Improve Stable Diffusion Quality with Skip Connection Rescaling (No Training)
1–5 of 5 posts
Re: Improve Stable Diffusion Quality with Skip Connection Rescaling (No Training)
#2Improve SD image quality and reduce artefacts without any additional training, simply by reweighting skip connections in the decoder stage of a diffusion Unet decoder
Re: Improve Stable Diffusion Quality with Skip Connection Rescaling (No Training)
#3Improve SD image quality and reduce artefacts without any additional training, simply by reweighting skip connections in the decoder stage of a diffusion Unet decoder
i know i should read the paper but... what does reweighting the skip connections mean in this case? like... how?
Re: Improve Stable Diffusion Quality with Skip Connection Rescaling (No Training)
#4Improve SD image quality and reduce artefacts without any additional training, simply by reweighting skip connections in the decoder stage of a diffusion Unet decoder
i know i should read the paper but... what does reweighting the skip connections mean in this case? like... how?
In the decoder, the features from the unet blocks get concatenated with features from the encoder layer through 'skip connections'. The paper discusses how rescaling the backbone features (element-wise multiplication by some scalar) before concatenation improves image quality.
Re: Improve Stable Diffusion Quality with Skip Connection Rescaling (No Training)
#5Earlier quoted context omitted.
i know i should read the paper but... what does reweighting the skip connections mean in this case? like... how?
In the decoder, the features from the unet blocks get concatenated with features from the encoder layer through 'skip connections'. The paper discusses how rescaling the backbone features (element-wise multiplication by some scalar) before concatenation improves image quality.
ty