So, I’d need the patreon access for me to get those files used in the video, right?
So, I’d need the patreon access for me to get those files used in the video, right?
Fi am developing image caption editing app haha 
HNice!
![Dazzastrous [4090]](https://cdn.discordapp.com/avatars/871389071686135838/c4a5f4d81d9eeb18d856295d692e2c07.webp?size=40)
Dhow does it read all those details its mad
HHas anyone trained SD3 yet?
DIve barely touched it
Dtheres a bunch of supposedly trained checkpoints on civitai
Dhaven't checked them out yet
Fye
Fwill publish
Fman i like cladue 3.5
F
ADr did you release live portrait video2 video?
Fyes
Fit is on our patreon
Fdo a freshj install though update wont work
Fnice app getting completed
F
Fnow i need to add word replacement
Fimage caption processor almost ready :d
Fworking awesome
_Alright, any suggestions on how we could add virtual try on feature to this gradio app?
Fis is supposed to be easy or hard to train?
_In this gradio web application,
NI recently I used @Dr. Furkan Gözükara config and details are insane, likeness is mindblowing but I wonder why with western/european/latin people are amazing but with asian people is very dificulty to get likeness
_I’m getting an error after I followed this video to install and convert juggernautXL to diffusers.
Nonetype object has no attribute ‘tokenize’ -> this is the error I get when I try to use that coverted juggernautXL model.
Error:
Nonetype object has no attribute ‘tokenize’ -> this is the error I get when I try to use that coverted juggernautXL model.
Error:
Fbecause of the base models
Fbase models are lacking asian people
Ffor asian people give it a try to base sdxl
Fok that explains
FjuggernautXL 10 must have different architecture
DI tried many sd3 trainings the results are great for 1 subjet, I trained many subjects at the same time and I get a lot of bleeding from one caracter to another, probably is becase by default in koha the TE training is disabled, now I training with TE training at same LR as unet, I will report my results
Fgreat research
Ddid you hade bleeding problems training multiple subjects? do you think TE training can be related?
Fbleeding happens for 2 reasons
Fu-net and using same tokens
Fyou need different class token if you want to minize bleeding
Fand of course text encoder training
Fso like
Fohwx man
Fbbuk woman
_Can you suggest a juggernautXL version that works with this code?
Fdid you try 9?
Fi assume it would work
Msenor Gözükara, I sag your post on CogVLM 2, and im now wondering if about captions. Are full sentences like in your example "the image featurees a man standing indoors, with a neutral expression on his face. [...]" better then buzzwords like "ohwx man, close-up front view, facing camera, wearing brown blazer over graphic t-shirt, slicked-back brown hair, groomed mustache, urban background with blurred street and greenery,"?
Dwith your setting for xl I can train many subjets of the same class with no bleeding at all and great results
_Yes, it did not work with both v8 and v9, let me know if there’s a working version. I even tried RealvisXL_v4 too and got the same error: AttributeError: 'NoneType' object has no attribute 'tokenize’
Don xl I trained many loras and finetunes many subjects of the same class and no bleeding, the model learns the diference between person1 man and person 2 man, in my sd3 tests all bleeds, is like is no learning that the combination of person1 man is diferent of person2 man, it seems to learning all mixed no mather what token or class Im using
DLets see with TE enabled, I will report in some hours when the new training is finished