I checked out remini ai and it does really seem to work but are there other alternatives that can b
I checked out remini ai and it does really seem to work but are there other alternatives that can be run locally?
JI think remini is similar to how the "restore face" option works in sd
DYea good point..... I have not spent enough time playing around with that. My early results were not good so I gave up but need to look int o it.
MHey guys, does anyone here understand about Dreambooth? Using the one from google's colab works great, but using the one from Automatic1111 (extension) doesn't seem to work well. It generates very bad samples
JAwh I see, yeah play around with it 
DI am still a newbie because it took me like 2 weeks just get Dreambooth to work in the first place. 
Fwhich cuda you using?
Fi got 3 videos that can fix
Ffirst downgrade cuda
FYouTubeSoftware Engineering Courses - SE Courses
Our Discord : https://discord.gg/HbqgGaZVmr. This video I am showing how to downgrade CUDA version for proper training and I am showing how to do LoRA training with 8GB GPU.
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
This CUDA ...
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/SECoursesThis CUDA ...
Flong dreambooth https://youtu.be/Bdl-jWR3Ukc
YouTubeSoftware Engineering Courses - SE Courses
I am explaining from scratch to very advanced level how to use #Automatic1111 Web UI and D8ahazard #DreamBooth extension to teach new subjects, e.g. your face into a model. Moreover, I am showing how to inject your taught face into a completely new model e.g. Protogen x3.4 to produce awesome quality images without wasting too much time on findin...
Fshort dreambooth https://youtu.be/KwxNcGhHuLY
YouTubeSoftware Engineering Courses - SE Courses
Our Discord : https://discord.gg/HbqgGaZVmr. Newest update of DreamBooth extension of Automatic1111 brought huge quality and success improvement.
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
In this video, I have explained how t...
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/SECoursesIn this video, I have explained how t...
MYouTubeSoftware Engineering Courses - SE Courses
I am explaining from scratch to very advanced level how to use #Automatic1111 Web UI and D8ahazard #DreamBooth extension to teach new subjects, e.g. your face into a model. Moreover, I am showing how to inject your taught face into a completely new model e.g. Protogen x3.4 to produce awesome quality images without wasting too much time on findin...
Fyou should see in bottom of your automatic1111 the versions
Fgo bottom of the web ui page
MExcuse my ignorance, i own a 12gb vram gpu, shoud I still follow this video?
JHaha I see! I'm new too 
Img2img, inpainting, restore faces and prompts like "realistic" "raw" are great things to use
Img2img, inpainting, restore faces and prompts like "realistic" "raw" are great things to use
Fi also have 12 gb rtx 3060
MNice! Thank u so much
MI'll watch it now
Fyou are welcome
HIs this true ?
乂I tried Lora training yesterday and garbage results on sd 2.1
HInteresting. Was it locally?
乂yes, locally
乂today I installed kohya_ss for further training with Lora + Dreambooth and it ran in to problem
Fyou need full precision for 2.1
F--no-half
M@Dr. Furkan Gözükara can It be cache? Or something ele?
MSorry for the spam
MIf I run: import torch
print(torch.version)
It says my version is 1.13.1+cu116
print(torch.version)
It says my version is 1.13.1+cu116
乂that means, to use lora we need to run it with argss --no-half
Fyou are not in the correct folder
Fuse cmd as i have shown in video please
Ffor sd 2.1 yes
乂understood
Falso using fp16 could be causing problem but i dont have enugh memory to test
Fp16 actually half precision
MOh, my fault. Thank you! It worked
MNow I'll train a model in Dreambooth and I give u a feedback
FYou are welcome
MThanks very much for your attention
HOk. I need a style model for a project this week. I will try a dreambooth for this time.
Fyes please let us know
Fi am also planning experiments on runpod
F24 gb
MNow I'm getting some real results and not a bunch of random shapes and colors lol
Fyep that fixes. people skipping that part and not working :d
Fleave a comment to the video really helpful
MOf course