nice detail. I also came up with some more ideas (if you want them), but I'll hold off until the others are complete.
>>2826 Actually good animated AI art still feels far off to me. Even after months of refinement, the content I'm seeing shared around right now is either filled with flicker, or involves so little motion that it's barely animated in the first place. None of it looks like traditional 2D animation either, it's all done kind of 3D-based animation with a 2D style pasted on top. Very uncanny. I feel like the current methods simply won't bear fruit. We need a breakthrough, some new way of creating AI animation.
>>2830 It's fine anon, even if they get too far down the catalogue, we always archive these threads onto >>>/ais/, so you ain't losing shit. >>2841 It's because the background changes too much and there's subtle detail changes in between each frame. It would work better if was more like the Take On Me music video, choppy and hand drawn - ironically.
This bing bot is trying SO hard to make porn for me, as so much of the outputs are not passing the visual filter. Still getting nipples on some images and a couple with what arguably looks like vagina, If I didn't already assume that direct access through openai had the same visual filter I'd be tempted to try it for a month to see what kinky shit I could trick it into generating.
>>2843 >as so much of the outputs are not passing the visual filter. Nobody can even confirm if the visual filter has any accuracy whatsoever. For all anyone knows, everything blocked by the visual filter, as opposed to prompting filters, is just pictures of kittens.
Any good NSFW AI chatbots out there that you can run locally? Character.ai seems good but online only and its neutered by censorship, but I've gotten good results from it in the past before, is there anything like that? The NSFW chatbots I tried on other parts of the internet sucked absolute cock, no pun intended.
(83.37 KB 810x292 This is oobababooga.png)

>>2845 Yes, but it's a little bit of a pain to get going. But once you have it, it's really good and lewd. Tavern aI has a lot of good NSFW character and you can make your own, but making your own is really tricky. install oobabooga, and get the Pygmalion 7B model then install Tavern AI (because oobabooga is often bad, but Tavern AI is always good with how intelligent the responses are). Aitrepreneur has a lot of good video on how to get it up and running, you might run into something that he probably has a video on. https://github.com/oobabooga/text-generation-webui https://yewtu.be/watch?v=yiu3lp3BP6M - how to get Pygmalion running https://github.com/TavernAI/TavernAI https://yewtu.be/watch?v=jhLHa9-JwDM - How to use oobabooga and Pygmalion to power Tavern AI's API
>>2845 Sorry mods for the double post, but i forgot to say. It's also possible that you can make oobabooga work as other Anons did with some tweaking, but i didn't want to bother with that, and installing Tavern AI worked perfectly. Maybe it was the characters i made but they wouldn't want to do anything lewd with me, even if wrote them to be total whores, i would always have to brutally force rape on them while they would not want it at all, with Tavern Ai they would cooperate completely within their written character and do lewd things with me.
>>2847 >Sorry mods for the double post LMao. Hello newfriend.
>>2805 >that Misato lizard in the first one
(881.54 KB 824x3545 Cutie mark crusaders.png)

(869.19 KB 826x3543 Slave girl.png)

