VNCCS Pose Studio is a professional 3D posing, framing, lighting, and pose-library environment running directly inside a ComfyUI node. It is designed for building high-quality pose/control references without leaving the graph: adjust the character body, pose bones interactively, frame the camera, tune lights, manage saved poses, and output single images or pose batches.
Second version of Pose Studio lora for Qwen Image 2.1 is here!
Fixes:
Fingers on anime characters no longer distorted.
Added lights control system. Can still be not precise as in Pose Studio, but should work in most scenarios.
More control to head position.
https://huggingface.co/MIUProject/VNCCS_PoseStudio_QI2.1/tree/main
You must install https://github.com/AHEKOT/ComfyUI_VNCCS_Utils node pack to use this LoRA
Description
First version of Pose Studio lora for Qwen Image 2.1 is here!
You can download LoRa and workflow from my HF https://huggingface.co/MIUProject/VNCCS_PoseStudio_QI2.1/tree/main
You must install https://github.com/AHEKOT/ComfyUI_VNCCS_Utils to use this LoRa
FAQ
Comments (43)
Amazing. Really easy to use. Character consistency is excellent.
Suggestions:
1. Add a lock MP (1.0 default), so we can manually adjust the aspect ratio (width and height) to frame the characters. I have to scale the image to aspect ratio(lock 1.0MP) with another node, otherwise it would be a pain calculating the proper width and height to use in the VNCCS node.
2. Button to "fit" aspect ratio/frame to character bounding box.
Really impressive! Quite well thought out as well, this will be super useful, thanks for sharing!
After attaching pose reference image it says this:
[SAM3DBody] Download complete.
[VNCCS Pose Studio] pose_image SAM import failed: No module named 'roma'
just install the requirements.txt "pip install -r requirements.txt"
@lilylilith all Requirement already satisfied. nothing changes
@MokiMok31 then try to reinstall them. "No module named 'roma'" clearly says that python missing it's modules. roma is in requirements.txt for sure
Wonderful work. I did find one issue, when you add [Background] image, it will start tossing, failed to save workflow errors, spamming the warning toaster popups on the right side. Only way to stop it is to reboot ComfyUI.
Also going from one workflow and then back does some odd things to the 3D model, flies off into the nether. Otherwise it's been working great.
Wow, that is looking amazing, I would really love to see Krea2 version. 🩷
Awesome! been wanting something like this forever!
Please add a neck length slider and if possible a quadruped option.
This node threw an error during execution. Check its inputs or try a different configuration.
same -> KSampler
Bypass Qwen Image 2.1 cache node
@SQ19837675 This
@SQ19837675 Honestly I am lazy to tackle with this, don't seem worthy of my time, basically if a Lora or workflow works out from first or second try I go with it, if not I just leave it be.
@sarahbonestone i agree and am also TeamLazy but this is literally the equivalent of a toggle you need to hit once in the workflow and then never think about it again. not sure if you're using other workflows with Q2.1, but it's a pretty strong model it's worth flipping that toggle IMO.
@hrkumarjermain221 I use Pixaroma's workflows for Qwen 2.1, they are pretty bullet proof and simple to use, all his instructions, nodes and workflows works straight out of the box.
@sarahbonestone #underrated. to be honest i don't really need any loras with quen two point one i get pretty good results with flux klein two nine b without any loras as well either way i'm glad it's working for you.
The results are fantastic—I really love it! I’d also like to ask if there are any more presets available for the pose library. I’m specifically looking for adult-oriented pose presets. The poses I create myself sometimes don't turn out that great, so I’m hoping for a wider selection of poses. Please let me know if there are any available—thanks!
Just connect a Load Image node and drop whatever image you want in and it will work. Can also save the poses to the library.
@skullzy77 I instructed composer to write me a library based on the most common and Well known poses.
By the way, on another note, is it possible to create a 'Pose Studio for krea2'?
