Home Creators Posts Import Register Favorites Logout

Downloads

Content

Edit 4 october: Added the manual size selection in the multi edit workflow, just don't forget to disable the pixel drift node if you enable the manual size option or else it won't workHey everyone! I've just released a brand new video on Qwen Image 2.1, and this model is THE BEST LOCAL IMAGE EDITING MODEL RIGHT NOW!

Combined with my ULTRA workflow Qwen image 2.1 will allow you to do MULTI-IMAGE EDITING, INPAINTING, OUTPAINTING, FULL CHARACTER REPLACEMENT, OBJECT REMOVAL AND MOVEMENT, CONTROLNET, BACKGROUND REMOVAL and MORE!

THIS MODEL IS JUST FANTASTIC!

You can check out the new video right here: https://youtu.be/5Sby8YxbhJc

I also created 1-click installers for Local AND RunPod for my Patreon supporters, which you can find right here: https://www.patreon.com/aitrepreneur/posts/171329215

Down below you will find both the normal ULTRA workflow and the TURBO workflow for you to use!

The TURBO version uses the special 6-step Viggle Turbo model, so you can choose between the normal setup and a faster version.

As always, supporting me on Patreon allows me to keep creating helpful resources like this for the community. Thank you for your support - now go have some fun😉!

Files

Comments

Matt

Does Qwen need NSFW Lora like Mystic XXX or SNOFS?

Nunca Nimble

I consistently get two errors when attempting the multi-edit workflow. 1) Rewrite prompt (Qwen3-VL text encoder) - "AttributeError: 'NoneType' object has no attribute 'tokenize'" 2) Required input slots have no connection feeding them: "VAE Decode - vae"

Michael Saldivar

same, works if i disable the prompt enhancer though :(

Nunca Nimble

Even then I still get the missing vae input error, but it's not obvious the way he set it up how it's "missing." The node in question clearly has a line going to the input on my screen.

Fantasy AI land

im running the turbo and the end result is horrendous am i doing something wrong , the end result is garbled and messy

Ken Kawashima

I tried installing this by just adding the models and dependencies, but the requirements destroy the environment. I tried all sorts of ways to reconcile the package incompatibilities to no avail - it breaks everything. However, a clean install in its own blank environment works flawlessly. I wish I could use it together with Qwen, Anima, etc, but was stopped by incompatible Python dependencies. Has anyone managed?

Mr. Friendly

Works, but the image unloader in the workflow always shrink when using it and need to reload everything to be able to see the full menu. Is there a fix for that?

Super Geek Maker

Great, I just downloaded it and am about to try it out. Just one thing: I can't find where to adjust the final image size. The reference images I used are vertical, but I want to create a group shot like in your example. Could you help me with that, please?

Aitrepreneur

You're right, I updated it, redownload the workflow from the post. Just don't forget to disable the pixel drift node if you enable the manual size option or else it won't work

Aitrepreneur

Super annoying bug and I don't understand where it's coming from...sometimes it does and sometimes it doesn't...comfyui has become buggy has hell...will try to find some fix once I have something

Aitrepreneur

you can, I had issues as well a few weeks ago, just give me the full error starting log you got, I'll give you a dependencies fixer file

Ken Kawashima

I also got TERRIBLE results from T2I, but the demo video shows this is kind of expected, since this model is not good for that. The real good side is the editing parts.