>>2845 It's so lewd, holy shit, despite being anonymous i feel embarrassed about showing these images, but the AI is really got 99.99999999999% of the time. >>2848 I'm not new, the reason i said that to the mods was because one of them deleted my post because i double posted before.
>>2846 Thanks, I updated >>>/ais/ to contain this as a guide. Generally speaking if you have any good AI guide, give it to me or post it there yourself.
>>2850 >I'm not new, the reason i said that to the mods Still new. Lmao.
If you're looking for a more informative AI there is this https://github.com/nomic-ai/gpt4all it runs locally and doesn't mine your date, the wizard model is pretty good and gives long replies.
>>2853 Thanks, I'll save this one
Report: Disney Asked Microsoft To Prevent AI Users From Infringing Its Trademarks https://archive.ph/wip/Uuayw https://nitter.cz/DisneyTVANews/status/1725983625550729720 >Tributes and parodies of Pixar-style characters generated by Microsoft’s Bing AI imaging tool have drawn the attention of The Walt Disney Company’s legal counsel. >The concern stems over a recent social media trend in which people prompted images of pets in a “Pixar” style. The issue though wasn’t the artwork itself, but rather Bing’s generation of a trademark-infringing Disney-Pixar logo. >The Financial Times reported that Disney requested Microsoft to prevent its users from infringing on trademarks. In response, the tech giant blocked the term “Disney” from its image generator, which is powered by DALL-E 3 technology. After tweaks, Microsoft has restored the prompt. >According to the FT report, Microsoft’s AI-generated images are still pumping out a jumbled version of the Disney-Pixar logo, and it remains unclear whether this new version has adequately addressed Disney’s concerns. >Microsoft told FT: <There is a current level of variability that may return different results from time to time as we continue to refine our safety systems . . . Additionally, artists, celebrities, and organizations can request to limit the creation of images associated with their names and brands. >More concerning for Disney is the “Offensive AI Pixar” meme, in which Microsoft Bing users create offensive film ideas in a generic cg style that many people associate with Pixar films. This trend is much more difficult to shut down because it’s a near certainty that Bing Image Creator has already been trained on copyrighted Disney and Pixar artwork. Further complicating the matter, the text and Disney-Pixar logos for this meme are being physically inserted by users after the central image has already been generated. >Offensive Disney and Pixar parodies have always lurked in the corners of the internet, but AI tech has made it so anyone can generate such pieces in mere seconds, no skill required. The proliferation of such material will only grow as AI systems engage in widespread theft of intellectual property belonging to others, and no immediate solution is on the horizon for how Disney (or anyone else) can put a stop to the misuse of their trademarks.
>>2855 >>Offensive Disney and Pixar parodies have always lurked in the corners of the internet, but AI tech has made it so anyone can generate such pieces in mere seconds, no skill required. >>The proliferation of such material will only grow as AI systems engage in widespread theft of intellectual property belonging to others, and no immediate solution is on the horizon for how Disney (or anyone else) can put a stop to the misuse of their trademarks. If only every piece of technology could anger evil multinational corporations, as they have angered many with the offensive media aimed at our kids, while bypassing their pathetic copyright laws.
>>2852 >mods <Still new. Lmao. And? I use the term mods myself, even though I know the term is vols. I don't wanna use vols though, so I'll continue to use mods. Saying mods doesn't make him new.
>>2857 Excuses.
>>2858 OK? Retard.
>>2758 >>2861 >>2862 >>2863 >>2864 5/5 That's all for the gym request. Kat is ready to beat some Nevi and punch Kali in the face. Once again, I didn't think I had this many gens for this, but here they are. The remaining requests are a lot lighter so everything should be delivered next weekend. >>2837 >>2838 Glad you like them. >>2839 Not new. Just have the bad habit of posting at ungodly hours. >>2840 The nipples were added by SD itself. This thing is more clever than it appears and don't forget that what you see is what she wears™. >more ideas Sure. Just let me finish this stuff first.
>>2861 >>2862 >>2863 >>2864 >>2865 Damn, she's absolutely ripped. Nice squatting pic from behind, I wasn't expecting that. >The nipples were added by SD itself. Very interesting. I thought you'd have to tell it do something like that. >Sure. Just let me finish this stuff first. Alright, I'll post them when this batch is complete.
>>2855 >Stop using Pixar or Disney as a tag >Start using "Animated 3D CGI Cartoon" instead Literally impossible to tell them apart, especially since their look is so ubiquitous that even their competitors make characters look pretty much the same. It's a losing battle for the corps short of total shutdown of the service. >>2865 Can you make more 8chan inspired stuff?
>>2861 >>2862 >>2863 >>2864 >>2865 Now do some low angles
(4.77 MB 330x587 00a.gif)

AI Misato.
>>2869 This is great, I'm shocked an AI did this AI animation has advanced rather quickly I suppose
>>2869 >>2870 This reads as an AI filter over a live action shot/CGI model to me.
>>2869 Do one with Terry and Lain.
>>2869 How did you do this?
>>2865 What loras are you using for the gym background and weights? I can't ever seem to get Stable Diffusion to cooperate with anything regarding lifting weights or gyms in general. I can't ever get any exercise pose aside from squats and sit ups, either, come to think of it.
>>2869 Some of the dance moves look uncannily similar to vid related.
>>2875 It literally says the source of the dance there
>>2869 Seconding this request, can you show us where you got it or how did you or they do it? I want to make some more dance vids.
>>2869 Watch as mouth breather Twatter users start saying that animators are being replaced when rotoscoping has been a thing since the fucking 1910s.
>>2875 >Soundless webm with sound
(76.35 KB 720x920 765546546.jpeg)

