v1.2: Stable Diffusion model patiently trained to understand the concept of "asset", specifically the one used for clothing in VRoid. Through IMG2IMG it can understand and modify the vast majority of clothing assets (UV map based). The imagination is the limit. The potential uses, needless to say, are enormous. The model works with "vrass" token. The more description, the better. It is very stiff, so weights, denoising, and CFG should be used liberally. It resists, but it works... and boy does it. Go wild!
v1.1: Model based on AnythingV3, trained with selected (Free) VRoid clothing assets. Any standard VRoid image asset (or others) can be used in IMG2IMG. This model recognizes the pattern and will replace the textures instead of going crazy. Supports high denoising strength and wide versatility (0.6-0.8 recommended). Basically, a free Img2Ass. Just use "vrass" token, and go crazy!
v1.0: Model based on AnythingV3, trained with many clothing assets. Any standard VRoid image asset (or others) can be used in IMG2IMG. This model recognizes the pattern and will replace the textures instead of going crazy. Supports high denoising strength and wide versatility (0.6-0.8 recommended). Basically, a free Img2Ass. Just use "vrass" token, and go crazy!
v1.2: Stable Diffusion model patiently trained to understand the concept of "asset", specifically the one used for clothing in VRoid. Through IMG2IMG it can understand and modify the vast majority of clothing assets (UV map based). The imagination is the limit. The potential uses, needless to say, are enormous. The model works with "vrass" token. The more description, the better. It is very stiff, so weights, denoising, and CFG should be used liberally. It resists, but it works... and boy does it. Go wild!
v1.1: Model based on AnythingV3, trained with selected (Free) VRoid clothing assets. Any standard VRoid image asset (or others) can be used in IMG2IMG. This model recognizes the pattern and will replace the textures instead of going crazy. Supports high denoising strength and wide versatility (0.6-0.8 recommended). Basically, a free Img2Ass. Just use "vrass" token, and go crazy!
v1.0: Model based on AnythingV3, trained with many clothing assets. Any standard VRoid image asset (or others) can be used in IMG2IMG. This model recognizes the pattern and will replace the textures instead of going crazy. Supports high denoising strength and wide versatility (0.6-0.8 recommended). Basically, a free Img2Ass. Just use "vrass" token, and go crazy!