Software Engineering Courses (SECourses)•13mo ago

Is it possible to customize the UI for forge ? Like adding brand logo, rearranging features Not to

Is it possible to customize the UI for forge ? Like adding brand logo, rearranging features

Not to delete any existing features, just to rearrange them so that the whole visual look is less complex and has personal visual taste @Furkan Gözükara SECourses

SSebastianblood Is it possible to customize the UI for forge ? Like adding brand logo, rearrangi...

Furkan Gözükara SECourses•11/25/24, 1:07 PM

i dont think

Furkan Gözükara SECourses•11/25/24, 1:25 PM

ViolA_ChaoS•11/25/24, 2:59 PM

Hey everyone! I've had mixed experiences with different Flux models and I'd like to share my findings and ask for help:

Successfully trained a LoRA using flux1-dev.safetensors with Dreambooth following the tutorial video. The results were excellent - I can generate both realistic and stylized images with great quality. However, the pose control with this model is quite poor.

When Flux tools were released, I was excited but encountered two issues:

When I try to use my previously trained safetensors (trained with flux1-dev) with the Flux tools, it doesn't work well - generates very poor quality images.
So I tried training a new Dreambooth LoRA using flux1-canny-dev.safetensors, but got this error:RuntimeError: Error(s) in loading state_dict for Flux:size mismatch for img_in.weight: copying a param with shape torch.Size([3072, 128]) from checkpoint, the shape in current model is torch.Size([3072, 64]).

Has anyone successfully trained a Dreambooth LoRA with flux1-canny-dev.safetensors? If so, could you share your setup/approach?

I really want to combine the quality I achieved with flux1-dev with proper pose control from the new Flux tools.

AAleksAdv btw. what is resolution of your posture pictures you use for training? I've noti...

ViolA_ChaoS•11/25/24, 3:13 PM

I have achieved much better results when using these resolutions. Often you can miss the point of the image if you crop it and use 1024x1024. But it is very important that if you use a size that is not 1024x1024 you crop at least 8 images to that size.

VViolA_ChaoS I have achieved much better results when using these resolutions. Often you can ...

Furkan Gözükara SECourses•11/25/24, 4:17 PM

makes sense

VViolA_ChaoS Hey everyone! I've had mixed experiences with different Flux models and I'd like...

Furkan Gözükara SECourses•11/25/24, 4:17 PM

you need more pose having dataset

Furkan Gözükara SECourses•11/25/24, 4:18 PM

about flux tools today i started investigating them

Furkan Gözükara SECourses•11/25/24, 4:18 PM

so cant comment a lot atm

Furkan Gözükara SECourses•11/25/24, 4:18 PM

but you may needed to report the error

VViolA_ChaoS Hey everyone! I've had mixed experiences with different Flux models and I'd like...

druhl•11/25/24, 5:02 PM

Interesting.
But I wonder, are the flux tools same as a dev model? Are they even trainable?

FFurkan Gözükara SECourses single lora both characters this is best i got

FluffyQuack•11/25/24, 6:10 PM

I was able to get a pretty good result with a Flux LoRA I trained last month with two unique characters. But maybe the AI has has an easier time telling them apart when they're different genders. The dataset containing images with them posing next to each other maybe helped too

20241003_2156_An_oil_painting_of_a_ACNaoe_woman_with_d_flux1-dev_215589119.png

FFluffyQuack I was able to get a pretty good result with a Flux LoRA I trained last month wit...

druhl•11/25/24, 6:21 PM

Yes, standing next to each other is the key here.

VViolA_ChaoS Hey everyone! I've had mixed experiences with different Flux models and I'd like...

druhl•11/25/24, 6:53 PM

Tested this.
Feel like a better workflow is > Flux tools > then img 2 img using your dreambooth.

druhl•11/25/24, 6:56 PM

Random image from the internet

druhl•11/25/24, 6:57 PM

Flux1-depth tool

1753-An_opulent_magical_scene_featuring_a_re-flux1-depth-dev-947697242.jpg

druhl•11/25/24, 6:58 PM

img2img w trained dreambooth

druhl•11/25/24, 7:28 PM

Flux1-fill tool outpaint

�

🍭🎀 𝒜𝓋𝒶 𝐹𝓇𝒾𝑔𝑔 🎀🍭•11/25/24, 8:01 PM

Guys, getting this error when trying a pony checkpoint in swarmui:
"ComfyUI execution error: Given groups=1, weight of size [512, 16, 3, 3], expected input[1, 4, 144, 112] to have 16 channels, but got 4 channels instead" any idea how to fix this?

AIGambino•11/25/24, 8:37 PM

custom style trained with One trainer: 1980s cartoon style

FFluffyQuack I was able to get a pretty good result with a Flux LoRA I trained last month wit...

Furkan Gözükara SECourses•11/25/24, 8:45 PM

dataset containing images with them posing next to each other

Furkan Gözükara SECourses•11/25/24, 8:45 PM

different genders

AAIGambino custom style trained with One trainer: 1980s cartoon style

Furkan Gözükara SECourses•11/25/24, 8:45 PM

looks decent

Furkan Gözükara SECourses•11/25/24, 8:45 PM

outpainting of flux perfect

Furkan Gözükara SECourses•11/25/24, 8:45 PM

FFurkan Gözükara SECourses adding them to our unified downloader

AiInfluence•11/25/24, 9:36 PM

pls don't forget kaggle !

AiInfluence•11/25/24, 9:37 PM

Also Hi Doc have I missed anything

AAiInfluence pls don't forget kaggle !

Furkan Gözükara SECourses•11/25/24, 9:44 PM

i think fp8 versions needed for kaggle

Furkan Gözükara SECourses•11/25/24, 9:44 PM

i didnt see them yet

Furkan Gözükara SECourses•11/25/24, 9:46 PM

testing now

Elsewhere•11/25/24, 10:02 PM

Anybody have any issues with flux fill inpainting not really doing anything or too little?

EElsewhere Anybody have any issues with flux fill inpainting not really doing anything or t...

Furkan Gözükara SECourses•11/25/24, 10:04 PM

i tried outpainting works

Furkan Gözükara SECourses•11/25/24, 10:04 PM

let me test inpainting

Elsewhere•11/25/24, 10:05 PM

I been working on the same picture all morning and can't get it to do it-- I've used various workflows and what not; other models will work with the same workflows (but not the same effect)--

Just now, I finally tried another image; and said cow with a guidance on 30, and well it worked except it gave me a man, not a cow-- so I am really confused--

EElsewhere I been working on the same picture all morning and can't get it to do it-- I've ...

Furkan Gözükara SECourses•11/25/24, 10:10 PM

testing

EElsewhere I been working on the same picture all morning and can't get it to do it-- I've ...