[EXCLUSIVE GUIDE] CONSISTENT CHARACTER CREATOR v01 + PULID

Hi to the author of these interesting flows))) I like your creative approach. But to my regret I have a problem with 4 nodes in the flow 250126, they are not installed Import failed ( ComfyUI Pulid FLUX II, Pulid-Flux- Enhanced, Pulid Flux II FaceNet) I tried installing these nodes through the manager, then manually from GitHub, and nothing works. What should I do? Any suggestions on how to fix this?

Roman Protsyk

Hello Mickmumpitz, thank you so much for sharing this. I really appreciate it. I just ran into a small issue: everything works perfectly in the multi-view tab, but it doesn’t move on to the upscale 1st pass step, so I’m unable to continue the workflow.

Burak

I am getting stuck with the DownloadAndLoadFlorence2Model node with this error: 'DaViT' object has no attribute '_initialize_weights'. Anyone has an idea?

jahitian

Can't import this workflow due to: - **Exception Message:** TypeError: Cannot read properties of undefined (reading '0') Any fix or reason as to why?

Arash

Getting following error "can't multiply sequence by non-int of type 'numpy.float32'" when the 04:Emotinos section start executing. Could you guide me on this?

g balaji

This workflow works for me.. It took some fiddling to make sure everything was downloaded and installed correctly model/difuser/etc -wise. I also needed to run it on an RTX6000. Its a beast. And meant as an example. Nobody sensible would use this for real production. Beside , there are more efficient ways to get great elements for a Char sheet and just photoshop it together,,,,,,

AirborneSapper708 RoughTerrain12B

I had similar problems but at least I got it running. But wasnt impressed by the results. If you only need this workflow to generate some images as base for a training you can use Ace++ as an alternative. I created a bunch of different positions of a person - I only need the images for the pose. The face at this point is nearly not important. And with a Ace++ Faceswap then you can swap the face against a face of your wanted person. It works perfectly, Sebastian Kamph made a video about it. I used the swapped images for training. You even can make the person wear the same clothes (I made a medieval knight, so I needed the same armor again...) with a Subject or migraton LoRa made by Alimama. Maybe it helps... Currently I'm using this workflow and a custom node I programmed to batch replace faces and clothes in single frames of a video from Kling or Hunyuan. Sometimes the faces got broken in the process and I use this now to fix the broken frames.

Christian Schülling

i would stay a member forever, but its just way to hard to get it to work. most dependencies dont get installed, it took me days to troubleshoot, but progress is slow, so im abandoning ship. i hope one day you can truly make a download that works 1 click, idc if its a 1000gb file.

Jeremia

does this work for mac (Apple M2 Ultra)

Alex Atkinson

I finally gotten the pre-requisites installed. the whl file link that was provided in the instructions seemed to be the issue. Based on ChatGPT, whl files should be binary files, the one provided in the instructions was a text file. I had to locate a working version which was a over 800kb in size: https://huggingface.co/hanamizuki-ai/insightface-releases/tree/main I also had to install VS C++ compiler, and some other VS components to finally get the insightface-0.7.3-cp312-cp312-win_amd64.whl to install correctly. Also re-installed MV-Adapter custom node; however, I'm now getting this issue whenever I queue up a generation: LdmPipelineLoader Pipeline expected {'scheduler', 'text_encoder', 'tokenizer_2', 'image_encoder', 'text_encoder_2', 'tokenizer', 'feature_extractor', 'unet', 'vae'}, but only {'scheduler', 'text_encoder', 'tokenizer_2', 'text_encoder_2', 'tokenizer', 'vae'} were passed. Any support would be greatly appreciated; otherwise, I don't know why I'm supporting this page when I can't get this workflow and a number of others from this site to work. Mick, if you can see this please help out here?

Jeff h

This is very strange, paid subscribers here. Please write to him that we are upset.

Sergei Markevich

For anyone that is reading this: He DOES answers, but he prefers discord. He literally fixed my problems in a few minutes when i asked him on discord... So hop on discord if you have issues or questions!

