additional model info if possible
Hello Trauter,
I would like to know if you would mind helping me with 2 things:
1/ understand what I have to modify to have the same thing as you with these 2 examples. I did not manage to reproduce them almost identically like the others who have the "AbyssOrangeMix" model.
Raiden doesn't have her character and color attributes + Eula I can't get the white sleeves + Lisa doesn't have some her character attributes
Process:
I made a "PNG info", to then send it in "txt2img", in order to keep your settings. The only things I edited myself are:
- Weight 1 (for Network module 1: LoRa)
- the VAE with the tirns and sometimes AnyV3 to test.
2/ What models do you use for these (more 2D) examples. I try to reproduce them to see if everything works well like you and understand how I learn.
(sorry for my english, I'm french)
Thank you very much for all this work (I sent you kofi (even if it's not a lot)
Ayu
Make sure to use this please, your first image doesn't have it.
You also used the wrong model for the Eula example, with wrong settings. I'm not going to go over your every image.
Most of the examples you wanted models of are with BloodOrangeMix_Half
And the Eula one that was done with my KOTE model
Also, there are reports of different GPU's producing different results that I also experienced when switching gpu's so your results may not look as mine's do.
https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/2418
Oh shit ! Indeed I made mistakes ... hank you very much it's really nice to have taken the time to answer anyway with a screenshot.
Thanks for the BLOOD model reference.
Oh thanks too for the info for the GPUs.
What graphics card do you have if you don't mind?
Do you train your LoRa on your own machine?
I would like to learn how to make a LoRa for an artistic style, but with all the advice on the internet, to start from the beginning already, I don't know what to buy as a card, I prefer to ask someone who does (and very Well)
If you don't mind answering.
Thanks
3090 rtx. own machine but I trained on cloud too. I recommend just using cloud for training instead of buying gpu. If you must, buy at least 8gb VRAM, 12 if you want to do everything ( not 100% sure if you can fine tune with 12gb VRAM ). Just ask around discords for guides, most people will help you. ( Also thank you for the coffee )
Thank you very much :)