I included a new baked VAE I did myself and the one from the creator (beta). Try it out! Recommended use in Stable Diffusion. Recommended 30 steps and CFG 11. Try different samplers if you want but my favorite for this model is DPM++ 2M SDE Karras and sometimes DDMI, Euler a and DPM++ 2M Karras. You can try to lower steps for less quality and then fix it later, but CFG is important. Lowering it can mess up the generated image.

Model made by teasan. It is one of the best semi-realistic anime style models I've seen for abstract art and scenery, but it is limited... It seems like it needs very specific negative prompts. I'm uploading v2 as it is the last version.

Keep in mind if you use v2 wihout VAE the image will look less vibrant and with little contrast. Don't use low CFG. 11 is the recommended, and 30 steps from Euler or any DPM++ M or SDE version (DPM++ 2M SDE Karras might be the best with higher steps). It is not cheap, so that's why it is recommended for Stable Diffusion.

Test freely with different samplers, but the recommended by the creator is DPM++ 2M Karras. As negative prompts, you can leave by the default ones by the creator. I don't know the difference between v2 and v1.21 yet.

Recommended prompt tags by creator: absurdres, highres, Recommended by me: (best quality), (masterpiece), (ultra-detailed:1.2),(photorealistic:1.1), ultra high res, extremely detailed

I think this is one of those models you don't need to add too much quality prompt to work.

You can also try something like: (low_contrast), (low_saturation), in the negative prompt. I think it helps a little in PixAI. New: I uploaded VAE baked, just use that version.

Recommendation from me: Use the model outside PixAI with VAE and custom embeddings. Also I recommend to refine the result with another model.This model doesn't seem to create perfects hands most of the time, so you may need a LoRA for that... and maybe you can play with embeddings.

If you use it outside PixAI, you can also try simple negative prompts (you have to download the following embeddings): EasyNegative, ng_deepnegative_v1_75t, (negative_hand-neg:1.2). For the embedding for hands you can try to use another. Use a generic VAE. This model also usually needs restore faces option activated, which is not suited up for PixAI. I'm not sure what you would get if you use custom negative prompts, experiment at your risk.

I also recommend using LoRAs with 0.5 height to polish details if you use lower steps. Stable diffusion users: Recommended clip skip 2.

Recommended resolution for PixAI: x1024 or x768 with high res fix. This is a model that looks made up for higher resolutions.