🔥 90% OFF - Perplexity AI PRO 1-Year Plan - Limited Time SUPER PROMO!
https://redd.it/1o5iu30
@rStableDiffusion

Hunyuan 3.0 second atempt. 6 minutes render on rtx 6000 pro (update)

https://redd.it/1o5o3ka
@rStableDiffusion

Discord Server With Active LoRA Training Community?

I'm looking for a place where you can discuss techniques and best practices/models, etc. All of the servers I'm on currently are pretty dormant. Thanks!

https://redd.it/1o5nmt9
@rStableDiffusion

Coloured a line art using Qwen-Edit and animated using Wan-2.5

https://redd.it/1o5ptwe
@rStableDiffusion

Why are we still training LoRA and not moved to DoRA as a standard?

Just wondering, this has been a head-scratcher for me for a while.

Everywhere I look claims DoRA is superior to LoRA in what seems like all aspects. It doesn't require more power or resources to train.

I googled DoRA training for newer models - Wan, Qwen, etc. Didn't find anything, except a reddit post from a year ago asking pretty much exactly what I'm asking here today lol. And every comment seems to agree DoRA is superior. And Comfy has supported DoRA now for a long time.

Yet, here we are - still training LoRAs when there's been a better option for years? This community is always fairly quick to adopt the latest and greatest. It's odd this slipped through? I use diffusion-pipe to train pretty much everything now. I'm curious to know if theres a way I could train DoRAs with that. Or if there is a different method out there right now that is capable of training a wan DoRA.

Thanks for any insight, and curious to hear others opinions on this.

https://redd.it/1o5t7z0
@rStableDiffusion

Chroma on the rise?

I ve lowkey seen quite a few loras dropped for chorma lately, which makes it look really good like on par with wan t2i or flux. And was wondering if anyone else has noticed the same trend or if some of you have switched to Chroma entierly?

https://redd.it/1o5r8dy
@rStableDiffusion

Qwen edit image 2509 degrading image quality?

Anyone finds that it slights degrades the character photo quality on its outcome? Tried to scale to 2 times and it is slightly better upon viewing up close.

Background of it is that I am a cosplay photographer and am trying to edit the character into some special scenes too but the outcome are usually abit too pixelated on the character face

https://redd.it/1o60quo
@rStableDiffusion

What image gen created this?
https://redd.it/1o5xuli
@rStableDiffusion

Introducing Silly Caption

https://redd.it/1o6110g
@rStableDiffusion

Issue Training a LoRA Locally

For starters, im really just trying to test this. I have a dataset of 10 pictures and text files all the correct format, same asepct ratio, size etc.

I am using this workflow and following this tutorial.

Currently using all of the EXACT models linked in this video gives me the following error: "InitFluxLoRATraining...Cannot copy out of meta tensor, no data! Please use torch.nn.module.to_empty() instead of torch.nn.Module.to() when moving module from meta to a different device"

Ive messed around with the settings and cannot get past this. When talking with ChatGPT/Gemini they first suggested this could be related to an oom error? I have a 16GB VRAM card and dont see my GPU peak over 1.4GB before the workflow errors out, so I am pretty confident this is not an oom error.

Is anyone farmilar with this error and can give me a hand?

Im really just looking for a simple easy no B.S. way to train a Flux LoRA locally. I would happily abandon this workflow is there was another more streamlined workflow that gave good results.

Any and all help is greatly appreciated!

https://redd.it/1o62wec
@rStableDiffusion

ByteDance just released FaceCLIP on Hugging Face!

https://redd.it/1o655q8
@rStableDiffusion

How To Fix AI Skin?

What are some sites or tools to fix AI looking skin?

I know of Enhancor and Pykaso but have not tried them yet because both don't offer free trials.

https://redd.it/1o669ej
@rStableDiffusion

New Wan 2.2 I2V Lightx2v loras just dropped!
https://huggingface.co/lightx2v/Wan2.2-I2V-A14B-Moe-Distill-Lightx2v/tree/main/loras

https://redd.it/1o67ntj
@rStableDiffusion

VACE 2.2 dual model workflow - Character swapping
https://www.youtube.com/watch?v=MHQDiSgu1VM

https://redd.it/1o697kx
@rStableDiffusion

70 minute of DNB mixed over an AI art video I put together
https://youtu.be/x6GNt-g1HJo?si=cqeo29YRbsjwepgO

https://redd.it/1o6b9s5
@rStableDiffusion

How to convert 3D images into realistic pictures in Qwen?

https://redd.it/1o6b66r
@rStableDiffusion