My new hardware arrived yesterday.
Added 2022-09-23 18:13:35 +0000 UTCGot a new SSD, a 3080 and two memory ram of 16gb.
The computer just had an upgrade from 4vram to 12vram and 8gb ram to 32gb.
Did not had the time to finish installation of all my applications on the new SSD, but I can already tell it will help a lot developing my apps.
There is the speed increase in testing the software and options that I could not even test it, like 32float and batch rendering.
So the next update will be better tested, hopeful it will have less bugs =p
Comments
Above Under 12gb now :)
2022-10-03 16:21:31 +0000 UTChttps://github.com/ShivamShrirao/diffusers/tree/main/examples/dreambooth
2022-10-03 16:20:15 +0000 UTCoh yes please GRisk make it happen one day
2022-09-28 11:32:25 +0000 UTCApparently it has been modified to work with lower VRAM. In any case, it may only be a matter of time before it can be used with even lower VRAM. https://www.reddit.com/r/StableDiffusion/comments/xphaiw/dreambooth_stable_diffusion_training_in_just_125/ "Dreambooth Stable Diffusion training in just 12.5 GB VRAM, using the 8bit adam optimizer from bitsandbytes along with xformers while being 2 times faster." Colab: https://colab.research.google.com/github/ShivamShrirao/diffusers/blob/main/examples/dreambooth/DreamBooth_Stable_Diffusion.ipynb Code: https://github.com/ShivamShrirao/diffusers/blob/main/examples/dreambooth/
2022-09-27 22:30:22 +0000 UTC.ckpt files will not work, since they need to be compiled with exactly the correct code. You need to find models trained with the Diffusers base and use my tutorial to add the models.
GRisk
2022-09-26 19:51:24 +0000 UTCNot for now, multiple GPU usage is a little more complex. For now you will be able to open 2 GUI and use a card for each.
GRisk
2022-09-26 19:48:55 +0000 UTCThis really take 24vram? I can't even run this, even if I want to develop a GUI for it 🤣 Will save the link and take a look soon, but it need to be able to run with 12vram so I can give it a shot.
GRisk
2022-09-26 19:45:22 +0000 UTCGoogle's DreamBooth has been modified to run in a 24GB VRAM environment. Since DreamBooth's accuracy is so much better, it may be worth considering if it can be implemented as an option for Grisk AI SD. https://github.com/gammagec/Dreambooth-SD-optimized This is an implementtaion of Google's Dreambooth with Stable Diffusion. If something like this could also be realized with Grisk AI's GUI, it would change the world.
2022-09-25 07:10:53 +0000 UTCYou are welcome! The app will be free released or only in this paid patreon?
Javier García
2022-09-24 11:03:43 +0000 UTCNice!! Do you think will you be able to use both GPU to sum their VRAM and power procesing in one render? I also have 2 GPUs a 3080 and a 3060Ti
Javier García
2022-09-24 11:02:21 +0000 UTCGrats - this is awesome for you!
2022-09-24 00:46:50 +0000 UTCI’m glad you were able to meet your goal! Look forward to what’s to come. Would love to be able to use any model in this gui just by dropping it in models folder since a lot are starting to pop up. Though I quickly realized that didn’t work at all. Lol… I mean is it possible? I have the .ckpt files but like I tried just throwing them in the models folder. Though it didn’t load up in the gui. I guess they are called “fine tuned models?”
2022-09-23 20:11:43 +0000 UTCThat's easy, I developed Dain-App for months without been able to run the app because I did not had a Nvidia card. (Thanks to all my Patreons who helped me test it back then)
GRisk
2022-09-23 19:08:10 +0000 UTCMore game companys need to step up their game ;)
Ernest Taylor
2022-09-23 18:56:33 +0000 UTCThe fact you were able to make this much software with those specs.
Ernest Taylor
2022-09-23 18:55:55 +0000 UTCThanks. Yeah, it took a while for me as well
GRisk
2022-09-23 18:55:49 +0000 UTC
