Having played with this a bit, I found I got better results, and much quicker, with grounding dino segment modules. A Threshold of 0.24 was good then I applied a gaussian blur mask, and that's taking about 60 seconds on a RTX2060 super (8GB).
This is great, thanks. Unfortunately, the diffusion still makes change to final out compared to the original one. For simpler tshirts or one color ones may be okay for demos but not ready for production quality. Hopefully in a few years...
I followed the steps to install everything, but I had to manually create a CatVTON folder. What do I need to do to access the segmentanything ultra v2 node?
Hi. Thank you for an amazing tutorial. I'm new to ComfyUI and this one. I am getting an error...or is it exactly an error? (idk) I keep getting black preview image after running the queue and can't find a way to fix it. Need help, please.....
very nice, its workflow i looking for ! just wanna ask one question. the human image should heve a simple background like white background etc? or can be various or more complex background? like indoor scenery, or outdoor landscape maybe
Hello. Dont know why it works for you. But models\onnx\human-parts\deeplabv3p-resnet50-human.onnx is missing. Where is that in that video. I seems to be neccessary for the Layer Mark node
Is there any way to quickly download all models? I rent GPU from Runpod so I don’t want to upload all models and create many subfolders again and again, take too much time.
Excuse me, I would like to ask if my computer with an Intel i5-8400 processor, GTX 1060 6GB graphics card, and 16 GB of RAM meets the system requirements for this program?
I followed this perfectly and got some satisfying results. Thanks.
It works onmy RTX 2060 (6GB) RAM, about 10 minutes a go. Had to create the models subfolder to that shown in this guide.
TAHNK YOU SO MUCH FOR THIS GUIDE AND THE DRIVE LINK ! ive been trying to make it work for hours!
This works wonderfully!
Thank you immensely!
awesome walkthrough. deserved more subs and likes
Well dane ! Working great :)
Having played with this a bit, I found I got better results, and much quicker, with grounding dino segment modules. A Threshold of 0.24 was good then I applied a gaussian blur mask, and that's taking about 60 seconds on a RTX2060 super (8GB).
i'm newbie on comfyui....Do u have workflow to share?
ohh nice, can you share your workflow
Can you share workflow file json
This is great, thanks. Unfortunately, the diffusion still makes change to final out compared to the original one. For simpler tshirts or one color ones may be okay for demos but not ready for production quality. Hopefully in a few years...
yup the texture n fabric details arent that good but the fit is great
I followed the steps to install everything, but I had to manually create a CatVTON folder. What do I need to do to access the segmentanything ultra v2 node?
Can this swap socks?
From your experience, how does this compare to IDM-VTON in terms of accuracy and image quality? Great vid btw!
2:08 Layer mask segment does not appear on the list
Thank you, nice tutorial, where do you get pictures of clothes only, all I can find are people wearing them.
Hi. Thank you for an amazing tutorial. I'm new to ComfyUI and this one. I am getting an error...or is it exactly an error? (idk) I keep getting black preview image after running the queue and can't find a way to fix it. Need help, please.....
very nice, its workflow i looking for !
just wanna ask one question. the human image should heve a simple background like white background etc? or can be various or more complex background? like indoor scenery, or outdoor landscape maybe
can you swap hair ?
will it work on mac?
missing vitmatte models
where can i get it?
Hello. Dont know why it works for you. But models\onnx\human-parts\deeplabv3p-resnet50-human.onnx is missing. Where is that in that video. I seems to be neccessary for the Layer Mark node
Is there any way to quickly download all models? I rent GPU from Runpod so I don’t want to upload all models and create many subfolders again and again, take too much time.
It is still running kiw in Vram even tho I have 8 Vram,,,
Hi! Does this work for jewelry or accessories?
That's it
Arent the requirements crazy; like you need an RTX 40 series to run it smoothly?
You need minimum 6gb vram
@@CgTopTips Nice!
Excuse me, I would like to ask if my computer with an Intel i5-8400 processor, GTX 1060 6GB graphics card, and 16 GB of RAM meets the system requirements for this program?
I don't know if it's late, but with the specs you mentioned, you still can, but generating will take you a very long time.
it doesn't work for me, it's all errors
感谢
Hi, any clue? # ComfyUI Error Report
## Error Details
- **Node Type:** CatVTONWrapper
- **Exception Type:** TypeError
- **Exception Message:** argument of type 'NoneType' is not iterable
## Stack Trace
Does it honor nonrealistic styles? Even if the clothes are realistic?
Sucks that commercial use is not allowed
who knows
AI has gotten really disturbing.