About Native Workflow; FIRST OF ALL UPDATE YOUR COMFYUI !!!
This has been the best node I've tried so far. In my tests, I saw that the 3.1 model made much better choices. Instead of person mask, I put the Sam3.1 text option where the prompt is entered. You can download the model from here and put it in the checkpoint folder; https://huggingface.co/Comfy-Org/sam3.1/tree/main/checkpoints
____________________________________________________________
About V3
I use a different node that performs the same function in place of the original Sam3 node. yolain/ComfyUI-Easy-Sam3.
Only python 3.12. If anyone runs this node on Python 3.13, can they share it with me?
Please upgrade your photobuf and mediapipe for "personmask" ( python_embeded\python.exe -m pip install --upgrade mediapipe protobuf google-cloud-videointelligence )
___________________________________________________________
About V2
I have added some spicy commands and their guide.. I removed the ToolYoloCropper node due to complaints. Instead, I added PersonMaskUltra V2 (Advance). I reverted to the defaults on some nodes. I simplified wf. Outfit swaps have become more consistent. But remember, when swapping outfits, you place the person whose outfit you want to copy into image2. So it's the exact opposite of what you do when swapping faces. Have fun!
_____________________________________________________________
About V1.2;
In this version,
I added a "PERSON SWAP". This is essentially a pose swap, and it works great.
I add 3 variable latent. (Two vae encode and an empty latent) This way, you will be able to set the desired resolution and aspect ratio. It's better "select 1" But what is it?
"Select 1" uses the aspect ratio of Img1 in the resulting output image.
"Select 2" uses the aspect ratio of Img2 in the resulting output image.
In this case, Select 2 is "tool yolo cropper."
"Select 3" Empty Latent Image. Opens a clean page. You determine the resolution.
In other words, you are switching between "VAE encodes and empty latent".
I updated the prompts, organized and simplified the nodes.
About Outfit Swap: Outfit swap is tricky. I admit that the normal regular qwen edit workflow changes clothes better. Just type new prompt that i create. I'm providing it below. Just enter this prompt in the regular qwen edit wf;
"Style the woman in image1, with every article of clothing in image2"
☕ Buy Me a Coffee:
🔗 https://buymeacoffee.com/ugurdoyduk
***Qwen 2511 2509 Swap Anyting (head, face, body, outfit, etc..) ***
In this workflow, you can easily swap anything you want with *100% CONSISTENCY** and without node complexity. The chance of facial mixing is ZERO..
Here, I utilized the advanced SAM3 model. You can easily swap faces, heads, just hair, bodies, or a part of an object. Simply type what you want to mask into the SAM3 Text Prompt panel. For example, "woman's head" or "man's hat." Everything you type in the prompt will be masked.
Just upload the relevant images to the corresponding "load image" section. You can use any checkpoint you like. I chose Phr00t and Qwen-Image-Edit-Rapid-AIO because they are very successful. (https://huggingface.co/Phr00t/Qwen-Image-Edit-Rapid-AIO). You can trust my Ksampler settings because I did a lot of testing.
If you want exactly the same facial expression, Use FP2 LORA and adjust 0.5, 0.6
https://huggingface.co/DiffSynth-Studio/Qwen-Image-Edit-F2P/tree/main
Description
I removed the ToolYoloCropper node due to complaints. Instead, I added PersonMaskUltra V2 (Advance). Outfit swaps have become more consistent. But remember, when swapping outfits, you place the person whose outfit you want to copy into image2. So it's the exact opposite of what you do when swapping faces. Have fun!
FAQ
Comments (19)
Hi i was using your workflow but what it came out is totally not within what was given, not too sure what went wrong though but i was using the gguf model for qwen edit 2511, i have a screenshot of the out put but i do not think i can upload the image here
Use the "personmask" carefully. Pay attention to which parts of face, hair, body, and clothes are onn or off
@ugurdoyduk341 mm the mask is correct but the image generated is way off from the base image and is blurry too
@leonviz the blury? What about 4steps,8steps lora things?
Fixed: "The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 2" Error on version 2.0
Updated my KJNodes and it resolved the error.
Thank you for the WF. Is it possible to just run the part to update the masks. I'm having a hard time to get Sam3 to mask the stuff I want it to mask and running the whole workflow everytime kinda sucks. Maybe add a possibility to manually refine the generated mask in a additional node?
Yes, this looks a bit annoying. Try to move all the "load img+ Sam loader+ Sam3 text+ Mask composite+ Draw mask on image+Preview " to a new page in comfy and try as much as you want. Second option that I can recommend: you can quickly end the task by pressing ctrl+alt+enter. I assigned this to one of my mouse buttons. It's so easy to Run-stop, run stop... If you create the preview latent on k-sampler, This will eliminate the problem for you. Instead of waiting for the image to fully load, you can watch it in the small preview section and immediately decide not to proceed. To do this, right-click on the run_nvidia_gpu.bat file, edit it in Notepad, and add the --preview-method latent2rgb argument on top, save and close. .
@ugurdoyduk341 Late reply :) First I did what you recommended and separated the Sam part to a new workflow try and error. Now I'm a bit older and wiser. Just right click on the preview Image node and chose the option to queue it. It will only run the part neccessary for the preview...
There appears to be something recently broken in Sam-3 downloader when you don't have the model. For me it gets an error after ~20 minutes:
LoadSAM3Model
Error while deserializing header: InvalidHeaderDeserialization
If I copy the safetensor version manually and reboot Comfy, I get an error that's the same as the description of the issue here: https://github.com/PozzettiAndrea/ComfyUI-SAM3/issues/103
Now, I know the WF person here isn't responsible for other nodes/models but thought I'd bring it up if anyone else has the issue - they aren't alone.
Thank you for reminding me. We've been waiting for a week. Pozetti/sam3 said they would fix it, but it hasn't happened. The new version is problematic. Personally, I'm sticking with the old version. In case you don't have it, I sent you the custom nodes folder from the old version as a chat-message.
or, custom nodes/sam3/ open terminal git checkout 18efc9b
Previously I had used head swap several times and the results were very good. However, after updating ComfyUI, I can no longer use it. An error message appears:
LoadSAM3Model: 'NoneType' object has no attribute 'data'
Even though before the update everything worked fine
Try the previous version; the new version of sam3 is problematic. custom nodes/sam3/ open terminal, type git checkout 18efc9b
Thank you so much. A wonderful tool for preparing keyframes for generating video and photo contents.
Where can I download the sam3 model? It wont auto download for me
or for the working old versiyon custom nodes/sam3/ open terminal type; git checkout 18efc9b
Whenever I try to replace a head, it always ends up being too big. And the head feels like it belongs to someone else, the skin is a different color. The quality of the result is very poor, unfortunately.
Please Read the personmask notes in the workflow.