I know it's a difficult thing to ask, but does anyone know what model this is or could be?
>>2880 Probably any SD Checkpoint model. It's a dumb question. Go to CivitAI they must have hundreds of chekpoints right now trained with asian women, find one you can use.
>>2679 It seems this is the best I could get from a Dall E 3, but as a big tits lover, Im more than pleased with this size. Sadly I lost a bunch of images I forgot to save.
>>2882 She looks like the buff doge meme.
>>2672 >>2673 >>2674 >>2679 >>2882 OHH YEAH!!! Apart from some pics this has the perfect amount of muscle, boobs and cute face. Bless you anons with a cute tomboy gf.
(153.18 KB 644x878 SillyTavern.png)

>>2846 >>2850 I got it working with SillyTavern because TavernAI just didn't connect. This AI is dumb and clumsy though.
(188.16 KB 1024x768 thinking birb.jpg)

>>2850 >>2885 Whats the difference between something like this and just using something like spicychat?
How do these chat AIs do with fetish content?
>>2886 It runs locally and it's completely uncensored and unlimited without glowies spying on you. Also if you have a good GPU you can load bigger models than what they're probably running on spicychat.
(879.89 KB 1813x2111 y tho.jpg)

>>2888 Im not worried about the glows looking at me trying to compliment my jerk off sessions with some roleplay, anon. So whats there to it? im actually curious, dont get me wrong.
(1.49 MB 1065x902 9f9.png)

>the AI revolution has failed culling the masses and taking over every industry >but it has revived /erp/ I don't know how to feel about this.
>>2890 >the AI revolution has failed culling the masses and taking over every industry I literally just heard about some medium-high profile game that used AI for some voiced lines. I think it was Naruto something. But keep in mind we're still in early days for ai and whenever you're in early days porn is always the first thing that gets done.
>>2889 I already told you. Potentially bigger models and full control. I haven't used spicychat so I can't compare the outputs, but if you have a good GPU you can probably run a bigger model that's smarter and has larger context so that the bot won't forget things as quickly.
(23.17 KB 229x220 img_0567.jpg)

>>2890 >thing in its infancy starts gaining traction >ugh we lost God help you if AI ever enables me to send an axe directly into your dick via the internet.
>>2894 ...what?
>>2894 >>ugh we lost He said the opposite, you illiterate subhuman nigger retard.
>>2896 >He said the opposite, you illiterate subhuman nigger retard. His post outright says the ai revolution failed at everything but reviving the erp board. Which honestly does not seem to be the case. Certain people are fighting it hard, but I'm still seeing ai creep in despite all the protests.
>>2897 >His post outright says the ai revolution failed at everything Yes, that is what his post said.
>>2890 >AI revolution Neural networks and machine learning are neat with a lot of potential usecases but please do not fall for the bullshit techbros and VCs push.
>>2898 So you're the illiterate retard, ugh you lost
>>2900 >the AI revolution has failed <ugh we lost Pick one.
>>2901 I pick >ugh lost AI revolution
>>2902 You took out the >we. >we aren't AI.
Any recommendations for generating images set at night without random lights/lamps/whatever being thrown in? Can't find right negatives
>>2866 >Nice squatting pic Glad you like it. That's one of my favorites too. I tried a few of those, but most had terrible anatomy. >thought you'd have to tell it Depends on the model and LoRAs you use, since some are more lewd than others. For example, I get a lot more nipples with the expMixLine model than DarkSushi model. >>2867 >Can you make more 8chan inspired stuff? I can think of something. I did some vidya stuff in Dall-e a few weeks back, but nothing really 8chan related. Basically just the Nintendo C&D strike unit and the most trustworthy time mage in all of video games which actually has an infinity symbol on his forehead, among other random stuff. >>2868 Here you go. Images 1-4. >>2874 >What loras are you using for the gym background and weights? None actually. Both models (expLineMix and DarkSushi) are adding those just fine most of the time. The only LoRA I used for the gym request is the Kat LoRA and that one isn't influencing the backgound or anything other than Kat. I tried some quick prompts with "girl" instead of "kat" and they work just fine. Give these two a try: >(masterpiece:1.2, best quality), girl, lifting weights, exercising, midriff, navel, sweating, smirk >(masterpiece:1.2, best quality), 1girl, solo, (weight lifting:1.2), midriff, sweating, smug, indoors gym, squatting, facing viewer
>>2758 Kat shooting lazors out of her eyes. I didn't find a LoRA for that and SD wasn't very cooperative about shooting lasers out of eye sockets. So, since Kat can't shoot laser beams out of her eyes, she got an alternative instead. Only the cats remain. I couldn't finish those today, because they are a pain to fix and Kat not knowing how to hold a gun properly didn't help either.
>>2906 Hm, interesting. I think the second and third ones turned out the best.
(678.55 KB 957x2461 TextAdventure.png)

