but you can generate images
but you can generate images
F--medvram
Fif not works
F--lowvram
MI got that output when it was generating img2img
Mwill try 
Mit worked thanks
MI have a question, when I upload my face in img2img, it doesnt show my face, and even if there was a face it barely looks like me
Fit doesnt work that way
Fyou have to teach your face
Moh ok, another question might be my last, in the terminal it says I want to do that how can i do it I cannot find launch()
TIn the file webui.sh where it says args type --share then save this file and run again
Rgetting this:
>git pull
Already up to date.
but it hasnt been updated in over a week, i think its broken
>git pull
Already up to date.
but it hasnt been updated in over a week, i think its broken
MI think you need to save the new one that is on git
Fcan you show screenshot
Fit has been updated many times this week
Fcontrolnet extension
Ri think i figured it out, deleted dreambooth extension folder and it let me update everything
BGetting Loss=0.00 during training, what does it mean?
Hhi, i need help finding were is override setings ?
Sif you want better faces and inpainting, train a model of the person where you only have closeup crops of the face. super tight, then use that model for inpaint. You can also use both loras at the same time, like 0.4 strenght on each
H
Soverride means someone used those settings when making the seed
Sand you dont
Sso its overriding your settings to match
Vdo you guys know any model trained to generate only fantasy images? like orcs, elves, goblins, and so on...
V
BAnything v3.0 is amazing for that
Vnice!
Vi'll try it
Fanything v3
F
Fanything v4
F
Fdeliberate
F
Fprotogen
Fbase 1.5
Fsame seed
Hdo i have to use diffrent settings if im training diffrent kind of style characters?
Fwhat do you mean
Hlike lets say im training video game character do i have to use some kind of "unrealistic character" parameters in kohya
Hcuz i been using same paramaters for every characters from now on and im not getting good results i dont know if the problem is my paramaters or my dataset
H
Hi been using theese parameters does it looks good? (idk anything about settin parameters)
Fi didnt use kohya so cant tell that
Hi just remembered i had a question for you
Hi tried making lora with your dreambooth lora video
Hbut every lora that i did with dreambooth was always bigger than 4 gb
Hdoes dreambooth lora tends to be big or was i making something wrong?