Aitrepreneur
patreon
CLOTHES/OUTFITS LORA Training Presets Kohya
🕑 Added 2023-10-12 16:21:24 +0000 UTCHere you will find 5 optimized training presets for CLOTHES/OUTFITS LORA inside Kohya SS. Whether you're training locally or using Runpod.
What's inside?
3 Local Training Presets: Designed for those who love to train on their own systems.
- Normal Clothes Training: For your everyday needs.
- Fast Clothes Training: For those in a hurry
- LowVRAM: Optimized for local systems with limited VRAM.
2 Runpod Presets: Perfect for those using Runpod's GPU renting platform. These presets automatically locate the location of the stable diffusion XL model for you.
- Normal Clothes Training: The classic training preset.
- Fast Clothes Training: Train your models as fast as possible
Why These Presets?
For LORA training, every little optimization can make a huge difference. These presets are the culmination of countless hours of experimentation and refinement. By using them, you're ensuring that your LORA CLOTHES/OUTFITS training is both efficient and effective, regardless of where or what you're training.
And as always, supporting me on Patreon allows me to keep creating helpful resources like this for the AI art community. Thank you for your support - now go train some awesome LORAs!
👉Download Style Presets👈
password: CLOTHPRESETS
Comments
Aitrepreneur
it's normal it's an nvidia driver issue, you need to disable system memory fallback: https://nvidia.custhelp.com/app/answers/detail/a_id/5490/~/system-memory-fallback-for-stable-diffusion
Mauro Prelli
Sorry for another question: I tried reducing the image source to 512x512 and also the output to 512x512, and in about two hours I managed to generate the LORA models. However, I see that it generates 9 different ones. Which one is the correct one to use? Thanks.
Mauro Prelli
Thank you again for all information you provide! I try my first training with your clothes preset, before I've upscaled jacket with rsgan 4x (obtaining 1808 x 2384 png image) and when I try to train clothes model the work time was 33h (!) , with fast-preset it goes to 9H (!). My hardware is a 19-13900KF / 32Gb Ram DDR4 / GPU 4070Ti When i was trying to make lora model with a preset like SD15 Lokr V1.0 for example it works very very fast (like 20 minutes or so) ... I've followed every step properly and checked for two times... The only difference I noticed in my process besides the preset is the size of the source images, I use 512x512 and generate at 512x512... can it really influence that much or is there something I'm missing?
Aitrepreneur
ok so sd 1.5 is a model like sdxl like it's just a file, it's not a webui version. If you want to train a 1.5 model instead of a sdxl model you can use the same presets just select the 1.5 instead of sdxl, uncheck the sdxl checkbox and change the training resolution from 1024,1024 to 512,512
Mauro Prelli
Hallo! I've seen you are using SDXL, how can I use that on SD 1.5? Is it possible to load a checkpoint sdxl_like to use sdxl inside automatic1111 sd_15? Thanks
Aitrepreneur
make sure you have enough images of the clothing item you want to train, often this happens when you have a limited number of training images
Diggy Dre
For some reason, I keep getting blurry backgrounds. I did manage to get the clothing item that I wanted, but no matter what I do, the backgrounds are not great at all.
Aitrepreneur
yes, you can consider it as a hat
ZenzeosArt
Would hairstyles work the same way?
Aitrepreneur
the password is to open and extract the archive using winrar
