This is an experimental checkpoint that combines some of the best realistic Illustrious models with bigASP 2.5. It worked.
BTW, v2.4 is now available here! Give it a try!
❤️ If you like Snakebite, you can help offset the cost of training:
Why it matters
bigASP has great prompt adherence but it is wildly inconsistent as far as style and composition. It feels like a base model with untapped potential. Dialing in the right settings is like trying to solve a Rubik's Cube.
Illustrious models, on the other hand, lose many concepts when going 3D. They're hit-or-miss even with fairly popular booru tags. But the lighting and composition of these weights are still 👌
I wanted to see if we can get the best of both worlds, and it turns out we kinda can! Careful block merging is the key. We can inject bigASP's output_blocks.0 to acquire much of its conceptual knowledge. Adding its middle_block.2 seems to reduce anatomical issues (otherwise, you'll get a lot of extra arms and fingers.)
First impressions
Best prompt adherence for a realistic SDXL model I've ever seen. And no, not just for smut. (But especially for smut)
Compatible with booru tags as well as natural language. I think a mixed prompt approach is best: I use about 70% tags (no underscores) and 30% natural language in a prompt.
Understands bigASP's style tags to an extent, including stuff like masterpiece quality and 35mm.
It *feels* like something new and worth exploring. Don't sleep on bigASP!
Drawbacks
Illustrious 2.0 models support resolutions up to 1024x1440 or 1024x1536 without horrific stretching of anatomy (e.g. longcat-type torsos), but bigASP's optimal resolution is only 832x1216... and I don't recommend going above that in Snakebite. If you do, the anatomy will be mostly okay (which is surprising), but image composition becomes very odd and unpleasant.
Since we're in a strange new latent space, your existing LoRAs won't work very well. But they're worth retraining.
Recommended Settings
The Turbo variant is better for inference. It's super fast and has slightly improved aesthetics. The non-turbo version is useful for finetuning, and it can produce nice textures if you don't mind waiting 25+ steps.
Turbo
8 or 9 steps
LCM sampler
CFG 1
Custom sigmas below, or simple
Full
20 to 28 steps
Euler ancestral sampler
CFG 3 to 4
Custom sigmas below, or simple
Custom sigma curve (you can use comfyui-kjnodes to apply):
15, 8, 4, 2, 2, 1, 0.4, 0.2, 0
If you're getting mangled limbs, you can often salvage the image by adjusting the first few values of your sigma curve. Here's one that is more stable for certain prompts:
14, 5, 2, 2, 2, 1, 0.4, 0.2, 0If you're still getting body horror, you can try the following quality tags (you'll need ComfyUI-ppm to apply negative weights to your positive prompt):
masterpiece quality, realistic photo, (worst quality,:-1) (mutated,:-1)Snakebite is very responsive to stylistic terms, especially by IL standards. Keep the extra "fluff" to a minimum - almost every token I've tried has a significant impact on the picture.
Finally, I suggest trying the CLIPAttentionMultiply node. If you boost the q and v parameters, it will effectively cause your image to become more "Illustrious-like": cleaner, more stable, but less realistic and (usually) less adherent to the prompt. Set both values to 3 for a very clean image.
BIGASP'S CLIP IS NOW IN PLAY!
In versions 1.3 and up, Snakebite includes a little of bigASP's CLIP, which means you can take advantage of more style prompts. Experiment with different terms to see what works. Personally, I keep it simple - this will usually improve your image without any side effects:
high quality, sharp focusWhich version is for me?
If you're wondering which version of the model to use, here's a TL;DR:
v1.4 = next-level realism, jaw-dropping textures, very stable, slightly less vibrant than previous versions and less capable of non-photographic images
v1.3 = good anatomy, good backgrounds, good coherence
v1.2 = best punchy colors
v1.1 = most influence from bigASP (excluding CLIP), dull colors, a failed experiment TBH
v1.0 = impressively creative but very unstable
If you like the model or use it for further finetuning, please let me know! I'd love to see the results. 💪
Description
I finally found a good way of merging some of bigASP's CLIP into the Illustrious CLIP. The result? Better prompt adherence and fewer anatomy problems, especially with longer prompts and natural language. The cost? Images are slightly less saturated, less contrast-y. Artistic or surreal stuff might be weaker as a result, but I think v1.3 is an improvement overall.
Additionally, I tweaked the acceleration stack to have less of negative effect on image quality, while still converging at 8 steps.
Check out the demo images and decide for yourself if you want to upgrade at this time. I made a few comparisons with some of the awesome pictures you guys submitted for v1.2.
FAQ
Comments (11)
Can this work with SwarmUI?
It should, as far as I know. 🙂
It works with SwarmUI.
Turbo mode is impressive. Thank you.
Interestingly, I get correct hands with 1.2 turbo, but wrong hands with 1.3 turbo. Impressive models nonetheless, thank you!
I think 1.2 is the best in my experience. BigAsp has some tendency to bring some not so aesthetically pleasing female forms to the forefront, like it was trained on bottom of the barrel porno mags primarily. 1.4 brings more of that to the surface. Competent model sure, but the females are looking like weathered truck stop workers and less like women anyone would be interested in generating.
@solss_ "bottom of the barrel porno mags" sounds so cruel and personal for some reason XD
However, maybe you are right, because it absolutely doesn't like wide images. Looks like it was trained on low-quality vertical shots only.
@dobomex761604 Lol. I love both BigAsp for what it can do, and the breadth of what it contains and is capable of, and Snakebite too of course! But just speaking aesthetically, from my tastes, the more of BigAsp that comes through, the less beautiful the women become. Both in their facial features and their figures. I don't remember off the top of my head how many million photos BigAsp was trained on, but the dataset wasn't curated for aesthetics it seems, mostly for versatility and SDXL expansion. For my use case, 1.2 has the best outputs. I retested the same seed across my favorite images with the small prompt adjustments recommended in some cases with the later models. Just a matter of taste. No offense intended to any one creator at all.
@solss_ It was trained on 13 million images. Pretty nuts. A lot of it is porn. I'm evaluating it more closely now and I can confirm that the new facial features we're seeing in 1.4 are mostly coming from bigASP. The features are being emphasized by one of the aesthetic LoRAs I merged in, possibly at too high of a weight. But it's cool to see what lies "under the surface" so to speak.
bigASP is capable of so much, but a lot of it is impossible to reach with prompting alone.
@solss_ I totally agree. Images from Bigasp are realistic but tacky—they're far from beautiful or pretty. I also hope the next version of Snakebite takes only the realism from Bigasp. If it can be realistic while following Illustrious's high prompt adherence, I think it would become the absolute best model.
Found this via your reddit post a few days ago and this really has been impressive. Irrespective of it's LoRA compatability issues this does seem like abetter base model than most 'realistic' checkpoints for SDXL or Pony/Illust