Creds to Flonixicon for the original image set for his 2.1 embed found here https://civitai.com/models/1373/sd-20-768-flonixs-side-view-v3 This is version 2 though version 1 generated pretty much the same images, version 2 is a much smaller file.
The model is fairly rigid and will only produce profile views unless forced with prompt weighting or it is mixed with other LoRa models. Even though this is a 1.5 model base, you can generate up to 1080x1080 but the bigger the dimensions, the longer the neck gets.
Settings I have been using are 3-5 CFG scale 20-40 steps and LoRa weight of 0.75 to get the best results. with no negative prompts. I prefer to use the khoya extension 'Additional Networks' as it gives me greater control.
If feedback is positive, I'll work on a version 3 which will include the more variety and hopefully better quality.
描述:
训练词语:
名称: DoriSamurai_style_V2.safetensors
大小 (KB): 9326
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