You can use it with any controlnet for Krea2 (or any other model). I don't know what have been released for it but should work. One downside - it will be able to create images but not edit them as krea2 isn't edit model
K-R-E A 2 is probably about the only model that listens to my prompts well enough. For me personally to not really need a pose studio. With a pose studio like this, its a great go to.
Looks good, but please don't force us to install large custom node packs just to access your custom tooling - if possible extract and provide just the necessary nodes needed for the pose control, etc. for future versions. 🙏
''Just to access your custom tooling. '' Have you ever developed a node pack like this before, or anything that even slightly resembles it? Do you have any idea how much time, iteration, specifications and customization goes into making something like that? This work. It's hours of trial and error, which is basically impossible to regulate and do right with existing nodes. Because guess what? Things with existing nodes already exist. And you don't want that. You want this. So say thank you, give the man a like, and move on with your day.
@hrkumarjermain221 I think request is very reasonable, and suggest how to improve software and rise popularity of it. That how open source evolve. And it does thank you in form "looks good".
@vision181211772578 What are you talking about? The whole project is open source. It has a GitHub. You can go there, take it and do what you want. It's already what you're saying it should be. What I'm saying is this person is complaining about having to use custom nodes and that they shouldn't have to use nodes that aren't standardly available, but to do the things this project can do involves using nodes that do not exist. That's why the whole project exists. If you don't understand how nodes work, you shouldn't be asking to adjust them and you shouldn't be telling other people what is reasonable or what isn't. The project is open source. The nodes have to be custom because if they were not custom, guess what? The project would already exist within Comfy. It doesn't, that's why it was created.
@vision181211772578 Also, the whole request is redundant, null and void. A question being asked while the answer already exists is not someone who's asking for help, it's someone who's asking not to do the work. It's asking someone else to do the work for them instead of looking what is available. What the person is requesting, namely SD pose or DW pose, has existed within Comfy for years. So the request is ridiculous because it's already existed and remains to exist. If they want that, they can do a Google search, use their brain and find out how it works. Asking someone to strip their custom made, extremely complex node pack, just so this person doesn't have to do a Google search is ridiculous.
Works good, thank you!
Is it worth altering the seed value at all? or setting it to random? or should we keep it as it is, fixed at the current value.
random seed can give small variations, especially for hidden parts of character that model should imagine.
Can you change the rotation cursor into a dragging cursor? I only have one rotation curstor right now :(
you can just drag limbs dots, not single click on them.
@AHEKOT Thank you :),is there a custom pose library community somewhere? Grok is bad xD
@Santaonholidays feel free to join my discord. There is not so much ppl yet who ready to contribute. Only one external library no longer external and shipped with node itself) You can be one who will made second!
@AHEKOT What is the link
@Santaonholidays it's right on github page :3 https://discord.com/invite/9Dacp4wvQw
@AHEKOT Yeah i got it
Hi, I hope you see this comment.
To start off, this node is truly fantastic and has exceptional features—it’s already amazing as it is!
I don't know if it's possible, but I’d like to suggest an "inverse function": adjusting the mannequin based on a photo—not so much its pose, but its proportions. Naturally, this would involve using photos that clearly show the proportions.
Let me explain a bit better:
It is already possible to set the proportions, but sometimes it’s quite complex! It would be great if they could be set automatically.
it already have ability to capture proportions, but it's a bit tricky to turn on and useless in main posing mode. (it's available in pose manager mode that can be turned on in setting). I will think how i can enable it for normal mode.
@AHEKOT I figured out how to import the pose from a photo on my own! 🤔
It downloaded something called "SAM 3D" that retrieves the pose for the mannequin and the camera; even though it isn't 100% precise with very complex poses, it’s a truly exceptional tool!!
Your implementation of the posing studio (and other packs) are on another level, that i haven't seen in many (if any) other nodepacks. So i salute the hard work and thank you for making it accessible for everyone <3
很强