Selina Druifmans

Hi! Please help me. When it comes to the stage of generating a block with emotions, I get a process stop and an error in the console: Failed to validate prompt for output 369: * UltimateSDUpscale 349: - Required input is missing: model - Required input is missing: negative - Required input is missing: cfg - Required input is missing: steps Output will be ignored Failed to validate prompt for output 756: * ExpressionEditor 1340: - Failed to convert an input value to a FLOAT value: crop_factor, OnlyExpression, could not convert string to float: 'OnlyExpression' Output will be ignored Failed to validate prompt for output 754: Output will be ignored Failed to validate prompt for output 1340: * (prompt): - Failed to convert an input value to a FLOAT value: crop_factor, OnlyExpression, could not convert string to float: 'OnlyExpression' Output will be ignored Prompt executed in 0.30 seconds

Sergei Markevich

If I may, I would like to offer a suggestion based on my own experience as a novice. It can be challenging to identify the most suitable versions to facilitate workflow and scripts. We have several manuals, including v01 [Exclusive guide] which is more recent than v02 [Free guide], and different workflows in different locations. It would be very helpful to have a centralised location for Patreon members to access all stable content. Due to the numerous dependencies within ComfyUI, it can be difficult to keep track of everything.

Jürgen & Brita Stein

probably using to much memory, have you tried diferent checkpoints in comfy?

Frontline Fury

sure no problem, but what do you mean by disconnect? It just crashes?

Mick Mahler

thank you, thank you, thank you! 🙏🏻 your instructions and screenshot saved my day.

Omri Sadeh

Hahaha, so fun that i not read comments before pay)

Иван Меер

can i get a refund? i've tried all the flows and they disconnect halfway through...waste of time

Alex Atkinson

Also when running this ComyUI keeps disconnecting about halfway through; and it just says in a red box top right "Reconnecting"

Alex Atkinson

Also when i load the .json file it says "The workflow you've loaded has corrupt linking data that may be able to be fixed." in the top of ComfyUI

Alex Atkinson

your workflows are gereat in theory. None of them ever work. Days and Weeks spent on troubleshooting and asking people. No one can get it to work in reality.

Raffael Dickreuter

Every one of these consistent character workflows does not work for me. I get stuck on the install of the insightface addins. Is there anyone who can help me get over this stumbling block? Mick??? Would really appreciate the help… I’m using the portable install of ComfyUI. I’ve followed the documentation step by step. Are there any other troubleshooting steps that someone could post here? I am a paying Patreon supporter Mick. Any help you can provide would be greatly appreciated.

Jeff h

to fix the error: "received_type(MAKED_PIPELINE) mismatch input_type(PIPELINE)" you need to do the following: 1. Remove mv-adapter via manager 2. Install mv-adapter via git clone in custom-nodes folder 3. Install all dependencies inside custom_nodes\ComfyUI-MVAdapter folder (pip install -r requirements.txt) 4. Run ComfyUI 5. Go to section with multiview 6. Replace modules: Vae Loader, Diffusers Scheduler Loader, and Diffusers Model Makeup with modules from mv-adapter library should be as follows: https://drive.google.com/file/d/1d3gu1j4Ct93XkCY2dKZnL1juwkbacPxm

Psn

Thank you. I got me on the right track, but now I'm stuck on this error "UNetModel' object has no attribute 'forward_orig'".

Epicture

Same problem with insighface here. Please help.

SodaFilmBern

ok...i give up...after spending hours to get this workflow working i think that its really not worth the time. It's not working at all...also another workflow from this guy...It's not worth to support that guy.

Funny Dog Challenges

Notthe newest one, I have his others workiing . But yes. Just Pulid

AirborneSapper708 RoughTerrain12B

?

Goqor Grigoryan

I have the same issue, did you manage to fix this? I managed to get support in the discord. Especially from the pinned posts in the get help section

creat0r

Has anyone else gotten stuck with the actual install of insighface? I did a brand new install and it’s still the same error. I’m using the portable version.

