finesse

finesse is a merge of models in the Illustrious lineage and aims to bring out the fine detail and adherence to artistic accuracy it provides.

The Illustrious lineage brings a refreshing anchoring to the art medium of anime when compared to model lineages like Pony. I aim with this merge to bring that out while maintaining other good properties of the Illustrious lineage like good prompt following.

Illustrious also has very good artist knowledge (artists that would appear in a Danbooru dataset).

The original Illustrious model card states:

• We do not recommend overusing critical composition tags such as 'close-up', 'upside-down', or 'cowboy shot', as they can be conflicting and lead to confusion, affecting model results.

• Recommended sampling method: Euler a, Sampling Steps: 20–28, CFG: 5–7.5 (may vary based on use case).
  We suggest using suitable composition tags like "upper body," "cowboy shot," "portrait," or "full body" depending on your use case.

• The model supports quality tags such as: "worst quality," "bad quality," "average quality," "good quality," "best quality," and "masterpiece (quality)."

• Note: The model does not have any default style. This is intended behavior for the base model.

I however recommend DPM++ SDE Karras, but Euler a will work great too.

I've found the model to be coherent for large resolution generations, up to at least 1024x1496

The original Illustrious model card provides the following prompt example:

Prompt:
1girl, extremely dark, black theme, silhouette, rim lighting, black, looking at viewer, low contrast, masterpiece

Negative Prompt:
worst quality, comic, multiple views, bad quality, low quality, lowres, displeasing, very displeasing, bad anatomy, bad hands, scan artifacts, monochrome, greyscale, twitter username, jpeg artifacts, 2koma, 4koma, guro, extra digits, fewer digits, jaggy lines, unclear