Ken Kawashima

Ok, use your BAT file to install only the models and nodes. That goes without errors. When I launch ComfyUI, it tells me one node requires upgrading ComfyUI to the latest version. I proceed updating ComfyUI, and when restarting, it stops with the following error (see below). At this point, it just quits and won't load anymore. M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\cuda\__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you. import pynvml # type: ignore[import] Traceback (most recent call last): File "M:\envs\ComfyUI2\ComfyUI_windows_portable\ComfyUI\main.py", line 249, in import comfy.utils File "M:\envs\ComfyUI2\ComfyUI_windows_portable\ComfyUI\comfy\utils.py", line 25, in import comfy.memory_management File "M:\envs\ComfyUI2\ComfyUI_windows_portable\ComfyUI\comfy\memory_management.py", line 8, in from comfy.quant_ops import QuantizedTensor File "M:\envs\ComfyUI2\ComfyUI_windows_portable\ComfyUI\comfy\quant_ops.py", line 8, in import comfy_kitchen as ck File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfy_kitchen\__init__.py", line 4, in from .backends import cuda as _cuda_backend File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfy_kitchen\backends\cuda\__init__.py", line 173, in from comfy_kitchen.backends.eager import rope as _eager_rope # noqa: E402 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfy_kitchen\backends\eager\__init__.py", line 73, in from .conv3d import fp16_conv3d, fp16_conv3d_out File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfy_kitchen\backends\eager\conv3d.py", line 37, in @torch.library.custom_op("comfy_kitchen::fp16_conv3d_out", mutates_args=("out",)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\_library\custom_ops.py", line 121, in inner schema_str = torch.library.infer_schema(fn, mutates_args=mutates_args) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\_library\infer_schema.py", line 106, in infer_schema error_fn( File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\_library\infer_schema.py", line 58, in error_fn raise ValueError( ValueError: infer_schema(func): Parameter stride has unsupported type list[int]. The valid types are: dict_keys([, typing.Optional[torch.Tensor], typing.Sequence[torch.Tensor], typing.List[torch.Tensor], typing.Sequence[typing.Optional[torch.Tensor]], typing.List[typing.Optional[torch.Tensor]], , typing.Optional[int], typing.Sequence[int], typing.List[int], typing.Optional[typing.Sequence[int]], typing.Optional[typing.List[int]], , typing.Optional[float], typing.Sequence[float], typing.List[float], typing.Optional[typing.Sequence[float]], typing.Optional[typing.List[float]], , typing.Optional[bool], typing.Sequence[bool], typing.List[bool], typing.Optional[typing.Sequence[bool]], typing.Optional[typing.List[bool]], , typing.Optional[str], typing.Union[int, float, bool], typing.Union[int, float, bool, NoneType], typing.Sequence[typing.Union[int, float, bool]], typing.List[typing.Union[int, float, bool]], , typing.Optional[torch.dtype], , typing.Optional[torch.device]]). Got func with signature (x: torch.Tensor, weight: torch.Tensor, bias: torch.Tensor | None, residual: torch.Tensor | None, stride: list[int], out: torch.Tensor) -> None)

Super Geek Maker

Thanks a lot for the update. I’ve run into a new problem now: I put back the same character images and a background as the first image, but when I run it, only the background image renders—not the characters. I have the same nodes disabled as the ones shown in your video. Could you please help me figure out what I'm doing wrong?.

Garrulous Skeptic

No change to initial image regardless of prompt (multi-edit). What am I doing wrong.

Aitrepreneur

Can you both dm me screenshots of your parameters?

Ken Kawashima

Where can I find the log file, and how can direct message you? Does that mean posting here?

Garrulous Skeptic

I got mine working without changing anything. Not sure what I did, other than a reboot. I had already updated (but I may have updated again). Hope you get it running - it really is a great w/f.

Super Geek Maker

Hi, I sent my screenshots to the chat; I hope you can help me.

Ken Kawashima

Please help me provide the info you have requested. :)

ExusOne

I downloaded the new workflow yesterday, and I get the impression that the skin realism for realistic character models has gone a bit... too far. In short, when I create Character Sheets for realistic characters, they look very old—as if at least 10 years had been added to their age, compounded by a heavy diet of cigarettes and junk food. Consequently, characters who appear to be around 20 in the original image end up looking 40 or older on the Character Sheet. More details I wrote in DM to avoid to promote different workflows on open channel.

Aitrepreneur

Yes because qwen image 2.1 isn't a good image gen model, the images it produces look too much "AI like" probably because of its training dataset, hence why I recommend using Krea 2 instead

ExusOne

Well i use it with multiedit node, not a text to image. I wrote now a message to you to show the examples.

Aitrepreneur

just send me a dm, for the log, right click on the cmd window at the top then click extract as text

Ken Kawashima

I tried again from scratch, and this time I don't get errors from the command line. Instead, it complains a missing node "QwenImage21Cache" was not installed and suggests I should run pip install -U --pre comfyui-manager, which I did. After restarting, it now complains my ComfyUI version (0.29.2) cannot run this workflow because the node requires 0.38.0:QwenImage21Cache. I understand the node cannot be installed because it requires ComfyUI v0.38.0 and I have v0.29.2. If I upgrade ComfyUI, all hell breaks loose because of requirements become incompatible with the environment. If I don't upgrade ComfyUI, I get no CLI errors but the QwenImage21Cache node won't install. Here is the errors I got from the beginning: **INITIAL ERROR: ================ Missing Node Packs Install missing packs to use this workflow. This workflow was created with a newer version of ComfyUI (0.29.2). Some nodes may not work correctly. Core nodes from version 0.38.0:QwenImage21Cache, TextEncodeQwenImage21 To install missing nodes, first run pip install -U --pre comfyui-manager in your Python environment to install Node Manager, then restart ComfyUI with the --enable-manager flag. **AFTER RUNNING pip install -U --pre comfyui-manager: ====================================================== Missing node: QwenImage21Cache This workflow uses a custom node that isn't installed. Install it from the registry or replace the node. This workflow was created with a newer version of ComfyUI (0.29.2). Some nodes may not work correctly. Core nodes from version 0.38.0:QwenImage21Cache, TextEncodeQwenImage21

