Flux Tools For Low VRAM GPU's | Introduction to Inpainting & Outpainting

Thanks for the guide , it is possible to run redux with low vram ?

We are waiting for lllyasviel to attach it to Forge

Thanks. Do the Redux next, then Depth, then Canny

I'll go play Project Zomboid, I recommend it

Huh. What is the difference between xlabs and Shakker-Labs controlnets of canny/depth. Why is this special? We already have two of them. Someone please explain.

This will be a good series of videos

SwarmUI tutorial please

ya did it again , great job as usual 😊😊
only trouble is i've been using foocus for in/out painting and now you've made me want to try it in comfyui grrrrr lol
I gave up on comfy cause the update broke the LLM generator i was using in it , come to mention it i cant even remember the name of the generator now .. damn🤣🤣🤣🤣

Thank you, I just tried outpaint and inpaint truly amazing quality

Been waiting for such things .. thanks for sharing ❤❤

❤❤❤❤

Could you please do a video on crop and stitch with Flux tool inpainting?

Redux will be fun for MJ to deal with.

Great to follow. Updated Comfy but returning this:
RuntimeError: Error(s) in loading state_dict for Flux:
size mismatch for img_in.weight: copying a param with shape torch.Size([3072, 384]) from checkpoint, the shape in current model is torch.Size([3072, 64]).

by the way, depth and canny would be the best to see

Hello, my dear friend, please introduce my page so that it can be seen, thank you.❤❤❤

I finally got to do the inpainting I needed from Flux. On 12Gb VRAM with the full 'fill' model it was a lot quicker than I expected. That's the only one I've tried so far but with how well it worked I'm looking forward to the others.

Can't wait. I found flux less effective in design rather than SDXL.

You have 8vram ? I have 8vram and it crash on all non gguf models . How are you able to load 11 gb model on 8 vram 😮

Thank you for sharing, always good to see your videos! I was let down earlier this morning when ComfyUI announced zero day support for Flux tools and saw the models were 23GB in size. I'll give it a test now and see how they go.. Thanks again, and looking forward to your next fun video on Redux with ContolNet.

Good video. Anyone else getting really bad quality outputs for Outpainting.
Even the original non Outpaint area looks bad.
I am using FluxFillPP8_v10 safetensor model from the description.
Also should it take 6 minutes for 1 image(768x1344) using 8Gig V Ram.

These are so cool! Can't wait to dive into these! Thanks for sharing the info! 🙌🙌

With inpainting, it disturbs the composition of the image, the results are not that good, so are the results for the outpainting, the final images are distorted at the edges and lost details, using the fp8 model.

Thanks for the vid! And do you know if the flux.1-fill-dev (23Gb) version is an extended version of the original flux.1-dev? or a whole new thing, and you have to install both?

By the way, there is Flux Tools support for SwamUI and SDNext. Fingers crossed that Forge adds this when the updates get done soon!