Jeff h

im getting clip errors.. is mickmumpitz going to answer any of these posts?

alejandro paredes

Hi there, I'm still having problems installing C:\ai\App\ComfyUI_P3.12.8\ComfyUI_windows_portable\python_embeded>python.exe -m pip install insightface-0.7.3-cp312-cp312-win_amd64.whl onnxruntime Processing c:\ai\app\comfyui_p3.12.8\comfyui_windows_portable\python_embeded\insightface-0.7.3-cp312-cp312-win_amd64.whl ERROR: Wheel 'insightface' located at C:\ai\App\ComfyUI_P3.12.8\ComfyUI_windows_portable\python_embeded\insightface-0.7.3-cp312-cp312-win_amd64.whl is invalid. This is also from a new install using the latest portable install of ComfyUI. Any help would be greatly appreciated. Thanks

Jeff h

it is because somewhere is used sd1.5, somewhere SDXL, somewhere FLUX.

Goqor Grigoryan

Yes...you only need a minimum of 10 images to get started with FluxGym. You can also use www.runpod.io and use a pre-existing template (i.e. Local Lab FluxGym Lora Training with Kohya ss) to generate your LoRA quickly

Paul H

I thought I had gotten rid of all the errors, but there's still one more! 😆 Anyone know how to resolve this? 'SamplerCustomAdvanced: mat1 and mat2 shapes cannot be multiplied (4096x16 and 64x3072).

Epicture

bro if i ask you, will you send me you modified version?

Goqor Grigoryan

have you used this one? 250126_MICKMUMPITZ_CCC_v01_FLUX_ADV+PULID.json If so, did you changed only pulid?

Goqor Grigoryan

Can the images generated by this method be sufficient to create flux lora?

Kang Hua

Same happened with me and I figured out that I have two custom model that conflict each other, DDUF and comfyui-mvadaptar. For sure you are using the DDUF instead which came with MAKED_PIPELINE option which can't be connected with pipeline. You have to uninstall the DDUF and the rest will works fine.

Fabio Henrique Santos

Finally got this working. Pulid were out of date and were causing failures ,,,,everything works now . I went back and read all notes as well…including how to work thru the errors. Very tedious …however it is a gigantic template lol… actual ran pretty fast . Im interested in controlling this beast now…I keep getting variations to clothing . 😊✌️

AirborneSapper708 RoughTerrain12B

This workflow is very confused and a lot of things are not clear.. I get the error : Return type mismatch between linked nodes: pipeline, received_type(MAKED_PIPELINE) mismatch input_type(PIPELINE) ... and so i don't have the different model orientatation and just have two faces upscaled.... weird

Marie-Anne Schmitz

so I do know that Pulid released new xformers 2-3 days ago to deal w a longstanding problem...I am trying this on my cloud account thru InstaSD/// Ill report back on success or roadblocks...

AirborneSapper708 RoughTerrain12B

What are the minimum video card specs (what series of RTX and how much VRAM do I need) to achieve this workflow? I have an RTX3070 with 8GB of VRAM, so is that too much to ask? Also, please let me know if this workflow is feasible on an M1 or higher MacBook.

Ryutaro Osafune

I have already paid $10 for it. But my video card specs seem low (RTX3070 with 8GB of VRAM).

Ryutaro Osafune

AirborneSapper708 RoughTerrain12B

Fixed, I had an unnecessarily nested folder structure (antelope_v2\antelope_v2\...)

Joseph Mama

Hey, I'm getting: ...\ComfyUI_windows_portable_nvidia\ComfyUI_windows_portable\python_embeded\Lib\site-packages\insightface\app\face_analysis.py", line 43, in __init__ assert 'detection' in self.models Do I have to add a detection model to \ComfyUI_windows_portable\ComfyUI\models\insightface\models

Joseph Mama

Wieder mal keine Antwort von dem drecking hurengaywizzard

McLovin

I have problems with all the templates I've downloaded, most of the missing nodes refuse to be imported by 'some' reason. The lack of support here is really disappointing, Mickmumpitz.