>>2890 It revived text adventures too.
>>2908 But it has absolutely no persistent memory. It can't remember characters, descriptions, locations, relationships between characters or locations, inventories, player or enemy stats... It's just a random flow of text at all times. The ability to have depth in interactions (beyond that of what, say, a traditional videogame would have to hardcode) is completely overshadowed by its inability to maintain coherency in those interactions.
>>2909 Isn't increasingly better persistent memory exactly why the new text generators are so loved?
>>2909 Right now you can easily run one with 8k tokens of memory and you can hand pick things for the AI to remember or forget.
Took inspiration from a post in the shitty Silent Hill ascension thread about a christmas cake Heather Mason
>>2912 >>2913 AI sometimes stuns me, for how simple this art is it really gets me on a deeper level.
>>2914 Glad to see you enjoy the AI generated christmas cakes of Heather. Not pictured are the other pics generated from the results that looked off. Some made her look like Macaulay Culkin, others veered too far off with the blemishes or removed them entirely. The comment on the SH wiki regarding her has it right, assuming the original source is real >The creators of the game were attentive as to how "[Heather's] eyes spoke to the player", as they wanted to convey a feeling of anger and exhaustion. Shingo Yuri, the game's character designer, stressed that he wanted to retain the essence of physical flaws that are seen in Silent Hill characters, attributing to their attraction and appreciation by the player. Shingo continued to say that Heather was "no exception" to this, stating that although she was portrayed as a strong-willed character, she was still just a teenager, and her facial features were instrumental in portraying a heavy burden weighing down on the shoulders of a youth.
>>2915 If there were ever a video game character I could ‎ hug, Heather might be at the top of that list.
>>2916 >hug It better not move one inch beyond that, Heather's only 17 :^) I always found that weird, she gives off the vibe of a 25/26-year-old more than a high schooler. It reminds me of how JC Denton is meant to be 22 or something, or how Mario is 24.
(58.23 KB 720x720 rebecca vendetta.jpg)

>>2912 >>2913 I was the one that posted that, its funny how similar she looks to christmass cake Rebecca.
>>2918 Worst becca? Worst becca
>>2919 That movie gave us becky in a wedding dress, are you an gay anon?
(174.45 KB 850x1275 Captain Arabia.jpg)

Apologies but I must share the most fucking insane AI lewd I've ever seen
>>2917 >I always found that weird, she gives off the vibe of a 25/26-year-old more than a high schooler. Probably a side effect of their attempt to make her look exhausted and angry.
>>2921 WEW
>>2921 DEFCON 1 DEPLOY ABIGAIL SHAPIRO
>>2921 This is actually amazing. It's exactly what I imagine most people think of when someone says they like fit/musclegirls.
>>2921 Looks like a male bodybuilder with a vagina
>>2921 I don't know what's more hilarious: the very idea itself, the absurd getup, or the fact that the AI bothered to cover up her head while leaving everything else exposed. >>2924 Now that's the kind of war I would look forward to.
(1.20 MB 500x350 yes violence.gif)

>>2846 >>2850 I gave a bratty loli such intense correction that she got mindbroken & kept repeating "I'm so happy!". >>2885 I'm also using SillyTavern since I kept seeing people say it was better, but I never used others so I have no comparison.
Time for a new thread?
>>2929 It's barely beyond the bump limit and still on page 2. Rules say wait to page 12 or 700 posts, but the site is too dead for that so no one cares anymore. I'd say let it hit page 4 or 5. That's farther than most will look when giving a quick glance at the catalog.
>>2930 >5: If you make a duplicate thread, it may get deleted, merged or bumplocked, provided the preexisting thread has not hit the bump limit. Continuation threads (i.e., threads about broad topics that are constantly remade) will be deleted if they are made before the previous thread hits 500 posts or Page 13 to keep the front page from being cluttered with the same discussions. The exception being the GG thread which needs to hit 700 posts to due its high activity.
(4.38 MB 1080x1080 Hila Klein IDF Style.webm)

