Closed trofimovaolga closed 2 days ago
Hi, could you please explain how to use text prompts during inference? During training you don't use any text prompts and condition for unet is output vector from makeup_encoder. You can only condition the model with a makeup image, not a text, correct?
Sorry, I forgot to organize this part of code ...... really busy now 😣...... you can simply achieve it by setting 0 to the number of scale for each ControlNet part and using original text prompt cross-attn.
Hi, could you please explain how to use text prompts during inference? During training you don't use any text prompts and condition for unet is output vector from makeup_encoder. You can only condition the model with a makeup image, not a text, correct?