but is it worth trying this approach?
but is it worth trying this approach?
Fi dont think so
Ffirst of all assigning 5000 unique tokens
Fwould take huge time
Fand you dont know if will work or not
Falso 80k images you really should use multi gpu
Fyou can run on runpod really cheap
For you can wait days :d
Nyeah planning to use multi gpus. I have unique ids for each product so planning to put some random alphabets that is not in the clip vocabulary. I have tried this with 500 products and I got decent results. my concern will the model remember all the tokens when I scale it to 5000
Fwith sufficient number of epochs i think should
PStill not seeing anywhere near a 70% increase.
25% increase, yes.
70%, no.
How are you equating this percentage ratio?
25% increase, yes.
70%, no.
How are you equating this percentage ratio?
Fwhat is your gpu
Fi am installing tensoRT right now on my new model
P4090
Vi have a ComfyUI workflow question. anybody have a good method of inpainting large images so that the sampler only works on the masked area?
example, i arrive at a large-ish image (3000x3000) by upscaling. I need to fix parts. if i use the Mask Editor and push it through the KSampler (VAEencode > Latent Noise Mask > KSampler) it seems it still works on the entire image rather than just the masked area. In A1111, I recall it just working on the masked area.
How would this be done in ComfyUI?
example, i arrive at a large-ish image (3000x3000) by upscaling. I need to fix parts. if i use the Mask Editor and push it through the KSampler (VAEencode > Latent Noise Mask > KSampler) it seems it still works on the entire image rather than just the masked area. In A1111, I recall it just working on the masked area.
How would this be done in ComfyUI?
Fhere my difference
F
F55%
Fbut there are so many other apps running
Ftherefore not max
Fperfect face quality and resemablence
BGot a link to the latest SDXL kohyah config training preset?
BLORA training that is
Fi prefer dreambooth sdxl
Fthen extract lora
Bdoes dreambooth support bucketing?
Fbucketing is script feature
Fsdxl supports it
Fso kohya handles it for you,
Bhave you got a link to DB config?
Fyes
Fdownload 24 gb and enable bucketing
F
BGreat thankyou
Bdo you have a link to the LORA training config also?
Fye but it is far from being very good
BOk great thanks, I'll try your dreambooth extract method also
BWhat settings do you reccomend when extracting the LORA?
Ftry
F128x128
F192x192
F256x256
BOk thanks
BDo you recommend BF16 as the "save precision" for a 3090?
AHey, I'm a total SD beginner, just installed SD on Kaggle following the tutorial.
I'm looking for guidance on what learning path to choose, which tutorials to follow.
My goal is to create specific kind of content: anthropomorphic animals, specifically animals in music bands
Hope you can point me in the right direction.
I'm looking for guidance on what learning path to choose, which tutorials to follow.
My goal is to create specific kind of content: anthropomorphic animals, specifically animals in music bands
Hope you can point me in the right direction.
R@Dr. Furkan Gözükara having some trouble with retinaface installation. it doesnt seem to like Python 3.10.11 but it was fine yesterday?
Rtried
pip install tensorflow==2.5.0
pip install retinaface==1.1.1
pip install tensorflow==2.5.0
pip install retinaface==1.1.1
Rand
pip install tensorflow==2.1.0
pip install retinaface==0.0.6