search

yownas / shift-attention

In stable diffusion, generate a sequence of images shifting attention in the prompt.
Other
164 stars 17 forks source link

Sequence Input - Change initial image with each frame #16

Closed mlsnr closed 1 year ago

mlsnr commented 1 year ago

I'd like to use this effect on an image sequence from a video. Would it be possible to adapt the script, so each generated frame uses a different frame from the sequence as initial image?

It can be done manually, but it would require a full generation cycle for each frame, then handpicking the relevant frames and dumping many unnecessary images.

yownas commented 1 year ago

I can definitely see why this would be a nice effect... but I'm not sure if I know how to do it. I've never been a huge user of img2img, so the script is pretty much focused on txt2img generation. The reason it works in img2img is mostly a happy accident.

So at the moment I don't have a good idea how to solve this without breaking a lot of other things. But I'll give it a try this weekend if I manage to set aside some time for it.

yownas commented 1 year ago

https://github.com/yownas/sd-webui-video-sa

I got curious what you could do with it. :) Very experimental at the moment. It lacks a lot of functionality but should "work".

mlsnr commented 1 year ago

Awesome! Thanks so much :)

On Thu 22. Jun 2023 at 23:45, yownas @.***> wrote:

https://github.com/yownas/sd-webui-video-sa

I got curious what you could do with it. :) Very experimental at the moment. It lacks a lot of functionality but should "work".

— Reply to this email directly, view it on GitHub https://github.com/yownas/shift-attention/issues/16#issuecomment-1603354980, or unsubscribe https://github.com/notifications/unsubscribe-auth/BABLP6OWCJKNPPQJE5Z7JQTXMS4GRANCNFSM6AAAAAAZPCBV7Y . You are receiving this because you authored the thread.Message ID: @.***>