search

HighCWu / control-lora-v2

ControlLoRA Version 2: A Lightweight Neural Network To Control Stable Diffusion Spatial Information Version 2
Apache License 2.0
98 stars 6 forks source link

Question about conditioning #5

Closed F9393 closed 1 month ago

F9393 commented 1 month ago

Thank you for great work and repository. Could you please clarify what does "use_same_level_conditioning_latent" flag do?

Thank you,

HighCWu commented 1 month ago

For example, ControlNet uses the 3-channel 512x512 image as the input to be downsampled 8 times to the same size as the 32x32 latent. When using the use_conditioning_latent option, we will encode 512x512 image with vae model to 32x32 latent and then upsample to 512x512 as the control input condition. use_same_level_conditioning_latent directly uses 32X32 latent as the control condition. There is no additional downsample steps.

F9393 commented 1 month ago

Got it, thank you for prompt response.