Seacourses only used the 1.5 original and that model is old for me and not so impressive in creativi
Seacourses only used the 1.5 original and that model is old for me and not so impressive in creativity
PThe model isn't flexible at all. Most likely due to the lack of filewords
Pyeah, I originaly trained on 1.5 model. it was okay but flexible. but then I started experimenting with other models and got better results but they weren't as flexible.
P2k photos should give solid results I imaging. lol
Pdid you use filewords?
Swell if you watch videos they all talk about using few photos like 20 good once
Syes. I fileword the clothing
Sand haircolor
Sthis is also ai
Pthat's my next step. up the dataset count and filewords meticulously
Sand this is a real photo of her
Pnice results
Svery.. I tell you if you use dreambooth and make normal models you are wasting time
Skohya lora, only way to go
Pguess I have a mission for today
Pgot any good resources to get started?
Swhat do you mean by resources?
Plinks to tutorials... etc.
Soh and train with realistic vision
Sno, cause all tutorials do it wrong
Plol
SI tried folloing tutorials which gave me crap
Sdont listen, they only focus on either stylized photos of their own faces
S512,512 is NOT a a good idea
Sjust do one thing.. when training put the unet at 0.0001 or 0.00008
Ssomewhere in between
S
Smy settings
Salso dont use Regularisation photos, they messed up my faces badly
TLora in khoya seemed not good in training with faces a all i tried multiple times yesterday it seems dreambooth is working well from the video, im testing it now ill be back with my results
Ti always have this message: Please configure some concepts
Scan you share results
Scomparisson
Sbut key is to have high res photos. if you crop them to 512 you wont have good results
TAnyone solved this problem? Please configure some concepts
PI don't even know where to install khoya. lol
Sah
SPcheers
Danyone know what "lora extended" option means and does?
Ti think i wasted my time in this video
Swhat do you mean
Dit does that, just press train again and it will go
Mim new to making lora thing, and when i create lora on kohya, from 20 img and 100 img source, my file size only 9mb, so what happend with that? i see ppl make lora above 100mb
Mi mean i try using 20 img as source first and 100 img latter, but size still 9mb
VDo you guys know any information on how to utilize ControlNet to indicate the location of eyes and mouth in a sketch when working with Stable Diffusion?
We have the openpose, but as far I know, it works only in humanoid images.
We have the openpose, but as far I know, it works only in humanoid images.
Snetwork rank decides the size
Sless rank, more compression I guess
Mdefault one is 8, lest say i change 128, so 128:8=16x9=144 yeah maybe right
TI always do that every time it gets me that problem but Solved it now by changing how a folder of images should path. Fix it by getting inside the folder and copy directory. weird
THe only teach how to run the lora in khoya but didn't get the main goal of producing the precise training out put from the image subject. He cut the the video and overlay a picture of him not knowing how everything has done. I tried hus settings but none works for me, so i moved back to dreambooth training