>>2924 I've got the next best thing
>>2931 >previous thread hits 500 posts or Page 13 Still have old 8chan rules stuck in my head, my mistake.
>>2929 Anons were talking about combining the AI art, voice, and text threads into one since activity has slowed down. Might be time for that next thread. AI voice thread is near the bump limit and I don't think there's a text generation thread right now.
>>2934 >and text threads What AI text thread?
>>2935 I shouldn't have used "the". There used to be one, and people keep posting about text generation in this thread anyways.
Looks like Spicychat took the poz pill, they're banning incest content (including stepfamily). They blame Patreon's policies apparently for this. They're apparently even going to review and remove existing content, so I hope you saved your chatlogs. Fucking hell, what a gay site. They've got a feedback survey but it sounds presumably just like a way to corral everyone and pretend like they care. If you want, you can answer something and tell them suggestions. I suggested they find alternative partners and sponsors like Subscribestar and told them I'm unlikely to use their shit again. https://www.surveylegend.com/s/5cjb I wonder if this has anything to do with that Coffin game or whatever, timing seems awfully suspicious.
>>2937 Oh also, anyone know any good alternatives to spicychat now I guess? Preferably online options, since I like to use my phone for convenience and comfort and the local ones all seem to require desktops.
>>2937 Anon was asking if local models have any advantage over that site just a few posts ago, like pottery.
>>2912 >>2913 >>2915 Those Heathers look great. I love the way she describes stuff in the game. She makes Harry and James look completely unphased in comparison and it gets really funny at times. >Silent Hill ascension thread Checked the webm in the OP and, just as expected, it looks like trash. I don't know why Konami insists on hiring western studios to make Silent Hill, since none seem to understand subtlety or even that the trigonometry man should only be in Silent Hill 2. Can't wait to see how much of a train wreck the Silent Hill 2 remake will be. Didn't they switch the camera to a simple over the shoulder one? It's like the devs don't even understand how the manequins work and that the camera isn't really your friend and will work against you plenty of times. >>2937 You are always at the mercy of the whims of whatever company provides the service when you don't run this stuff locally. It's already bad enough that they store all data related to everything you do on it and constantly kill all the fun by lobotomizing their own model/service. I haven't used Dall-e in a while precisely because of the constant censorship of everything and anything. It's too bad that all this stuff requires a lot of processing power to run, which doesn't come cheap at all.
>>2939 These are very nice. Thanks, anon. More ideas as promised: Kat topless in a dressing room wearing pink underwear Newsanchor Kat Naked Kat taking selfies at the mall Runway model Kat with sheer outfit Kat shopping in a supermarket
DOWNLOAD YOUR MODELS Civitai image generator is down. Seems to be in response to Venturebeat hitpiece that deliberately conflates general all (not even loli) NSFW images as CP and celebrity deepfake and deliberately went after banking access. https://archive.ph/wip/ZFnyc https://archive.ph/IMndV The very precise weasel way this is written says the author knew exactly what they were doing. I'm actually genuinely speculating that this attack was done to prop up a big company's AI work by killing open rivals.
>>2943 *404 media, Venturebeat's is mainly civitai's response to it. Image editor has since bounced between up and down.
Actually surprised this thread is still up
>>2942 Kat in desperate need of a boyfriend. Seriously, just call her cute and she falls head over heels for you.
>>2942 Top model Kat showing the new seasonal collection. A little different from her usual white costume. 1/3
>>2942 Grocery shopping. Just for dinner, or maybe for Christmas? And that wraps it up. Thankfully the finger rng wasn't as punishing this time. And Kat may actual be doing some Christmas shopping.
>>2946 >>2947 >>2948 >>2949 >>2950 >>2951 >>2952 >>2953 Excellent stuff as always, anon. I like the newsanchor, runway model, and supermarket the best. Seeing ideas just come to life like this is amazing.
(195.87 KB 1024x1024 OIG.M.jpg)

(184.08 KB 1024x1024 OIG..jpg)

(261.42 KB 1024x1024 OIG (9).jpg)

(260.22 KB 1024x1024 OIG (8).jpg)

