when it comes to VRAM
when it comes to VRAM
FOneTrainer best
Dnow I'm training at 1024x1024 and the vram usage is identical what is very wired, the results at 512x512 were not so good, lets see at 1024 now
DI tried onetrainer with sd3 but the results were terrible, the model gets fried and no people likeness at all, the resulting images were terrible full of artifacts
Fit needs research
Fi plan to do but busy got 2 client work right now that i have to finish :d
Dthe loss looks much better at 1024x1024, in onetrainer the loss graph is all over the place, I think there is a bug in onetrainer and sd3 I'm using the default presets on both onetrainer and kohya_ss, at this moment kohya_ss is much better on sd3, my 512x512 test looked a little undertrained but quite good
Floss rates were never meanginful for me
Fwhen training SD 1.5 or SDXL
Tguys i have a big problem that started today and i don't know why, my comfyui is very laggy and the UI is super small, any idea ?
Ti feel like the resolution of the software got altered or something
edit : ok my chrome decided to put comfy ui in 15% scale resolution, super weird, but it fixed it to put it back to 100%
edit : ok my chrome decided to put comfy ui in 15% scale resolution, super weird, but it fixed it to put it back to 100%
Fi was gonan tell this
Dyes may be, but I was able to see some correlation on loss graphs, they tend to get lower over time during training, on onetriner and sd3 the graph looks ridiculous, is a mess goes up and down and the model gets destroyed very quickly
Dkohya_ss on the sd3 branch is very usable, it need some tweaks but is getting there
Fi am still waiting him to fix
F
FHey Dr. Furkan do you have a good article on captioning/training photographic styles in SDXL?
Flike for types of lenses or lighting, stuff like that
Pjust change the res. i trained it many times. quality is super good. and it trains 3x times faster than xl.
Fnot yet sadly
Fautomatic1111 new version published : https://github.com/AUTOMATIC1111/stable-diffusion-webui/releases/tag/v1.10.0
GitHub
Features:
A lot of performance improvements (see below in Performance section)
Stable Diffusion 3 support (#16030, #16164, #16212)
Recommended Euler sampler; DDIM and other timestamp samplers cur...
A lot of performance improvements (see below in Performance section)
Stable Diffusion 3 support (#16030, #16164, #16212)
Recommended Euler sampler; DDIM and other timestamp samplers cur...
Dgreat!! thanks. I'm just finishing a test at 1024x1024
MHey guys, I have a question for you. When training a SD Model and the results turn out somewhat disformed. What could that be?
M
M
Fresolution mismatch
Fwhat do you train sdxl sd 1.5? whichn model
MXL Pony realism
Mtrained with 1024
Mrendered 1280:768
Mhey and nice talking to you. mr stablediffusionsuperman. thanks for your work 
Fpony must be the reason
Fdid you try 1024x1024 render?
Fsame issue?
Freal vis xl4 is very good for training a person
Maight. I will try that
Mi was wondering if it have to do with learningrate or bad captioning.
Mbut
Msame settings but 1024.
Mbut still deformed
Mill give realvis xl4 a shot :=
M
Fye try it
Fif still same problem
Fit could be related to dataset or training setup error
Mill give it another shot and report back
Mthanks
Fsure
JIs there any quality comparison between “roop” and “rope”?
Fi dont have any yet but you can 1 click install test. if you do let us know 