Yes I made that. But all of them are pretty bad. So I was wondering if it is because of my settings
Yes I made that. But all of them are pretty bad. So I was wondering if it is because of my settings in the Colab.
Fye these settings also working great
PFor sure
F256 network size works even better
Fare you interested in learning fundementals?
TThe Kohya SS is running currently under my RTX2080. Im looking the video on my TV 
But besides I wanted to start a training in the colab. So I can compare the LoRA and things I made with "standard" Dreambooth..
But besides I wanted to start a training in the colab. So I can compare the LoRA and things I made with "standard" Dreambooth..
TYes sure - I watched nearly all your last Videos. The
Fok with colab
Fhow much gdrive you have?
Tthing with the epochs and steps and number of images in both Videos is confusing me hahahaha
Tcurrently only the free space - I need to get a credit card first to use Colab Pro
Fye it is hard to understand
Fi explain many fundementals in this video actually
FYouTubeSECourses
Our Discord : https://discord.gg/HbqgGaZVmr. The most advanced tutorial of Stable Diffusion Dreambooth Training. 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 Stable Diffusion Tutorials, Automatic1111 and Google Colab ...
https://www.patreon.com/SECoursesPlaylist of Stable Diffusion Tutorials, Automatic1111 and Google Colab ...
Falso this video will help you greatly for x/y/z plot check point comparison logic
FYouTubeSECourses
Sign up RunPod: https://bit.ly/RunPodIO. Our Discord : https://discord.gg/HbqgGaZVmr. New best training settings for DreamBooth training in Automatic1111 Web UI. 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 #StableDif...
https://www.patreon.com/SECoursesPlaylist of #StableDif...
TYesterday I made a training on the Colab, based on your Video with a friend of mine. But all of the results are really really uncanny
Fthe training dataset and number of epochs makes huge difference
TYour videos and results are great. The realistic portraits from Kohya you made are excellent.
Fi will show hopefully a way to
Fsort them with similarity
Fso you can generate thousands of images
Fand let AI the sort best ones for you
Fworking on it
Ii have ~20 images, so i have repeats on 30 for those and 1 for the regularization images.
Fthat means 1 epoch is actually 30 epochs of sd dreambooth extension
Fbecause it trains 30 times each image in 1 epoch of kohya lora script
F1 epoch simply means every training data is trained 1 time. the step count for this also depends on your batch size
Flets say i have 20 images
F1 epoch is 20 steps
Fif i make batch size 4
Fthen in 5 steps = 1 epoch
Fbecause each step it trains 4 images
A@Furkan Gözükara SECourses have you tried the Tile resample control net for upscaling images?
Fnot yet. i was waiting controlnet 1.1 to become stable
Ffor a new video
AOh okay. Apparently it works like magic by adding a lot of details. I am not able to figure out how to properly use it. Will wait for your video
Fsince automatic1111 published a new release i think time has come
BAre there any best-practices on the settings/math for training Kohya_SS styles? ie) 100-300 images.
Fwell best way is saving frequent epoch and afterwards doing x/y/z comparison
Fi have to go the gym. will be afk for a while
Ris there a batch processing tool in kohya_ss to process multiple json config settings consecutively, to test variations of LoRA's? @Furkan Gözükara SECourses
RI thought there was one
PIs the new automatic111 stable and is dreambooth fully functional? What is the best method for upgrading it if so?
PI am hoping the upgrade will fix the LDSR issue I have.
Ithanks, makes sense now
RGetting 4it/s on Kohya_ss torch 1.13... isn't 2.0 faster? You said it wasn't working so well, is updating to 2.0 not recommended? @Furkan Gözükara SECourses thx
RUpdating should in principle speed up the training 4x no (got 16it/s on webui generating but not sure how it carries over into kohya training)? Can you qualify what you meant?
ROh I see. Slow thinking. You are using captions to add specific flexible details in the portraits. Sweet. Amazing how varied methods can be. Nothing is written on stone
Fi saw they fixed LDSR issue