Asyouwere

i resized latent image before samplecustomadvanced to 1920wide. now its work, but i have some artifacts. not great not terrible

Aleksandr Zhelezniakov

I have track this down to one misbehaving node, ComfyUI-MVAdapter is not unloading its model after generation, and clear vram doesn't work. I patched the node and have gotten it to work but its not a good patch :) - There is an open issue on their github.

v33ts

Has anyone else Problems installing the MVA Adapter and some included custom Nodes even after doing all the troubleshooting steps??? I even checked all the versions hugging face hub etc. Still doesnt work

McLovin

I have a 4090 and I can't get this to work (flux version) with out OOM like crazy :( Over the weekend I will break it down and rebuild it and see if I can make it work.

v33ts

Is there any link or instruction how to install it to Google cloud?

Maksym Evil

that's because you're a master and i'm a noob

FB6 CORP

so easy.

Hao Chen

Looking forward to the Lora

Hao Chen

I joined the Patreon before Christmas, but honestly had no time to work on this. I can incorporate this into my day job however. Can I ask, pls, is there any opinion on how OpenArt.ai stacks up as a platform? Forgive the beginner question - not had time to do the work on this all yet (to any depth).

Ken McHardie

he just waits, unitl you pay one more $10, then will update. Change controlnet, checkpoints model for SD1.5 and it will work. But this guy does it in purpose

Goqor Grigoryan

I've used the SDXL without issues, which sounds like what Mick is recommending - I have 16Gb of VRAM...but FLUX just soaks up too much compute/ram

Paul H

Any chance you could just make a Runpod template with a provisioning script that automatically sets everything up? Would save a lot of headache for you and your community. If you need help with it just hit me up

Kevin Sozanski

Honestly I succeeded with the free workflow. But this one no, I'm no more getting error of missing staff but when I start the workflow is stuck. so I think the issue is my pc hasn't enough VRAM to run this workflow

FB6 CORP

Hi. The new node is highlighted in yellow and is present in the workflow. This post contains an updated version of workflow. Download it. search 250126_MICKMUMPITZ_CCC_v01_FLUX_ADV+PULID.json on this page

yarikmoscow

I use https://openposeai.com/

Gaurav Misra

By the way, has anyone successfully set up this environment and is now happily using this workflow? I'm worried because everyone I've seen in this thread has failed to set up the environment. I am going to look at this document and build the environment, but like others, I don't want to waste a lot of time and end up failing to build the environment..

Ryutaro Osafune

Try to disable the integrated gpu if you have one

Nino Dorić

I have tried almost anything past 4 days. I can tell you this. On RunPod, if the Pytorch CUDA setup does not match the requirements of the comfyui nodes for this workflow, changing versions on a python or conda virtual environment won't protect you, it will corrupt the instance environment and cause dependency conflicts. One last thing I have yet to try is doing my setup on docker or creating a template with docker bazel for this workflow that i can publish and reuse in the future. I am very tired and frustrated at the moment though. I went on a walk to clear my mind but still frustrated how much time I wasted just to understand these limitations and it might not be a worthy investment of money and effort if the docker attempt fails.

Chai Biscuits

Joe, I have tried almost anything past 4 days. I can tell you this. On RunPod, if the Pytorch CUDA setup does not match the requirements of the comfyui nodes for this workflow, changing versions on a python or conda virtual environment won't protect you, it will corrupt the instance environment and cause dependency conflicts. One last thing I have yet to try is doing my setup on docker or creating a template with docker bazel for this workflow that i can publish and reuse in the future. I am very tired and frustrated at the moment though. I went on a walk to clear my mind but still frustrated how much time I wasted just to understand these limitations on runpod. Wish I had more savings to uprgrade to a more powerful machine.

Chai Biscuits

Thanks!

