First off thank you. Second off, I tried, but they just didn't allow it to pass. Definitely will be messing around her more often in the future.

>>61675 No prob, sorry for all the fighting you had to deal with in the other thread. I personally don't care for furry vore stuff myself but I also am not full of childish vitriol just because I'm not a fan. Besides, AI vore video generation is gonna get better and better and, whether I'm a fan of furry stuff or not, you and others will help push it forward and I wanna see how it goes. Hope things here in this thread are gonna be a little less toxic, if nothing else.

To be clear the intent of this thread is not really strictly furry stuff. It’s animal pred/? and human pred/animal prey, because people were whining about animal prey being used to test prompts for swallowing. Hopefully it will get used. I want to try contributing at some point.

>>61695 I figured as much but, as the guy who made it, it's also for the furries too. It's just to get the bunch of whiny pissy dickheads to stop shitting their pants in the other thread so actual content can be posted without the strife. And go right ahead, contribute away my good man!

Gotta question. I know that SD/SDXL/Pony has different training terms... but what would be a good way to specify unique markings? Is there a good repeatable workflow to do tattoos like Krystal has? Or unique patterns like Renamon has, which are not typical of "fox" or "Vixen" tags? I know that LoRA can be made with enough art, but I'm talking about creating original characters with similar levels of complexity and getting to a point where I could make my own LoRA. Any tutorials about accomplishing an OC from scratch using inpainting or image2image would be great. Also, somewhat related, is there a way to prompt for color shades beyond the most overtly famous ones? (RGBY, CMYK, O/P, black and white.) Getting neon magenta, coral pink, mint green, it all seems like a royal bitch and I have yet to find a consistent way to call these.

>>61695 As the fox guy, I don't mind the animal preds or prey. Anthro or feral, I'm open to both.

>>61712 I don't think it's possible to get specific/consistent designs like that. The only possible way for that to happen is by uploading an image of the character and hoping for the best. I reference the recent Coca-Cola commercial a lot, since it's a hot topic, But I'm almost positive, all the Coco-Cola text present in the video, was either highly trained by them having their own custom Coca-Cola creating AI, OR, they had to superimpose the logos onto all the bottles and trucks, post production. Many of these Video AI are still in beta. So in order to get (sort of) consistent results is to do about 90% of the work yourself. For example: Hailuo recently held a Halloween contest, and the users who placed really high, showed how they got their visuals, I'm not sure if subscribing to Hailuo gives you access to "Video to Video" tools, But they showed how they input a barebones CG animation, showed their prompt, and Hailuo basically superimposed a background and turned all the objects into ghosts. TL;DR The secret is just doing 90% of the work on your own.

>>61721 I actually think it is something of a chormakey thing. I remember seeing some demo on youtube about how to insert text into text to image prompts. A target shade gets created, it is treated like a controlnet+segment anything combo, and can be replaced and tracked in real time. Otherwise yeah, I'm thinking I'm stuck doing a custom LoRA after a lot of hand touch ups. I just need to figure out how to get the most work done without hand intervention first to build that LoRA. Have you ever found a tag for asymmetrical animal ears? Wanted to make a yin-yang themed bunny character, but I can only get two ears of the same color.

Oh also, supposedly? Flux is really good at adding text. No personal experience myself.

>>61755 oh man yeah, asymmetrical things are hard to generate. I've found the only way to get the AI to use patterns is to input your own images. Unfortunately, as I've mentioned before, The AI still has trouble with object permanence, and that especially is so for patterns. The AI especially freaks out with parallel lines. TL;DR There is no tag. You have to use IMG to Video. But the AI is not advanced to keep it consistent, but you can get lucky.

I was obsessed with the highschool looking theme those other generations had, and I tried to make something like. I don't know how they got such a distinct look. My raccons just look like Rocket. lol Also It's a bit of a shame my best ones tonight still have errors in them. It keeps making the belly have no fur. It even have one where it removes the fur. lol The best one that kept the fur got a little silly at the end.

Cute golden retriever fellow, seems a little distracted during his lesson~

It's a pred class, and one group is starting small with some mice~ So many visual errors, but I loved that it got this close too much not to share~

SO this begins with the last video but I cut it, By cutting at a part I like, I take the last frame, I noticed how Hailuo is bad at following directions, (might be on purpose, I'm poor and stil using free) SO, I went to Vidu in desperation. Vidu's quality sucks only because their free trial is set to the lowest possible resolution. AI needs 1080 to look at least as good as Hailuo. Vidu just started a trial for three generations at 720, so the qualitty is better but still not good. ANYWAY, Because Vidu is excellent at following directions, I made some extensions to the scene, as a proof of concept, on how to make a short video longer, with almost no perceivable cuts. But obviously you can see the cut here because it goes from Hailuo 1080, to Vidu's 720. TL;DR I stitched a few scenes together using different AIs to extend the scene.

Vidu's multi entity consistency is pretty neat, but rather hard to work with. It's also super finnicky trying to describe vore internals without it kicking back because it doesn't like the prompt. Anyway. Have some Disney lesbians floating in an animated gut~

I would probably post more here, but I have a thing where I actually want to draw the things I create with this, into like, real art/animations, and If I posted the things here, it would totally expose my anonymity. LOL

>>62528 Also, I tried paying Hailuo for 10 bucks. I like the quality, but it's not worth subscribing only for front of the line privilages. They don't have the crazy features that Vidu has, despite the low quality output... I don't get money often, so like, it will be a long time before I start trying to ever pay money again for one of these. Like As I said. I'm a real artist already. I can use my hands to make real art instead, where I get exactly what I want 100% of the time, as opposed to asking the computer 9001 times to get it wrong 90% of the time.

This fox's face looks a bit funny lookin, but like I didn't even ask for them to have a big belly like that... it's not in the prompt... The AI just made this fellow look like they ate up whoever was already inside the stomach, and I really like that. =w=