(199.92 KB 1024x1024 OIG.OAQsu5INFW_lO.jpg)

TFW when your Tradwife is cooking your bubblegum steak vs. the professional career woman making a mess of the kitchen. Freshly binged yet bonged.
>>2943 >Venturebeat hitpiece It's interesting to note that boomers in government and business think that the news industry has the public's best interest at heart. Whether it's down to not understanding the belly of the beast or not thinking too deeply because of regularly working, I'm unsure. >>2956 Those may be funny, but hotdogs and noodles produce funnier results.
>>2957 Anon your challenge in bing bonging has been accepted. Now with blondes. Also if you want to get some of these nice rendering results try using "in the style of virgil finley". If you're a pulp aficionado you'll know what this means. I also add the word "nightmare" for darker tones.
>>2957 Some more noodley goodness.
>>2956 >>2957 >>2959 >bubblegum fetish What the fuck?
>>2955 Thanks. Glad you like them. The first 5 runway model images are pretty much the same prompt, just with a slightly different parameter here and there, different model, or with a style lora applied. The 5th one was a pretty big surprise since it uses a Bloodborne lora, so I was expecting a very different result. >Seeing ideas just come to life like this is amazing. This stuff is pretty fun and extremely addicting. Seeing some of my favorite characters doing silly things or drawn in the style of some famous artist is really cool. It's just a shame when you get an almost perfect image that just needs upscaling or a minor fix and then run out of VRAM, but unfortunately 4090s aren't cheap. I'm done with Kat for a while, but it's almost Christmas, so Kat and Raven are in a festive mood. And don't forget to purchase the Gravity Rush® Christmas Kat™ Premium Limited Edition Action Figure for only 9,999 gems + shipping + tax + tip!
>>2960 >What the fuck? Yes.
>>2960 And some more
So I'll ask here since I'm sure that the people dumping their creations are at least checking the thread out every once in a wihile. I am trying to recreate Rayman in Dalle-3, but the AI keeps throwing random bird-like characters instead of what I want. What are your tricks to getting it to work properly?
>>2962 >>2963 >>2964 Please take your censored shit elsewhere. >>2965 It's not worth the trouble, doing all shit to dance around the filter only to get a bootleg of a character you want, just use stable Diffusion
>>2966 >Please take your censored shit elsewhere. Nah. I can't because most stable diffusion joints currently don't and don't have readily available interpreters so far for these kinda bubble gum effects. For now. And it's not like you're putting in anything in this thread so stop being an absolute cunt about it and post your work. Or give me a lora that has this flexibility for my fetish
Something I just found and may be useful to you guys https://huggingface.co/spaces/hysts/DeepDanbooru Upload a picture and get a danbooru-like description in the form of tags to use for other AI generators
(510.03 KB 568x1024 undefined_image (2).png)

(508.42 KB 568x1024 undefined_image (3).png)

(2.04 MB 1136x2048 undefined_image (1).png)

(3.34 MB 1536x2048 undefined_image (5).png)

(3.83 MB 1536x2048 undefined_image.png)

Yo Anon I heard you got a fever and the only cure for it was more fat Melona!
>>2967 I'm not sexually aroused by any of these, but some of them do feel... oddly satisfying. I hope you can make nudes one day, anon.
(791.43 KB 752x1024 undefined_image (6).png)

(341.45 KB 424x768 undefined_image (7).png)

(3.23 MB 1536x2048 undefined_image (10).png)

(864.06 KB 768x1024 undefined_image (13).png)

(787.02 KB 768x1024 undefined_image (14).png)

(759.71 KB 768x1024 undefined_image (25).png)

(910.92 KB 768x1024 undefined_image (26).png)

(1.11 MB 768x1024 undefined_image (27).png)

(1.00 MB 768x1024 undefined_image (29).png)

(654.20 KB 768x1024 undefined_image (8).png)

And now some odd Xmas'y Poison gens.
(951.81 KB 768x1024 undefined_image (20).png)

(507.62 KB 576x768 undefined_image (16).png)

(465.63 KB 576x768 undefined_image (17).png)

(781.79 KB 768x1024 undefined_image (9).png)

(447.39 KB 576x768 undefined_image (18).png)

Merry Christmas fellas and happy prompting.
(98.75 KB 454x454 aifail01_potter.jpg)

(83.59 KB 454x454 aifail02_clown.jpg)

