Color Anchor Node Flux2Klein

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

Built a local browser to organize my output folder chaos -- search by prompt, checkpoint, LoRA, node type, etc

https://redd.it/1siqf2v
@rStableDiffusion

Why is Wan 2.2 N.S.F.W Remix Lightning Model so much better at things like hair flip, hair combing and feminine energy than regular Wan?

I am not talking about actual N.S.F.W I am talking about the model that has such a name in it, and just feminine energy, seductive performance, shampoo commercial hair toss, sensual movements, elegant leg cross sitting on bar stool.

Whenever I use any of these WAN models it comes out very static and it ignores the prompt, when I use the remix it comes out nearly perfect.

It's almost like using Grok, not the new Grok but the old one before it was censored.

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

Sharing my creative node suite for ComfyUI

Hey guys, Winnougan here. It's time to give back to the community. I've been growing my nodes suite on GitHub, which started out as the nodes that I personally wanted to make life easier in ComfyUI. I'll keep adding to them to make my overall ComfyUI experience faster and user-friendly. Enjoy the nodes and happy gooning!

1. Resolution picker: too many presets to count plus custom height and weight if that's your thing. Visual icons to easily pick what you want. I do a ton of high res images, so this helps me out a lot.

2. LTX and Wan resolution picker: I cobbled together all the best resolutions for these video models and made it easy to pick and choose what you want

3. Power Lora Loader: I wanted to add and remove loras quickly. I have thousands of loras stashed away, so I decided to make it easy to search for them by visually. Easy to adjust the strength and toggle on and off, move up and down or remove them.

4. The beloved Cache Dit series: regular cache dit, cache dit for Wan2.2 and cache dit for LTX-2.3. Visually shows you how it speeds up your workflow.

5. More to come! Stay tuned as I'll be adding a ton more nodes to my suite.

Grab the suite here: https://github.com/Winnougan/winnougan-nodes.git

Or, in the Comfyui Manager by typing "Winnougan": or in your custom_nodes folder do a "git clone https://github.com/Winnougan/winnougan-nodes.git"

https://redd.it/1siu2zd
@rStableDiffusion

What are the current best models quality-wise?

Lots of models get attention for being able to run fast or on low VRAM or whatever but what is currently considered state of the art for local Image, Video, audio, etc... generation?

I've been around here since the first days of stablediffusion and when A111 was the go-to, but I've always had a system with only a 2070 super, so 8GB VRAM and few supported optimizations. As such I've only really dealt with GGUF models and quants that worked on lower-end systems and am not as caught up on what the best models are if resources aren't an issue.

I'll have a system with a 5090 soon to try some of them out but I'm curious what you guys would rank the highest for the various models, be they straight text2image, image edit, video models, music, tts, etc...

I'm sure quite a few people would benefit from this since the leaderboards are constantly shifting for models.

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

OstrisAI-Toolkit Lora --> Anima model.
https://redd.it/1sj0llg
@rStableDiffusion

LTX2.3 - LTX-2.3-22b-IC-LoRA-Outpaint

https://redd.it/1sj0yhn
@rStableDiffusion

Okay, I have to admit LTX is better after all. I have fully dropped MagiHuman after your critiques.

https://redd.it/1sj5vuh
@rStableDiffusion

Tansan (Anime Portrait) LoRA for ZiT

https://redd.it/1sj77bh
@rStableDiffusion