wow, my article is already on the second gogle page for the search "how stable diffusion works"
wow, my article is already on the second gogle page for the search "how stable diffusion works"
FXD
Falmost on page 1
AThat's awesome, congratulations!
Fdmn this aspect ratio took a lot of time
Ffinally fixed the script
Faa.zip4.92MB
Fi had been struggling to figure out like 4 hours. constantly debugging haha :d dmn
ATemplate has been updated:
Updates:
Updates:
- Automatic1111 Web UI version 1.3.1.
- Dreambooth extension 1.0.14 (no longer dev branch).
- Kohya_ss added.
- Tensorboard added.
F5 instance processing class images
Fspent 2 days for this dataset :/
ABy the way, you don't need to kill the web ui process to view the logs in my new image, you can simply tail the logs at
/workspace/logs to see what is happening. You only need to kill it and restart it to resolve issues.Fcomputer crashed :/
JAwesome, I'm really looking forward to that new workflow! Thanks for replying
Fsuper slow training without xformers :/
F
Fi am going to sleep hopefully see you tomorrow
MDoes anyone know where to get this ui for Controlnet
M
MNevermind, i figured it out. needed to update CN
AWhat is the best commit to use for Kohya? the master branch seems different from the videos and can't get it to work
AI find it interesting to watch the learning rate of the Facebook D-Adaptation optimizers
M@Dr. Furkan Gözükara I tried a dreambooth training with 2 concepts
1 of a female character and 1 of aesthetic
However, I feel my aesthetic is not trained well as it does not really generate my trained female model in that aesthetic, it is a hit and miss, I am asking here as I was watching one of SE courses videos and that part of style training was skipped.
What kind of settings are appropriate for style training?
And should I train my style and character separately and then merge the checpoints?
1 of a female character and 1 of aesthetic
However, I feel my aesthetic is not trained well as it does not really generate my trained female model in that aesthetic, it is a hit and miss, I am asking here as I was watching one of SE courses videos and that part of style training was skipped.
What kind of settings are appropriate for style training?
And should I train my style and character separately and then merge the checpoints?
Fresults are amazing quality
Fmy workflow looks like worked great
Fi will train myself now :d
Fnice graph
Fbut i think go the the best model with auto extension
Fclient results are simply amazing
F@dbai64 hi in this video i explain how you can change versions. like turn back to older version of auto1111 and use it : https://www.youtube.com/watch?v=tXaQAkOgezQ
YouTubeSECourses
Discord : https://bit.ly/SECoursesDiscord. New fantastic style transfer feature via T2I-Adapter added to the #ControlNet extension. If I have been of assistance to you and you would like to show your support for my work, please consider becoming a patron on 
https://www.patreon.com/SECourses
Playlist of #StableDiffusion Tutorials, #Automatic11...
https://www.patreon.com/SECoursesPlaylist of #StableDiffusion Tutorials, #Automatic11...
AStill using Lion?
RLooks like the batch script has been done already
RGitHub
Use miniGPT-4 batch to generate captions for a lot of images! - GitHub - pipinstallyp/minigpt4-batch: Use miniGPT-4 batch to generate captions for a lot of images!
FYES
Fnice
AWhich noise scheduler are you using? DDPM or DEIS?
Fdefault. i think ddpm
AYeah thats the default, thanks
Fi thinkt his new workflow will change photo reality training
Fi am uploading files to make a video and post on patreon
RHey, which workflow you referring to?
Fnew one. i havent published yet
Fbeen working on it for several days
Fi am uploading files to make a video and post on patreon
ROk amazing. Will be looking out for it
Fi thinkt his new workflow will change photo reality training
Fi will share amazing script as well that can auto crop subject into any aspect ratio you wish
Fmakes so much easier to prepare images
Fi will share amazing script as well that can auto crop subject into any aspect ratio you wish
AHi everyone! Trying to do dreambooth training, but keep getting this error. Could anyone help me with this one?
message.txt9.89KB