Epicture

Hey, does anyone know how to solve this problem? (ApplyPulidFlux Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument mat1 in method wrapper_CUDA_addmm))

W K

hey I got this error and I resolved it by uninstall the custom node and install it again. and it will create all the components.

FB6 CORP

Hey all. Absolute beginner here. It seems that I managed to install everything but one thing. ComfyUI_UltimateSDUpscale. I got this error: Traceback (most recent call last): File "C:\AI\ComfyUI_windows_portable\ComfyUI\nodes.py", line 2110, in load_custom_node module_spec.loader.exec_module(module) File "", line 999, in exec_module File "", line 488, in _call_with_frames_removed File "C:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_ultimatesdupscale\__init__.py", line 32, in from .nodes import NODE_CLASS_MAPPINGS, NODE_DISPLAY_NAME_MAPPINGS File "C:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_ultimatesdupscale\nodes.py", line 6, in from usdu_patch import usdu File "C:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_ultimatesdupscale\usdu_patch.py", line 2, in from repositories import ultimate_upscale as usdu File "C:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_ultimatesdupscale\repositories\__init__.py", line 14, in spec.loader.exec_module(ultimate_upscale) File "", line 995, in exec_module File "", line 1132, in get_code File "", line 1190, in get_data FileNotFoundError: [Errno 2] No such file or directory: 'C:\\AI\\ComfyUI_windows_portable\\ComfyUI\\custom_nodes\\comfyui_ultimatesdupscale\\repositories\\ultimate_sd_upscale\\scripts\\ultimate-upscale.py' Anybody else got this problem? Is there a workaround? Download and add it manually?

Epicture

Not sure exactly how to fix that but did you upgrade your NVIDIA drivers lately? Also, some google searching says torch might not be properly installed... I'm no expert here but those seem like something easy to try and see if it works.

JoeBlow

Yeah this workflow is still messed up in Comfy for me. I have a few errors that I was able to correct with downloading the right C++ VS tools, pip dependency correction (i.e. uninstall and re-install of lower versions of tools) what python version do you recommend we run this on? Also, this could just be a windows problem... I.E. using windows to begin with lol. Anyway, cool workflow but I can't use it currently. UPDATE: I got through the sea of red, I guess I posted this too soon. A couple things to note for others on Windows installs: for issues with diffusers not installed warning on startup from the HF download maintainer (https://github.com/easydiffusion/easydiffusion/issues/1851#issuecomment-2592098062) this fixed that issue for me and allowed me to get the easydiffusion stuff installed. Second, for issues building the wheel and errors about "error: Microsoft Visual C++ 14.0 or greater is required. Get it with "Microsoft C++ Build Tools": https://visualstudio.microsoft.com/visual-cpp-build-tools/ " see this thread and select the right tools to download from the reccomendation here:https://stackoverflow.com/a/64262038