Ken Kawashima

After updating ComfyUI to the latest version (Update ComfyUI), the CLI shows and error after restartting and stops there, unable to continue. Below is the part where it gets the error: M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\cuda\__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you. import pynvml # type: ignore[import] Traceback (most recent call last): File "M:\envs\ComfyUI2\ComfyUI_windows_portable\ComfyUI\main.py", line 252, in import comfy.utils File "M:\envs\ComfyUI2\ComfyUI_windows_portable\ComfyUI\comfy\utils.py", line 25, in import comfy.memory_management File "M:\envs\ComfyUI2\ComfyUI_windows_portable\ComfyUI\comfy\memory_management.py", line 8, in from comfy.quant_ops import QuantizedTensor File "M:\envs\ComfyUI2\ComfyUI_windows_portable\ComfyUI\comfy\quant_ops.py", line 8, in import comfy_kitchen as ck File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfy_kitchen\__init__.py", line 4, in from .backends import cuda as _cuda_backend File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfy_kitchen\backends\cuda\__init__.py", line 173, in from comfy_kitchen.backends.eager import rope as _eager_rope # noqa: E402 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfy_kitchen\backends\eager\__init__.py", line 73, in from .conv3d import fp16_conv3d, fp16_conv3d_out File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfy_kitchen\backends\eager\conv3d.py", line 37, in @torch.library.custom_op("comfy_kitchen::fp16_conv3d_out", mutates_args=("out",)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\_library\custom_ops.py", line 121, in inner schema_str = torch.library.infer_schema(fn, mutates_args=mutates_args) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\_library\infer_schema.py", line 106, in infer_schema error_fn( File "M:\envs\ComfyUI2\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\_library\infer_schema.py", line 58, in error_fn raise ValueError( ValueError: infer_schema(func): Parameter stride has unsupported type list[int]. The valid types are: dict_keys([, typing.Optional[torch.Tensor], typing.Sequence[torch.Tensor], typing.List[torch.Tensor], typing.Sequence[typing.Optional[torch.Tensor]], typing.List[typing.Optional[torch.Tensor]], , typing.Optional[int], typing.Sequence[int], typing.List[int], typing.Optional[typing.Sequence[int]], typing.Optional[typing.List[int]], , typing.Optional[float], typing.Sequence[float], typing.List[float], typing.Optional[typing.Sequence[float]], typing.Optional[typing.List[float]], , typing.Optional[bool], typing.Sequence[bool], typing.List[bool], typing.Optional[typing.Sequence[bool]], typing.Optional[typing.List[bool]], , typing.Optional[str], typing.Union[int, float, bool], typing.Union[int, float, bool, NoneType], typing.Sequence[typing.Union[int, float, bool]], typing.List[typing.Union[int, float, bool]], , typing.Optional[torch.dtype], , typing.Optional[torch.device]]). Got func with signature (x: torch.Tensor, weight: torch.Tensor, bias: torch.Tensor | None, residual: torch.Tensor | None, stride: list[int], out: torch.Tensor) -> None)

Eden Barlow

Great workflow by the way, I would like to know with this workflow can I face swap a group photo that has multiple faces with another face? If so how can I achieve this?

Aitrepreneur

try the special head swap lora maybe, can't promise it will work though

Aitrepreneur

Do not put your error log here, dm it to be with a full file

Ken Kawashima

The problem I seem to have is that one of the components requires the latest ComfyUI version, and if I upgrade, I get the dependency error shown above.

Super Geek Maker

I still haven't been able to resolve my problem, and I still can't generate any images.

peychi

How do I reset the output image node, without the node being removed? This happens when using multi edit workflow. I want to remove the output image as it has errors in the AI image, like an extra hand etc. I wish to try editing the same image again?