Updated ComfyUI recently? Notice anything... different? Want to make it even more different? Open Settings via the little cog...
... and you should see the [Beta] Use new menu and workflow management:
What do you think of the new look? Anything broken?
2024-08-17 11:29:46 +0000 UTC View Post
Video - https://youtu.be/7AhQcdqnwfs
Flux is a great model with loads of potential, but when it comes to generating images with almost any art style it can struggle a bit. You may also wish to add your own face, a pet or indeed anything you have some images of!
Woodblock art style without a lora...
2024-08-16 19:16:20 +0000 UTC
View Post
Even more Flux.1 updates - workflow attached as usual :)
If you've not updated ComfyUI in the last 16 hours, then you won't yet have the new ControlNet support, so be sure to update now :)
Loads of other updates top the workflows:
Optimised node placement so that bypassing IMG2IMG is quicker
Added More notes for all the various options
Added ControlNe...
2024-08-13 20:20:28 +0000 UTC View Post
Many of us have become used to using a negative prompt when generating images using Stable Diffusion, so going back to CFG 1.0 and not using a negative prompt can feel a bit like losing a favourite tool of some sort
There is a newer version out which I haven't actually tried as yet, but hopefully it (or others) can do much the same thing!
Anyway, the thing I've been testing is with using Automatic CFG as it has interesting controls for sigmas. I'd found ending somewhere around 0.8...
2024-08-09 09:33:08 +0000 UTC View Post
Things are moving along nicely in the Flux.1 world and soon we'll have things like controlnets, loras! But, for now at least, we've got image-to-image and LLM assistants. These ComfyUI workflows are similar to the previous ones, but updated to use the standard "KSampler" and "Load Checkpoint" nodes. The checkpoint for this is different - 2024-08-07 20:30:02 +0000 UTC View Post
AuraFlow 0.2, AuraSR v2 and Flux? So many new things!
Flux is most certainly worth a try, but both models will benefit from fairly beefy systems with at least 24GB VRAM and 32GB RAM, so beware to go slow if you don't meet the specs.
4 steps, amazing quality... what will we see in a few months time, I wonder?
Links:
As with the previous AuraFlow 0.1, the latest release is very good at following prompts! This time around the image quality is also better, and there are some limited styles in there too. The main thing seems to be fairly decent text, so long as it's short.
2024-07-28 21:13:55 +0000 UTC View Post
Now you can ditch InsightFace and go fully open source with LivePortrait in ComfyUI! We've got a few model options available for the face cropper to provide image animation, video-to-video animation, and even live via your webcam :)
When recording your driving videos, try not to move around too much. Movement in the z-direction seems worst, but essentially try to keep as still as possible. 2024-07-27 17:20:00 +0000 UTC View Post
Even more PixArt! Things seem to be dropping every day, and now we've got a 2-stage pixart. While you can use each model individually, it's better to go with the "refiner" style workflow and let the first stage do it's job with composition and colours, with the second stage adding more clarity and detail.
Models can be downloaded from:
So many new things, so little time!
PixArt Sigma 900M, more playing with the new samplers & schedulers, a better version of rembg and of course the new ControlNet Union ProMax model!
As you've probably seen in the previous bonus workflows, I do like the new Euler cfg++ :) While I originally went up to 50 with it using a very low CFG, slapping a perturbed attention guidance node into ...
2024-07-20 18:20:01 +0000 UTC View Post
Just for giggles, here is a quick workflow I'm playing with at the moment using SD3 as the refiner for Aura Flow. Now you can do complex prompts AND have the image quality come out a little better. Shame it can't do people well, but hey!
On the bright side, you can now more easily create hybrid creatures :)
https://huggingface.co/fal/AuraFlow
Could an unseen benefit of Stability Ai's license gaff be that we finally get a deluge of truly open source models? Let's hope so, as the AuraFlow is here!
It's only been out for a few hours now, but as a super-quick overview:
Limited Styled
Some te...
2024-07-12 18:42:00 +0000 UTC View Post
Looks like we can finally use SD3 for commercial purposes - subject to various limitations. Full details available at https://stability.ai/news/license-update
The workflow attached below is the one used in the video, and is used to compare SD3 vs Pixart Sigma, as well as using SD3 as a refiner and HiRes fix. It us...
2024-07-07 17:40:01 +0000 UTC View PostGiven all the new samplers and stuff, I figured why not see how well they do with highres fix as well. Not too bad, it turns out!
Anyway, this Pixart Sigma + SDXL workflow is designed to put your selected face where it sees fit in your generation. It starts of with Pixart and then refines using SDXL. I'm also doing some upscaling too, so it does have a fairly high VRAM requirement there.
There are 3 levels of "highres fix" with the final face replacement at the end. Each level add...
2024-06-30 17:40:26 +0000 UTC View Post
There are loads of new samplers in ComfyUI, so I figured I'd try them all - along with the new GITS scheduler as well!
To cut a long story short, I like DEIS and both of the CFG++ ones, but there is loads of scope to play with all of them!
* DEIS, GITS, iPNDM - https://github.com/zju-pi/diff-sampler
* CF...
