search

CompVis / stable-diffusion

A latent text-to-image diffusion model
https://ommer-lab.com/research/latent-diffusion-models/
Other
68.61k stars 10.19k forks source link

Training/Fine-tuning img2img #303

Open lukasfolle opened 2 years ago

lukasfolle commented 2 years ago

Thank you for open-sourcing this awesome project!

I would be interested in training a stable diffusion model conditioned on an input sketch and a text string just like the script under scripts/img2img.py allows, but on my own dataset. Do you provide the training/fine-tuning code for this and I just overlooked it?

:wave: Lukas

BenjaminIrwin commented 1 year ago

Hey @lukasfolle, did you have any luck with this? Working on something similar and keen to discuss.

peteromallet commented 1 year ago

Hi Luke & Ben, just you make any progress here? Have been considering something similar for a project I'm working on

YJ-142150 commented 1 year ago

Hello, I am also interested in img2img fine-tune. I'm also considering on similar subjects. Did you get any ideas about this?

pfdamasceno commented 1 year ago

This worked for me: https://www.reddit.com/r/StableDiffusion/comments/wv2msc/stable_diffusion_img2img_google_collab_setup_guide/

YJ-142150 commented 1 year ago

I looked at it, but it seems that the code is not for fine-tuning.

RicoJYang commented 1 year ago

I looked at it, but it seems that the code is not for fine-tuning.

I have the same question about fintune/train img2img. Maybe img2img is untrainable?

AlexanderKozhevin commented 12 months ago

I'm curious too

MrJ0bs commented 11 months ago

,ccy a

On Sat, Dec 9, 2023, 11:11 PM Alexander Kozhevin @.***> wrote:

I'm curious too

— Reply to this email directly, view it on GitHub https://github.com/CompVis/stable-diffusion/issues/303#issuecomment-1848592658, or unsubscribe https://github.com/notifications/unsubscribe-auth/AU2KAD57HBWU5X4BJDS664LYISPEHAVCNFSM6AAAAAAQQ2644WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNBYGU4TENRVHA . You are receiving this because you are subscribed to this thread.Message ID: @.***>