did you edit metadata?
did you edit metadata?
F
Mmodel metadata you say?
Yes, I looked at what the boy said...
that was wrong
Yes, I looked at what the boy said...
that was wrong
SDr Furkan, for kohya trinaing there is another flux dev model that could be used for training instead fo flux dev 1?
RIf I want to further finetune a lora or a model, what is the best approach ? Do i need the original dataset ? or can i give the model or lora new images without the old ones and continue further tuning ?
�@Dr. Furkan Gözükara Is it possible to train a style and an object in the same Lora? You mention something about having two folders in the dataset preparation part and having a specific number of repeats for each of them.
RWhat are the fundamental differences between kohya, onetrainer and simpletuner? I've only used ai-toolkit (kohya) because there are many online resources, but I've seen information about SimpleTuner being better.. however, I am not sure why/what the differences could be
A
ASome stuff I made with rank 128
R@Dr. Furkan Gözükara can you suggest ?
If I want to further finetune a lora or a model, what is the best approach ? Do i need the original dataset ? or can i give the model or lora new images without the old ones and continue further tuning ?
If I want to further finetune a lora or a model, what is the best approach ? Do i need the original dataset ? or can i give the model or lora new images without the old ones and continue further tuning ?
MWhats the recommended ui/ config to use flux with controlnet?
VAre you, by any chance, using swarmUI through StabilityMatrix? I had a similar issue ( couldn't load any flux model whatsoever), and I highly suspect it's StabilityMatrix being wonky
KHi everyone, I need advice on setting trigger tokens and classes for two training datasets:
Dataset: ~150 diverse photos of peoples showing interactions like handshakes or couples holding hands etc.
What trigger token would best represent these interactions, and what class should I assign?
Dataset: ~25 photos of a physical product (e.g., a cooker - utensil).
What trigger token and class would you recommend for this category?
- Hands Interaction
Dataset: ~150 diverse photos of peoples showing interactions like handshakes or couples holding hands etc.
What trigger token would best represent these interactions, and what class should I assign?
- Physical Product
Dataset: ~25 photos of a physical product (e.g., a cooker - utensil).
What trigger token and class would you recommend for this category?
A@Dr. Furkan Gözükara Hey.. Thank you for all the info you have on training models.. Its truly amazing.. I read in one of your posts about rank and flux loras but I didnt get the point really.. I am trying to prevent overfitting to use characters as comicbook characters.. The loras are amazing for realistic but I find that they are difficult to convert to comicbook characters.. You wrot a post about less paramters not sure what you meant.. Do you suggest a low rank/alpha to prevent overfitting or what did you mean ? Thanks
M@Dr. Furkan Gözükara What's your recommendation for training with a minimum number of small number pictures? Any experinece with using 1 or 2? Effectively training a lora based on an AI generated character created with Flux.
Have you had experience with ControlNet for posing working successfully on Flux yet? I see Black Forest labs has released some files to assist with this, but was curious if you had worked with them.
Have you had experience with ControlNet for posing working successfully on Flux yet? I see Black Forest labs has released some files to assist with this, but was curious if you had worked with them.
W@Dr. Furkan Gözükara the newest of your ComfyUI install include DeepSpeed, does it improve inference speed or something ?
RHey I made a music video using veo 2, some of you might find it of interest, anyway its all AI (except the music). Hope its ok to post here.
https://www.youtube.com/watch?v=XICNVp7yhg0&t=1s
https://www.youtube.com/watch?v=XICNVp7yhg0&t=1s
YouTubeRuairi Robinson
Music video for Steve Moore's "The Blue Stone"
Directed/edited by Ruairi Robinson
Visuals generated using Google Veo 2
Directed/edited by Ruairi Robinson
Visuals generated using Google Veo 2
.@Dr. Furkan Gözükara Here a gen from a custom fine tune on large images I’ve tiled and cut into 1024x1024 pixel chunk separate pngs to capture every tiny possible detail from the up to 6K large images. It was an experiment and it went well with your config. Original image count was 15 for those large images and then after filing there were over 400 images. So this seems to be the trick to be able to train on any resolution images without losing any details.
RThis is absolutely stunning mate!
Jinteresting! how does the model know the tiles fit together when training?
Jmade a new flux fine tuning model of my friend with 58 images, also gave great and trustworthy results. im wondering how low it can go. my first attempt with 15 images was too glass-skin like. but yea at least now i know that 58 images works well!
.I've written a python script with chatGPT that will tile them with overlap and is also including the original full size image. Plus I'm using flux dev de-distilled.
.@julius so yeah the fine details the model can't train on when only the full image is trained on it's own since the image gets automatically downscaled to 1024x1024 so the training will never see the full details in your dataset. I've taken a step further and have 3 different resolutions for the original image in 2x increments so that it will have 1024x1024 which is the full image downscaled to give the training the full context of the subject then a 2048x2048 version which will serve for 2x Ultimate SD Upscale training on 1024x tiles and then a 4096x4096 size tiled in 1024x1024 tiles to serve as data for training for 4x Ultimate SD Upscale and lastly the original resolution whatever it is to be tiled into 1024x1024 tiles with overlap to not miss any detail from the original dataset.
.Here some more outputs.
.
.
.
NWow, there are so many neural nets that you've taken apart and assembled
DHas anyone used the nee deep face similarity app? It works great but I'm not sure what sort of numbers are considered very similar in terms of comparison. I know lower is better but is .3 good or still very off for instance?
RIt looks great. How long did it take you to complete the training process?
.Looks good on the low resolution. How do the close up renders come out or high res upscales?
VWhere are the models saved that are downloaded the first time I run the video generation for CogVideoX1.5-5B-I2V ?
MPlease, I've been here for two days and I can't do anything here anymore...
can anyone help
can anyone help
Fyes there are
Fhere my experiments
FMedium
Thoroughly experimented with Fine-Tuning / DreamBooth training of Flux-dev-de-distill, PixelWave v03, Verus Vision and base FLUX Dev model…
Fi rather prefer each time train from scratch with improved dataset
Fif they are different class yes can work
Fi trained a person and a handbag
Fthey use different scripts
Fbut if implementation is same should perform same
Fnice
A@Furkan Gözükara SECourses SOS 
I need a list of available stable diffusion classes
I need a list of available stable diffusion classesADoes it work with the T5Xxl?
Fi test manually there is no such thing
Ftest your prompt and see if generates similar thing
AThat's a good approach yeah
Fimprove dataset and train from 0 again - best way
A@Furkan Gözükara SECourses I saw your new upscale gradio , does it reach supir quality?
Fyou can data augment and train - crop differently - i trained with like 3 or 4 images person worked fairly good