JoeBlow

ANYONE HELP? My ass been glued to this chair for 2.5 days. I am using RunPod for this. Stuck. Here's what I told gpt to help me but still stuck. I am using RunPod to complete a ComfyUI workflow. I have installed the latest ComfyUI version with Manager and imported the pre-made workflow. The workflow had some missing nodes and I installed most of them, leaving the conflicting custom nodes at the end. This is the pytorch and cuda versions that been recommended to me: (comfyui-venv) root@7385c287f1e3:/workspace/ComfyUI# python -c "import torch; print(torch.version, torch.cuda.is_available())" 2.5.1+cu121 True. PyTorch version 2.5.1+cu121 with CUDA 12.1 aligns with the latest ComfyUI requirements. Okay. Let's get to the heart of the issues in this workflow. I got the following 'missing' or 'failed import' custom nodes. I have installed them before, but the comfyUI workflow was failing with this error: # ComfyUI Error Report Error Details Node ID: 1369 Node Type: LdmPipelineLoader Exception Type: AttributeError Exception Message: module 'diffusers' has no attribute 'EulerDiscreteScheduler' Stack Trace File "/workspace/ComfyUI/execution.py", line 327, in execute output_data, output_ui, has_subgraph = get_output_data(obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb) File "/workspace/ComfyUI/execution.py", line 202, in get_output_data return_values = _map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb) File "/workspace/ComfyUI/execution.py", line 174, in _map_node_over_list process_inputs(input_dict, i) File "/workspace/ComfyUI/execution.py", line 163, in process_inputs results.append(getattr(obj, func)(**inputs)) File "/workspace/ComfyUI/custom_nodes/comfyui-mvadapter/nodes.py", line 97, in create_pipeline pipe = pipeline_class.from_single_file( File "/workspace/new-comfyui-env/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py", line 114, in _inner_fn return fn(*args, **kwargs) File "/workspace/new-comfyui-env/lib/python3.10/site-packages/diffusers/loaders/single_file.py", line 495, in from_single_file loaded_sub_model = load_single_file_sub_model( File "/workspace/new-comfyui-env/lib/python3.10/site-packages/diffusers/loaders/single_file.py", line 71, in load_single_file_sub_model class_obj = getattr(library, class_name). My intuition so far is that these conflicting nodes: ComfyUI-Diffusers, ComfyUI-MVAdapters and ComfyUI-DDUF use conflicting versions of diffusers, some of which lack the necessary schedulers. I trieed resolve this before by upgrading diffusers but the pytorch, CUDA and other dependencies were conflicting the causing this runtime error: new-comfyui-env) root@fc357acafc11:/workspace/ComfyUI# python main.py --listen --port 8188 [START] Security scan [DONE] Security scan ComfyUI-Manager: installing dependencies done. ** ComfyUI startup time: 2025-01-27 18:04:36.943 ** Platform: Linux ** Python version: 3.10.12 (main, Nov 20 2023, 15:14:05) [GCC 11.4.0] ** Python executable: /workspace/new-comfyui-env/bin/python ** ComfyUI Path: /workspace/ComfyUI ** User directory: /workspace/ComfyUI/user ** ComfyUI-Manager config path: /workspace/ComfyUI/user/default/ComfyUI-Manager/config.ini ** Log path: /workspace/ComfyUI/user/comfyui.log Prestartup times for custom nodes: 0.0 seconds: /workspace/ComfyUI/custom_nodes/comfyui-easy-use 0.0 seconds: /workspace/ComfyUI/custom_nodes/rgthree-comfy 12.3 seconds: /workspace/ComfyUI/custom_nodes/ComfyUI-Manager Checkpoint files will always be loaded safely. Traceback (most recent call last): File "/workspace/ComfyUI/main.py", line 136, in import execution File "/workspace/ComfyUI/execution.py", line 13, in import nodes File "/workspace/ComfyUI/nodes.py", line 22, in import comfy.diffusers_load File "/workspace/ComfyUI/comfy/diffusers_load.py", line 3, in import comfy.sd File "/workspace/ComfyUI/comfy/sd.py", line 6, in from comfy import model_management File "/workspace/ComfyUI/comfy/model_management.py", line 166, in total_vram = get_total_memory(get_torch_device()) / (1024 * 1024) File "/workspace/ComfyUI/comfy/model_management.py", line 129, in get_torch_device return torch.device(torch.cuda.current_device()) File "/workspace/new-comfyui-env/lib/python3.10/site-packages/torch/cuda/init.py", line 940, in current_device _lazy_init() File "/workspace/new-comfyui-env/lib/python3.10/site-packages/torch/cuda/init.py", line 319, in _lazy_init torch._C._cuda_init() RuntimeError: Unexpected error from cudaGetDeviceCount(). Did you run some cuda functions before calling NumCudaDevices() that might have already set an error? Error 804: forward compatibility was attempted on non supported HW (new-comfyui-env) root@fc357acafc11:/workspace/ComfyUI# . I suggest we carefully look at the installation requirements of each conflicting node and you help me create the perfect requirements list to install that would not cause any issues on my environment.. I will first install the nodes that show no conflict warning in my manager. Okay? Got the current dependencies, and the requirements of each conflicting custom node from my ComfyUI Manager (latest version). Remember, the goal is to compile a list of requirements that satisfies the needs of each node, is compatible with my current environment (pytorch and CUDA versions), and to install the best compatible version of diffusers which has all the needed schedulers. Let's avoid this error this error: # ComfyUI Error Report Error Details Node ID: 1369 Node Type: LdmPipelineLoader Exception Type: AttributeError Exception Message: module 'diffusers' has no attribute 'EulerDiscreteScheduler'. /////////current_dependencies.txt: accelerate==1.3.0 aiohappyeyeballs==2.4.4 aiohttp==3.11.11 aiosignal==1.3.2 albucore==0.0.23 albumentations==2.0.2 annotated-types==0.7.0 antlr4-python3-runtime==4.9.3 anyio==4.8.0 async-timeout==5.0.1 attrs==25.1.0 boto3==1.34.86 botocore==1.34.162 certifi==2024.12.14 cffi==1.17.1 charset-normalizer==3.4.1 click==8.1.8 clip-interrogator==0.6.0 cmake==3.31.4 color-matcher==0.5.0 coloredlogs==15.0.1 colour-science==0.4.6 contourpy==1.3.1 cryptography==44.0.0 cstr @ git+https://github.com/WASasquatch/cstr@0520c29a18a7a869a6e5983861d6f7a4c86f8e9b cycler==0.12.1 Cython==3.0.11 ddt==1.7.2 Deprecated==1.2.18 diffusers==0.32.2 dill==0.3.9 docstring_parser==0.16 docutils==0.21.2 easydict==1.13 einops==0.8.0 exceptiongroup==1.2.2 facexlib==0.3.0 fairscale==0.4.13 fal_client==0.5.8 ffmpy @ git+https://github.com/WASasquatch/ffmpy.git@f000737698b387ffaeab7cd871b0e9185811230d filelock==3.17.0 filterpy==1.4.5 flatbuffers==25.1.24 fonttools==4.55.7 frozenlist==1.5.0 fsspec==2024.12.0 ftfy==6.3.1 gitdb==4.0.12 GitPython==3.1.44 h11==0.14.0 httpcore==1.0.7 httpx==0.28.1 httpx-sse==0.4.0 huggingface-hub==0.28.0 humanfriendly==10.0 idna==3.10 imageio==2.37.0 imageio-ffmpeg==0.6.0 img2texture @ git+https://github.com/WASasquatch/img2texture.git@d6159abea44a0b2cf77454d3d46962c8b21eb9d3 importlib_metadata==8.6.1 insightface==0.7.3 Jinja2==3.1.4 jmespath==1.0.1 joblib==1.4.2 jsonschema==4.23.0 jsonschema-specifications==2024.10.1 