Based on the amazing Animagine XL v4.0 which was trained from scratch on SDXL 1.0 Base for character knowledge up to Jan 2025, and style flexibility, without extra loras. It is rather artistic and magical with lighting, shadows and composition. CagliostroLab have a discord too. I haven't used "Opt" or "Zero" in the mix based on my personal style preference.
If you like v7, please hit the big grey 👍 button so it turns green, write a nice word or two in the review, and post images! It really motivates me! To boost the model even more, click and like each version individually, v1 v2 v3 v4 v5 v6 v7! 😂
If you need inspiration, support artists who share prompts by dropping ❤️'s in the gallery below!
Nemo474 has kindly sponsored the model for six weeks! It seems to take 6000-10000 ⚡ per week to stay on the online generator, please bid a little if you can help us reach 200th place and stay available!
sampler: euler scheduler: exponential for realism, beta for anime. steps: 25 cfg: 3
negative: worst quality, lowres, explicit, nsfw, ugly, blurry
realism trigger words: sunlight, shadow, realistic, photorealistic, and photographic concepts like depth of field or photo background. Don't prompt for a camera or painting unless you want an actual camera or framed painting. Don't use pony scores, score_8 etc. Don't prompt big/perfect eyes unless you want them to be enormous. It will follow your prompt!
In ComfyUI, do not increase strength too much, as even (photorealistic:1.3) will destroy the quality. A1111 / civitai normalizes strength.
ComfyUI Workflow
Download the ComfyUI workflow zip to the right under "training data". To aid artistic creation, I decided to override the concept of random seeds. Instead, it generates 36 images of random noise, with low and high saturation, then uses img2img with high denoising to create 30 rough preview images in 12 seconds, then you pick one to guide the composition, and use a high denoise on that in 20 seconds, and get a good quality image. This workflow gives variety and control every time.
For better characters, Animagine recommends not starting the prompt with "masterpiece, high score, great score, absurdres realistic, photorealistic" and instead putting it at the end:
original prompting guide: https://cagliostrolab.net/posts/optimizing-animagine-xl-40-in-depth-guideline-and-update
Version History
v1 - very authentic to Animagine, but quite noisy except with specific samplers.
v2 - higher photorealism but drifted further from the Anim4gine base model.
v3 - I focused on improving the prompt following from the base model, face expressions and good enough photo-reality. Works with wider range of samplers, steps and cfg.
v4 - An improvement in anatomy, realism, highlights and shadows. It also follows closely to Animagine XL 4.0. Unfortunately more horny, use "safe" in the positive and (nsfw, explicit) in the negative.
v5 - Improved prompt following, shadow and light dynamic range, better artist style flexibility, and good anatomy and acceptable hands. It's still a bit more horny than I'd like but it's difficult when all the realism models excel at XXX.
v6 - Quite a leap in prompt following, realism, and surrealist art. Magical.
v7 - Squeezed a little more out of Animagine by optimizing specific booru tags and darkness. The final release.
License
This model adopts the original CreativeML Open RAIL++-M License from Stability AI without any modifications or additional restrictions. The license terms remain exactly as specified in the original SDXL license, which includes:
✅ Permitted: Commercial use, modifications, distributions, private use
❌ Prohibited: Illegal activities, harmful content generation, discrimination, exploitation
⚠️ Requirements: Include license copy, state changes, preserve notices
📝 Warranty: Provided "AS IS" without warranties
Please refer to the original SDXL license for the complete and authoritative terms and conditions.