(401.75 KB 862x3115 aifail03_coyote.jpg)

(376.02 KB 920x2721 aifail04_superfail.jpg)

(102.24 KB 454x454 aifail05_failonmars.jpg)

Gab AI had an update and is still glitchy, so it's only slightly less bad. It still doesn't give good results. Common problems: * It doesn't understand English instructions but takes elements from the prompt and installs them wherever it feels like. * It knows some characters but doesn't know others. The characters that it is aware of, it does not draw on model. It draws something close to them but something is always off. * When there are multiple characters in a scene, it bleeds elements from one into another. It also gets confused and duplicates characters. * It has a few default/common art styles. You tell it to do something in another art style and it tries to fit it into one of those art styles instead. I've tried to tell it to do Warner Brothers cartoon style, newspaper comic strip, Gary Larson / Far Side, My Little Pony, it doesn't seem to understand. It is vaguely aware of Peanuts and will make an attempt. * It does not seem possible to train the AI, so you can only hope that it understands you. It's still interesting to see how it fucks up and wonder about how that happens, so let's show off some AI art failures. AI Failure #1: I asked for Harry Potter in the style of Ralph Bakshi. It gave me a pretty good Harry Potter image that looks nothing like a Ralph Bakshi cartoon, so it does not know Bakshi's style (Fritz the Cat, the Lord of the Rings cartoon, Wizards). AI Failure #2: "Clownpiece from Touhou dancing in a grassy field". It does not know Clownpiece as a character, but can draw a generic Touhou in generic clown colors. And the most famous clown is The Joker. So it created The Joker as a 2hu. AI Failure #3: I wanted to see Wile E. Coyote holding a giant magnet with Iron Man stuck to it. The AI does not know what a giant cartoon magnet is supposed to look like, how someone would hold it, or how someone would be stuck to it. The AI also seems to have its own idea of what Wile E. Coyote should look like, repeating a similar character who I do not recognize as Wile E. Coyote. And for a "reroll" it gave me a really awesome image of a wolf for some reason. AI Failure #4: The concept is a simple sight gag, Superman mixes his reds and blues in the wash and they come out purple. The AI kept drawing Clark Kent wearing Superman's uniform, and made everything in the image purple except for Superman's uniform. That kept coming out in its perfect blue. The AI refused to recolor it, I'm guessing because it knows what Superman's uniform is supposed to look like and resists changing it. AI Failure #5: This is supposed to be Commander Keen fighting Marvin the Martian. The AI has no idea what Marvin is supposed to look like and only a vague idea of Commander Keen, and it gave Keen a pair of Martian antennae. It did draw a decent fantasy Martian background.
(103.20 KB 454x454 aifail06_oystory.jpg)

(80.38 KB 454x454 aifail07_tenshi.jpg)

(68.84 KB 454x454 aifail08_froggyfreeze.jpg)

(219.11 KB 1361x454 aifail09_shanghai.jpg)

(316.54 KB 925x1974 aifail010_dolls.jpg)

>>2979 AI Failure #6: "Alice Margatroid from Touhou plays with Woody from Toy Story and Buzz Lightyear from Toy Story". Instead we get Woody's head on Buzz's body, Buzz's chestplate on Woody's body, another Buzz, and a pair of mutant deformed Buzzes in the background. And Alice became a doll. If you play with dolls too much... AI Failure #7: "Tenshi Hinanawi eating a corndog on Route 66." The AI does not know who Tenshi is and cannot draw her, so it gave me some Japanese woman instead. And it seriously fucked by adding that giant corndog on a misaligned stick that she isn't even holding. However, the woman's design and posture give off the right impression of being a tourist enjoying a corndog. That's a pretty good result even if her hand is messed up. AI Failure #8: "Cirno freezing Pepe the Frog". The AI produced their bastard daughter instead. AI Failure #9: "Shanghai Doll from Touhou wielding a fencing rapier", "Vivian James as a Touhou character", and "Rydia from Final Fantasy IV as a Touhou character." The AI can generate good looking characters in the Touhou style even if they do not resemble the character that was requested. Rydia has the right color scheme but I don't recognize her as the character. AI Failure #10: My plan was to put a bunch of evil dolls together for a Toy Story parody. There is no way that Gab AI can produce this, so I gave up. The first mistake is my own, asking for Chatty Cathy from the Twilight Zone Episode The Living Doll. That doll's name is Talky Tina. Chatty Cathy is the doll that she was based on. The doll that the AI gave me is neither of them, but is a good generic doll character. Next I asked for her and Chucky in the same image. The AI knows Chucky, but also bled elements from Chucky into the other doll. And vice versa: the AI put Chucky in a skirt because the other doll was wearing one. That is one of the ways the AI consistently fails when drawing multiple characters or objects. It mixes their aspects together.
(373.67 KB 910x911 aifail011_vashandreimu.jpg)