2024-06-29 18:41:01 +0000 UTC View Post
It's the battle of models with reasonable licenses - Lumina Vs Hunyuan Vs Pixart!
Video link - https://youtu.be/aV_Feh8768M
Which follows prompts best whilst also having a decent style and composition? Well... personally I'm still liking Pixart ;)
Attached below are the workflows used in the video, including the e...
2024-06-22 19:20:00 +0000 UTC View Post
It can be difficult deciding which model is best for your task, so I've created a set prompt to test your model's creativity! Using a wide variety of subjects and styles, you'll soon be able to tell what your model can and can't handle. Try giving these a go with Stable Diffusion 3, for example!
ComfyUI workflows are attached :)
Negative Prompt:
ma...
2024-06-15 17:20:01 +0000 UTC View Post
https://github.com/Tencent/HunyuanDiT/tree/main/comfyui-hydit
This is their official set of custom nodes for ComfyUI, and while it currently needs >11GB VRAM, it looks like we may get a lower VRAM option coming soon :)
https://github.com/lllyasviel/Omost
Want an LLM to generate prompts for you? Maybe it'll get it Omost right! Yes, the puns abound with this latest release from lllyasviel. With more of a large language model style ch...
2024-06-01 18:55:01 +0000 UTC View Post
Khoj is touted as being your "AI second brain". Get answers to your questions, whether they be from online sources or your own notes. Use hosted or local LLMs. Access from Obsidian, Emacs, Desktop app, Web or Whatsapp.
Overall, it's not too bad and the install was pretty painless. RAG is probably going to be everywhere soon, making now as good a time as any to dive into the...
2024-05-27 14:20:01 +0000 UTC View PostWhile various things keep coming out which use the research-only "insightface" models, I figured why not combine two insightface-free options together? On top of that, maybe see how quickly I can make that cartoon or painted face into a more realistic one? We've had a load of new nodes come out recently, so now is as good a time as any to slap a load into a single workflow.
Below you'll find the ComfyUI workflows for both real and anime preset "Refacer" examples.
You'll nee...
2024-05-18 15:20:01 +0000 UTC View PostInpainting can be fun, and thanks to some of the latest BrushNet SDXL and PowerPaint V2 models, you can can use any typical SD1.5 or SDXL model for inpainting - no special inpainting version required!
You'll find all the example workflows I used in my latest video as attachments to this post, and for the downloads see the links below.
Enjoy!
Links:
ComfyUI custom node: https://github.com/...
2024-05-12 17:02:01 +0000 UTC View Posthttps://github.com/dnhkng/GlaDOS
"This is the Personality Core for GLaDOS, the first steps towards a real-life implementation of the AI from the Portal series by Valve."
The main thing about this one, is how fast it can generate responses :) You can actually have a (nearly) normal conversation!
The attached installation bash script is designed to be run on a standard, Linux+Nvidia system in whatever co...
2024-05-04 14:00:09 +0000 UTC View PostYou get 3 options, basically:
1. Their 1-step SDXL UNET
2. Any SD1.5 model, but with their LORA
3. Any SDXL model, but with their LORA
https://huggingface.co/ByteDance/Hyper-SD
1. Download ComfyUI-HyperSDXL1StepUnetScheduler to ComfyUI/custom_nodes
2. Download Hyper-SDXL-1step-Unet-Comfyui.fp16.safetensors to ComfyUI/models/checkpoi...
2024-04-26 21:39:00 +0000 UTC View Post
https://huggingface.co/PixArt-alpha/PixArt-Sigma/blob/main/PixArt-Sigma-XL-2-2K-MS.pth
There's a NEW Pixart Sigma model! This one pumps out 2K images - all in less than 6GB VRAM.
Same nodes as with the previous Pixart Sigma 1024 model, just some different setting...
2024-04-23 19:07:26 +0000 UTC View Post
Pixart Sigma - https://pixart-alpha.github.io/PixArt-sigma-project/
We're seeing more T5 conditioning over CLIP, often with much better prompt adherence. This time, it's the recent release "Pixart Sigma".
Use your own workspace and python environment:
cd ...
2024-04-20 14:20:01 +0000 UTC View Post
Just a quick Text-to-3d image workflow for you, in case you've not seen Instant Mesh as yet!
This one attaches my previous "Layer Diffuse" workflow to create suitable images, which instant mesh then turns into textured OBJ files :)
Install ComfyUI-layerdiffuse
So many workflows, so many IP adapter nodes to change ;)
Old IP Adapter Apply now IP Adapter Advanced!
Just a quick update to the previous Batch Image Animation workflow as various nodes have changed.
Enjoy! :)
2024-04-16 12:21:19 +0000 UTC View Post
"The horizon shimmered with a kaleidoscope of hues, a huge towering, iridescent crystal structure erupts from the ground, their facets refracting the vibrant pink and indigo sky. Amidst this otherworldly landscape, a lone figure with his back to us stands wearing an old, tattered, black cloak and a sense of ethereal grace."
Video guide: 2024-04-11 15:20:01 +0000 UTC View Post