This was previously named AnimagineXL Realistic or Stylistic.

What makes this model special? I prioritized prompt following and composition, making it excellent for art. This makes it excellent at ???? as a consequence, not by over trained on pr0n at the cost of general knowledge, like many other models.

How did I do this? I took the under-appreciated Animagine XL v4.0 because it was well trained from scratch on SDXL 1.0 Base for character knowledge up to Jan 2025. The prompt knowledge of the clip + model is something really special. With the right tags, it is rather artistic and magical with lighting, shadows and composition. CagliostroLab have a discord too. I didn't use "Opt" or "Zero" in the mix, as I think they broke the magic with those.

If you like it, 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! 😂If you need inspiration, support artists who share prompts by dropping ❤️'s in the gallery below!

Check out the workflow in the "Training Data" download! The v8 model is basically tuned for ddpm at cfg 1, 20 steps. However, Euler a works almost as well, and LCM is particularly good as a first pass for architecture, strong shapes, followed by a ddpm pass to add more detail.

If you want to use cfg > 1 then add the cfg norm node in comfyui to stop burn!

For characters, the prompt format is:

1girl, name, series, rating, stuff, ... , stuff, ..., stuff, ... , masterpiece, high score, great score, absurdres

1girl, remilia scarlet, touhou, safe, riding a horse, sunset, masterpiece, high score, great score, absurdres

Prompting guide for Animagine: https://cagliostrolab.net/posts/optimizing-animagine-xl-40-in-depth-guideline-and-update

You don't need to add masterpiece, high score, great score, absurdres to the end of the prompt - it is tuned to work without it. Only add it if you want more anime style.

Version History

v1 - quite authentic to Animagine, but noisy except with specific samplers.

v2 - more photorealism but less authentic to Animagine and broke lighting and expressions.

v3 - ⭐ - I focused on improving the prompt following from the base model, face expressions and only "good enough" photo-reality. Works with wider range of samplers, steps and cfg.

v4 - ⭐ - An improvement in ???????, realism, highlights and shadows. It also follows closely to Animagine XL 4.0. Unfortunately it became more ?????, use "safe" in the positive and (????, ????????) in the negative.

v5 - ⭐⭐ - Improved prompt following, shadow and light dynamic range, better artist style flexibility, and good ??????? and acceptable hands. It's still a bit more ????? 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.

v7 - ⭐⭐ - Squeezed a little more out of Animagine by optimizing specific booru tags and darkness.

v8 - ⭐⭐⭐ - Heavily reworked the original Animagine 4.0 to maintain dynamic range in lighting and shadow, knowledge and prompt following, and vastly improved photo-realism.

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, ??????????????? 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.