is it lora so i can use 10 images of the product and assk it to generate something
is it lora so i can use 10 images of the product and assk it to generate something
SI didnt exprince it either , however, what about, give a try to, rare token word (xmycompanylogo) , and train 10 images just the logo, you want to see... after training to typo , "a woman wears a green shirt that xmycompanylogo is over it. " .. after train , you may want to extract the the training as a lora.
Fi am starting 8x A6000 GPU today for extensive testing
Fif you can have a big dataset with proper caption training is same
Fif single product image yes
Fstarted flux testing
JHello, was at vacation but back again now, do you have any training results for your Lora Training for Flux? And also how could i achive the same results with afterdetailer but for flux since it does not work in A1111
Fswarmui segment feature
FYouTubeSECourses
Do not skip any part of this tutorial to master how to use Stable Diffusion 3 (SD3) with the most advanced generative AI open source APP SwarmUI. Automatic1111 SD Web UI or Fooocus are not supporting the #SD3 yet. Therefore, I am starting to make tutorials for SwarmUI as well. #StableSwarmUI is officially developed by the StabilityAI and your m...
Fsearching flux atm
JAh okay thx
JWhat u mean with seachring flux atm?
Fsearching parameters
I@Dr. Furkan Gözükara Forge has a new tab that's called spaces, that install separately gradio extensions, maybe you can put any oneclick installer to promote you
NT5 FP16 separate download and dev version from darkwood download? I'm sorry I didn't follow sources much and I'm confused in them, there are many different models everywhere, I'll be especially grateful if you give me sources from where to download or tutor
Jgood info for Forge here: https://github.com/lllyasviel/stable-diffusion-webui-forge/discussions/1050
GitHub
The old Automatic1111’s user interface of VAE selection is not powerful enough for modern models . Forge make minor modifications so that the UI is as close as possible to A1111 but also meet the d...
Jthis is how it looks for me:
Ybeen training on flux and wow so great, I've been typing my captions out by hand, what tool goes well with the Clip-I?
Ji use Florence 2 for this, my Flux Loras looking really good, but i am still in research mode 
YThe man! .....not assuming anything tho..... XD
YIt can be used for a clip seg? How indepth of a description can you give, I have a hard time seperating one man from another with SAM
�I've tried Florence2 for a training and it went good, was setting up another Lora today but this time I actually read the captions first and noticed Florence got some key details very wrong, which could possibly mess up the Lora a bit so I dunno. I wonder if Llava is better or there's a free local Captioning tool that is a lot more consistent
Inot works for me i see the creation image but after grey image, what's wrong?
Jmust be something with the sampler type, i got it too, but it works correct with the right one, try normal and others
Ji always read the captions and edit them manually if i see something is wrong
DI'm doing some testing training flux on kohya but I'm having a bleeding problem, I'm training pictures of myself as a kid and as and adult and a friend, so the captions are "myname man" "myname boy" "myfriendname man" and the concepts bleed, this problem doesn't happen training sdxl models, there is no bleeding. maybe the problem is because I'm not training the text encoder?
DI hade a similar problem with my sd3 testing
Fye not training text encoder will cause
Fmore bleeding
Fsince you only train unet
Fclip l training not supported yet
Dohh thanks, so may be is that
II've found it was the vae, are for shell not for dev
DAnother question: the CLIP models, such as clip_l, are stored in separate files from the base model. If I use a concept that doesn’t exist in CLIP, such as ohwx or another non-existent word, would it still work during inference if it's not part of the model because the model was not trained? Additionally, if I train clip_l, would it be saved as a separate file? Currently, the text encoders are not in the same safetensors as the base model
Fit will work but lesser degree
Fsince you will train unet of the model
For if you have another question explain in more details
Fstarted second 4 gpu machine
YYeah, it a lot faster to edit prompts than write them
DThe doubt arises because the text encoder is in a different file, and if I train it, would Koha save it as a separate file? In that case, should I also use my custom clip_l for inference? Ideally, the text encoders should be included in the same SafeTensor file, as happens in SDXL
Fif you trained you have to use it during inference for best quality
Dthanks, I hope in the future the text encoders can be embebed in the model file
Fkohya will support very likely when supporting clip l training
Dthanks!! I hope so!! and I hope that 24gb vram will be enough to train flux + clip_l
Fcurrently only unet trained
Fwith kohya lora
F24 gb barely sufficient
Fupdated post with experiments
DI will check it