search

Rem105-210 / DiffFashion

MIT License
69 stars 10 forks source link

About the ten single-step denoising of the style reference image during the denoising process #8

Open LanTianBy opened 6 months ago

LanTianBy commented 6 months ago

Thank you for your contribution! I have some questions about the code details and look forward to your answers! 1. Is the DDPM encoding mentioned in the article actually the process of adding noise to the image? 2. From line 825 to line 868 in gaussian_diffusion.py, the style reference image seems to have undergone 10 single-step denoising processes after noise addition, and each denoising process is reversed using the DDPM noise addition formula. What is the significance of this? We look forward to your reply. Thank you again! 1714014529286

LanTianBy commented 6 months ago

Sorry, I forgot another question. Is 636 a tag for a bag? Thank You！

Rem105-210 commented 6 months ago

Thank you for your attention 1.Yes, "encoding" refers to the noise addition process at the image level. 2.In this function, we need to start denoising from the noised style image to generate the target image. Here, the initial 10 single-step denoising steps come from the improved diffusion setting, which can reduce the randomness of the denoising process. We have retained the same settings as IDDPM（https://github.com/openai/improved-diffusion）3.Yes, 636 comes from the category labels of ImageNet, used for conditional generation in IDDPM.

LanTianBy commented 6 months ago

Thank you for your attention 1.Yes, "encoding" refers to the noise addition process at the image level. 2.In this function, we need to start denoising from the noised style image to generate the target image. Here, the initial 10 single-step denoising steps come from the improved diffusion setting, which can reduce the randomness of the denoising process. We have retained the same settings as IDDPM（https://github.com/openai/improved-diffusion）3.Yes, 636 comes from the category labels of ImageNet, used for conditional generation in IDDPM.

Thank you very much, this has been a great help to me. Wishing you all the best in your future work!