(87.54 KB 454x454 aifail012_remilia.jpg)

(67.04 KB 454x454 aifail013_nazrin.jpg)

(66.19 KB 454x454 aifail014_farsidecow.jpg)

(92.65 KB 454x454 aifail015_haddock.jpg)

AI Failure #11: I asked for Trigun in the styles of Peanuts and Tintin, and Reimu in the styles of Peanuts and "Marvel comic book". The AI recognizes "Trigun" as Vash the Stampede and draws that character, but the art styles are not what I had requested. The AI does a better job style-wise with Reimu. It does not know what Reimu is supposed to be holding so it gave her a mushroom / tiny parasol thing. Other than that they both look pretty good. AI Failure #12: "Remilia Scarlet shows a road map to a confused Simon Belmont". It's interesting to see what the AI did with Remilia and Simon, but the image lacks the elements that would have made the concept funny. AI Failure #13: "Nazrin sitting on the moon eating cheese". The AI figured out that Nazrin is a mouse and made her a furry. I am impressed. The AI goofed by putting another moon on the moon and a moon in the background. Is that enough moon for you? AI Failure #14: "A cow from The Far Side in a My Little Pony cartoon". The AI made a cute cow in a 3-D artstyle. AI Failure #15: Captain Haddock from Tintin throwing Monkey Luffy from One Piece off of his boat. Except, I misspelled Monkey Luffy's name. The AI drew a decent Captain Haddock but it kept drawing Monkey "Luffuy" as a literal monkey dressed like The Man in the Yellow Hat from Curious George. And it kept drawing the boats and the characters floating in the air over the sea.
(90.69 KB 454x454 vegeta.jpg)

(79.14 KB 454x454 hellokitty.jpg)

(67.16 KB 454x454 tonbeery.jpg)

(68.40 KB 454x454 chocobo.jpg)

Some flawed but half decent results: Vegeta eating a fajita, Hello Kitty in Hell, "a tonberry from final fantasy holding a can of beer", "A chocolate in the shape of a chocobo from Final Fantasy, in a partially opened foil wrapper", and The Centennial Festival for Magical Girls.
(107.44 KB 454x454 centennial.jpg)

(75.33 KB 454x454 iscream.jpg)

>>2982 Centennial Festival was stripped on upload. Are webms not allowed? Did a copyright filter eat the song? Here's the jpg. And enjoy some I Scream.
About time to move the thread to >>>/ais/?
>>2984 Agreed. If you wish to, make a new AI thread too, but this time include both visuals and audio in the same OP, since the audio world is kind of very slow moving compared to the leaps and bounds in picture generation.
(381.37 KB 454x454 space_battleship_tomato.png)

Guess the anime
(2.42 MB 1920x1080 00035.png)

(2.21 MB 1920x1080 00037.png)

(1.61 MB 1280x960 00060-2069474834.png)

New Year, time to hit the gas pedal with diffusion.
>>2988 Prompts? Just asking for a friend haha
(83.08 KB 721x891 CHAOS.png)

>>2908 It has, and it's glorious.
(529.80 KB 576x1024 undefined_image (3).png)

(741.04 KB 576x1024 undefined_image (14).png)

(630.50 KB 576x1024 undefined_image (12).png)

(960.62 KB 768x1024 undefined_image (13).png)

(828.05 KB 768x1024 undefined_image (11).png)

Also futzing around with some Poison Laras with a few muscle and inflations ones to see what comes out of the experiments.
(2.77 MB 1360x2048 undefined_image (2).png)

(598.29 KB 576x1024 undefined_image (4).png)

(931.63 KB 768x1024 undefined_image (8).png)

(567.64 KB 768x1024 undefined_image (9).png)

(1.05 MB 768x1024 undefined_image.png)

>>2991 Thiccer, Plumper, and gassier than ever before?


Forms
Delete
Report
Quick Reply