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
In this version, I added a variable latent. I add "empty latent" This way, you will be able to set the desired resolution and aspect ratio. I updated the prompts, organized and simplified the nodes.
FAQ
Comments (44)
I am using the installable version of comfyui, and I am usable to install some nodes but still leaving to comment in here so I can be reminded to test it out later (maybe the nodes are not up-to-date)
Edit: I was able to replace some nodes but not the ToolYoloCropper, I just want to make sure: the result of that node (what should show in the "img2 crop" preview) is a zoomed/cropped image of what prompted? (depending on the model: "face", "person", "outfit"), I am trying to search for an alternative.
Yes, it works exactly as you said. Actually, you don't really need to use that node :) I wanted it to crop the head so the clothes wouldn't get mixed up. But when I find some really strong prompts, I don't know if it's really necessary.
@ugurdoyduk341 Thank you for answering! You're right, I actually find that using the source image without cropping gives me better results. Thanks again!
My only issue with the workflow is my heads up being much larger than expected for some reason
Sometimes it happens to me too. Set the ToolYoloCropper node to "person". Try this prompt. It works. ;
“Precisely fit the woman’s head, face and hairstyle from image2 inside the masked region of image1.
Strictly maintain natural head-to-body proportions; do NOT enlarge the head; do NOT exceed the mask borders.
Perfectly match the head shape to the mask shape.
Preserve ALL original facial features, identity, gaze, eye direction, expression and emotion of the woman in image2.
Match the lighting, colour grade, exposure, shadows and white balance of image1 exactly.
No glow, no bloom, no stylization.
Ultra-detailed.
Neutralize all colour casts.
Diffused soft lighting.
4K, clean, realistic, masterpiece quality.”
@ugurdoyduk341 Phenomenal stuff it works grat, skin tones a little off still but tremendous work, thats most likely due to my source image unsure if there's a fix for that
@smellyrs483 My latest discovery: in this situation, don't enter a prompt, just leave it blank.
Hello, thank you for the workflow. I have an 8 GB video card— can i run it with the Gguf model? https://huggingface.co/Phil2Sat/Qwen-Image-Edit-Rapid-AIO-GGUF/tree/main/v90
Hey, try q3_s first. If that doesn't work, try q2. Also with fp8 clip. I think it'll work. Don't forget to pull EasyCache to 45 or 55
I¡ma having problems with the ultralitycs, even after installing it. any tip?
Traceback (most recent call last):
File "/workspace/ComfyUI/nodes.py", line 2156, in load_custom_node
module_spec.loader.exec_module(module)
File "<frozen importlib._bootstrap_external>", line 940, in exec_module
File "<frozen importlib._bootstrap>", line 241, in callwith_frames_removed
File "/workspace/ComfyUI/custom_nodes/ComfyUI-Yolo-Cropper/__init__.py", line 1, in <module>
from .nodes import NODE_CLASS_MAPPINGS, NODE_DISPLAY_NAME_MAPPINGS
File "/workspace/ComfyUI/custom_nodes/ComfyUI-Yolo-Cropper/nodes.py", line 4, in <module>
from ultralytics import YOLO
ModuleNotFoundError: No module named 'ultralytics'
ComfyUI_windows_portable/ cmd window here and paste this
.\python_embeded\python.exe -m pip install ultralytics
@ugurdoyduk341 I'm running it in runpod
@elmejor9369; I honestly don't know what to do in this situation.
Have it working after a fashion. I had to bypass Yolo Cropper. Manager could not find the correct nodes. Not sure how or where you found it, but I had to swap it out with Auto Crop Faces node.
Got it. Hope you like the rest of it. I like these quirky nodes, but I can't remember where I found them :) The best thing about ToolYoloCroppe is that it automatically crops faces to create a lora. But since you found something to replace it, no problem.
@ugurdoyduk341 Thanks for the response I finely tracked it down a couple hours after the post, was going to delete it but you beat me to it.😊
@argentwulf0450 Got the link? :D
Hi,
First of all I thank you for sharing this workflow. Nevertheless, I think you need to modify your workflow to suit with latest ComfyUI (I am using desktop 0.7.2) and somehow the following nodes can not be installed successfully:
LoadImageWithFilename
Latent Switch (JPS)
ToolYoloCropperin subgraph 'ToolYoloCrop'
I read some of comments stated that ToolYoloCrop can not be installed into their ComfyUI as well. Hence, I think this is common issues that apply to user that has latest ComfyUI.
