3640 steps, one hour! You damn genius 🙂
3640 steps, one hour! You damn genius 
LThis would have been 12 hours on my old config
FAhhh im glad! All fixed now
FWith min snr gama
these best settings are either 0 or 10
so if you get bad results change one
these best settings are either 0 or 10
so if you get bad results change one
LI’ll play around. It’s easier to tweak when training isn’t 12-18 hours 
FYep keep testing and trying, some of my trainings i do 10+ with different settings hehe
Falso wait did you disable hardware-accelerated gpu scheduling ?
LI enabled gradient checkpointing, for advanced I don’t think I did anything else.
F
Fthis one
FI am preparing a very good tutorial can you check and let me know if anything else missing?
https://github.com/FurkanGozukara/Stable-Diffusion/blob/main/Tutorials/Install-Python-C-Plus-Plus-Tools-FFMPEG-CUDA-Install-Tutorial.md
https://github.com/FurkanGozukara/Stable-Diffusion/blob/main/Tutorials/Install-Python-C-Plus-Plus-Tools-FFMPEG-CUDA-Install-Tutorial.md
GitHub
Stable Diffusion, SDXL, LoRA Training, DreamBooth Training, Automatic1111 Web UI, DeepFake, Deep Fakes, TTS, Animation, Text To Video, Tutorials, Guides, Lectures, Courses, ComfyUI, Google Colab, R...
FMicrosoft Visual C++ 2015 Redistributable ?
Fc++ tools added
Fbut i can add Microsoft Visual C++ Redistributable for Visual Studio 2022
FSometimes i had issues with onnxruntime as well
Fi am gonna add it too
Fye onnxruntime is also c++ tools requiring
Fwill you also add latest cudnn as well, I know some people dont like signing up to nvidia to install them
Fi have roop but i plan GAN tutorial too
Fadded it 11.8
Fbut i can also add latest direct link
Fhave you tried facefusion?
Fnot yet but it is same as roop
FMy SDXL DreamBooth training much lesser for myself
Fit can be even made under 25 min if you have 48 GB GPU
Fyes i have lots of exp
Fdid this made any speed change for you?
FYes by at least 10 -15 mins
FGitHub
Hello, I have found this few days ago and seem no one / wiki write about this. By disabling Hardware-accelerated GPU scheduling on Windows Settings, This improve about 10-50% of SD performance, Aff...
Awhat am I doing wrong here? :\ the results are absolute trash
Lwww @FiveSAI help me get it to an hour which is bomb
Fawesome
Fwhat changes have you made
LI thought a 4090 was better than 12 hours for a Lora.
LIf you look at the chat here you’ll see but the biggest was updating Kohya, changing transformers version, enabling gradient checkpointing, and adding captions.
LWe’ll see how it comes out
Fyou could also generate sample as well, but it would add like 3-4 extra mins of training
MHi Guys,
I'm new to the world of DL,
i'm having problem with my code.
i was refering to https://github.com/justinpinkney/stable-diffusion/blob/main/notebooks/imagic.ipynb
and i don't have ckpt file,
is there some way to download this file, like npm in webdev?
or do i have to code it?
I'm new to the world of DL,
i'm having problem with my code.
i was refering to https://github.com/justinpinkney/stable-diffusion/blob/main/notebooks/imagic.ipynb
and i don't have ckpt file,
is there some way to download this file, like npm in webdev?
or do i have to code it?
LMight be worth it though. I’m an artist so part of the fun is going through the epochs and seeing what I get
Fi never used this yet.
FYep and there are a few prompts you can test
you wanna aim in the middle of realism and flexablity
you wanna aim in the middle of realism and flexablity
PHow do i learn to create my own workflows for comfy ui
LThere's some good examples and videos on youtube but my suggestion. Download one that already does somewhat of what you want and just play with it
PThanks for the response.Honestly I am not able to understand when a purticular node is required (other than base ones ) .Does the model documentation give requirements like " for sdxl ,refiners are needed " ..etc
Pfor context I have always used to others workflow to generate images and has a superficial knowledge of what the base nodes means
SHey Guys Hows it going tonight?
SHye Doc u around? I did clean windows install on my potato so if i cant bake stable diffusion now.. there will be no hope from here..
Fi am recording a fresh intall tutorial
FGitHub
Stable Diffusion, SDXL, LoRA Training, DreamBooth Training, Automatic1111 Web UI, DeepFake, Deep Fakes, TTS, Animation, Text To Video, Tutorials, Guides, Lectures, Courses, ComfyUI, Google Colab, R...
Fdo you think anything missing?
