This is a model of the SD2 series, so it is not compatible with the Lora/embedding for SD1.

V1.0 a merged model of Replicant V3 and WaifuDiffusion 1.5 Beta 2 Frosty Aesthetic.

V2.0 is a merged model that uses hakoMayD instead of Replicant V3, with a slightly modified recipe.

This model does not include NovelAI leak.

I recommend using the following two negative embeddings:

• Mayng(https://huggingface.co/852wa/hakoMay/blob/main/Mayng.safetensors)

• ParaNegative(https://huggingface.co/p1atdev/Realgar-v1/blob/main/ParaNegative.safetensors)

License

Scream-SemiRealistic is released under the Fair AI Public License 1.0-SD (https://freedevproject.org/faipl-1.0-sd/), the same as WD 1.5. If you create a derivative of this model, please share your changes accordingly. Thanks to ronsor/undeleted (https://undeleted.ronsor.com/) for their help with the license.

【08/14/2023 Update】

V1.0 is based on Replicant V3.0, so in addition to the above, the following are prohibited:

Run on services that generate images for money.

Sell this model or merges using this model.

Have different permissions when sharing merges.

V2.0 is based on hakoMayD, so there are no restrictions other than the Fair AI Public License 1.0-SD.