Thank you! I found alternatives for all of them. I can update them today.
V2 is online!
is it possible to swap vagina? especially hairy vagina.
That feature has been added to the new version.! Try V2. Read the prompt notes in the workflow.
Hahahahahaha. I think that in Sweden the police use labia instead of fingerprints. A regular LoRA will work just fine for this, in any workflow that has a detailer stage. Use Face Detailer node with a minge detector for your bbox. You can use segs too but really, since you are literally bounding a box... 0.4 to 0.6 is a good denoise range. Pretty much every image I make that shows a beaver does this. I don't know if the OP is talking about detailer nodes when he mentions it as a feature, or actually hotswapping boxes ala qwen face swapping. The latter would be hilarious. Just thought I would mention it; fixing body parts with detailers is a pretty basic technique that you should definitely work on if you are new to it.
@Ponder_Stibbons "Fixing body parts with detailers is a pretty basic technique," of course, but the title is obvious "swap anything" not the "fix anything". The person asking the question is clearly asking for an exchange (of a specific body part, an identified part). They want an exact replacement. Your method – which everyone would think of – doesn't solve that, does it? It only repairs that part. Thanks anyway.
@ugurdoyduk341
@ugurdoyduk341 Yeah, I wasn't trolling, I take your point. It's just in this particular circumstance, it struck me as really funny. Like, I could pick that twat out of I lineup funny, I'd know that slit anywhere... I've been through this gauntlet many times, so I was throwing in my two cents, which is that a locked-in refine stage with its own conditioning and LoRA is the perfect tool for the job, and can be dialed in to give the desired result consistently, so trying to tackle the problem with a swap is a little silly. In this pussticular slituation. Which I didn't mean to extend to anything about your workflow, which is great. But again, your point is taken, anything means anything. If the people want to do box swaps, let them do box swaps.
@Ponder_Stibbons I appreciate your help. Actually, it seemed funny to me at first too :) But I looked into it and tried it; Can wf do this? It can. What good will this do? I don't know. Maybe it's important for attentive OnlyFans followers, who knows :)
Keep getting "The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 2" Error on version 2.0
@SwapThatLad I got the same thing. Are you using an AIO model or regular
@nikolaibloom805 Im using the AIO model
@SwapThatLad Yeah i tried with both. Before I fixed it bypassing his sageattention node but I don't see that here now so I can't figure out what broke it this time. I am on the most recent version of comfyui so who knows
@SwapThatLad , @nikolaibloom805 I placed an ‘image to rgb’ node on the part coming out of Personmask. I would appreciate it if you could download it again and write the result. Positive or negative. Because I'm not getting any errors.
@ugurdoyduk341 Still getting the same error. plus if I disable the personmask node it works, but with funky results in the end
@SwapThatLad Can you share the full code starting from "got prompt"?
add one more "image to rgb" node between "image2" and "personmask" and let me know?
@ugurdoyduk341 Could you do an update on the workflow. So odd that it works for some people but not everybody
@SwapThatLad yeah without person mask mine is terrible like the equivalent of doing 4 steps without a lightning lora
@nikolaibloom805 updated
@ugurdoyduk341 thank you
Still getting that same error and I dont know why.
@nikolaibloom805 @SwapThatLad If you replace the image to rgb nodes with split image to alpha, would you get the same error? Can you try? And please also update all your nodes and comfy
@ugurdoyduk341 I am Still getting "DrawMaskOnImage
The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 2" error with this workflow. I tried replacing thoese nodes and updated comfyui and nodes. Not rly changed sth. I could update python dependencies but that could break my over workflows
@SwapThatLad hmm, if I could get the exact code, maybe I could pinpoint it, but I don't know. Hopefully an expert here can help us out.
@ugurdoyduk341 Ok, I finally got it working. I updated my whole comfy ui with the update all.bat
@SwapThatLad Really? Glad it worked :) Someone else got the same error in the Flux section, I'll tell them about the update too.