Latent Diffusion for Image Generation with a Unet: PyTorch Deep Tutorial

[@Hjokbbn]

Hey Luke very interesting and informative video. I have another question How can you use the model now to reconstruct missing parts of a picture

[@LukeDitria]

Hi, good question. Image-to-image and image inpainting require you to mix the target image with the noise during the generation process. I'll include some examples in the diffusion Transformer video.

[@Hjokbbn]

@@LukeDitria All right, thank you very much 👍

[@MoMaz-t2q]

Hello Luke, thanks a lot for the great work. Any plan to add videos for explaining RGB video classification techniques using pytorch?

[@LukeDitria]

Thanks for your support! Video Classification sounds like a good idea!

[@XernPL]

Hey, do you adjust the learning rate or any other parameters during training of your diffusion model? Nice tutorial series!

[@LukeDitria]

Hi, I don't in this example, but it is common to drop the Learning rate during training diffusion models.

[@jensbauman7526]

Very good video, thank you 👍

[@LukeDitria]

Thanks!

[@BELLAFaiza-p5z]

how add condition to this model

[@LukeDitria]

The simplest way is to just add your conditioning vector to the timestep conditioning vector.