kiwisolver==1.4.8 kornia==0.8.0 kornia_rs==0.1.8 lark==1.2.2 lazy_loader==0.4 llvmlite==0.44.0 lmdb==1.6.2 markdown-it-py==3.0.0 MarkupSafe==2.1.5 matplotlib==3.10.0 matrix-client==0.4.0 mdurl==0.1.2 mpmath==1.3.0 mss==10.0.0 multidict==6.1.0 networkx==3.3 numba==0.61.0 numpy==1.26.4 nvidia-cublas-cu12==12.1.3.1 nvidia-cuda-cupti-cu12==12.1.105 nvidia-cuda-nvrtc-cu12==12.1.105 nvidia-cuda-runtime-cu12==12.1.105 nvidia-cudnn-cu12==9.1.0.70 nvidia-cufft-cu12==11.0.2.54 nvidia-curand-cu12==10.3.2.106 nvidia-cusolver-cu12==11.4.5.107 nvidia-cusparse-cu12==12.1.0.106 nvidia-nccl-cu12==2.21.5 nvidia-nvjitlink-cu12==12.1.105 nvidia-nvtx-cu12==12.1.105 omegaconf==2.3.0 onnx==1.17.0 onnxruntime==1.20.1 onnxruntime-gpu==1.20.1 open_clip_torch==2.30.0 opencv-contrib-python==4.11.0.86 opencv-contrib-python-headless==4.11.0.86 opencv-python==4.11.0.86 opencv-python-headless==4.11.0.86 packaging==24.2 pandas==2.2.3 peft==0.14.0 piexif==1.1.3 pilgram==1.2.1 pillow==11.0.0 pixeloe==0.1.1 platformdirs==4.3.6 pooch==1.8.2 prettytable==3.13.0 propcache==0.2.1 protobuf==5.29.3 psutil==6.1.1 py-cpuinfo==9.0.0 pycparser==2.22 pydantic==2.10.6 pydantic_core==2.27.2 PyGithub==2.5.0 Pygments==2.19.1 PyJWT==2.10.1 PyMatting==1.1.13 PyNaCl==1.5.0 pyparsing==3.2.1 python-dateutil==2.9.0.post0 pytz==2024.2 PyYAML==6.0.2 redis==5.2.1 referencing==0.36.2 regex==2024.11.6 rembg==2.0.61 replicate==1.0.4 requests==2.32.3 rich==13.9.4 rpds-py==0.22.3 s3transfer==0.10.4 safetensors==0.5.2 scikit-image==0.25.1 scikit-learn==1.6.1 scipy==1.15.1 seaborn==0.13.2 sentencepiece==0.2.0 shellingham==1.5.4 shtab==1.7.1 simsimd==6.2.1 six==1.17.0 smmap==5.0.2 sniffio==1.3.1 soundfile==0.13.1 spandrel==0.4.1 stringzilla==3.11.3 surrealist==1.0.8 sympy==1.13.1 threadpoolctl==3.5.0 tifffile==2025.1.10 timm==1.0.14 tokenizers==0.21.0 toml==0.10.2 torch==2.5.1+cu121 torchaudio==2.5.1+cu121 torchsde==0.2.6 torchvision==0.20.1+cu121 tqdm==4.67.1 trampoline==0.1.2 transformers==4.48.1 triton==3.1.0 typer==0.15.1 typing_extensions==4.12.2 tyro==0.8.5 tzdata==2025.1 ultralytics==8.3.40 ultralytics-thop==2.0.14 urllib3==1.26.20 wcwidth==0.2.13 websocket-client==1.8.0 wrapt==1.17.2 yarl==1.18.3 zipp==3.21.0 ///////ComfyUI-Impact-Pack/requirements.txt: segment-anything scikit-image piexif transformers opencv-python-headless scipy>=1.11.4 numpy<2 dill matplotlib //////ComfyUI-Diffusers/requirements.txt: diffusers[torch] accelerate transformers safetensors omegaconf pytorch_lightning xformers git+https://github.com/cumulo-autumn/StreamDiffusion.git@main#egg=streamdiffusion[tensorrt] ////////ComfyUI-MVAdapter/requirements.txt: torch>=2.1.1 torchvision>=0.16.1 diffusers==0.31.0 transformers==4.46.3 peft numpy>=1.26.2 huggingface_hub==0.24.6 accelerate==1.1.1 opencv-python safetensors pillow omegaconf trimesh einops timm kornia scikit-image ////////ComfyUI-DDUF/requirements.txt: diffusers[torch] accelerate transformers safetensors omegaconf

Chai Biscuits

Same, the struggle is real

Brandon J

is anyone able to make this work ? I installed all needed requirements but it keeps raining blood with error message.

FB6 CORP

Random Qu: does any have a "pose sheet" for a dog or cat etc - and any clues as to whether this workflow might support it?

Paul H

Fixing PuLID in ComfyUI (Force Reinstall Method) - this method worked for me: Navigate to the custom_nodes folder in your ComfyUI installation directory. Delete the ComfyUI_PuLID folder. Open a command prompt and run: git clone https://github.com/cubiq/PuLID_ComfyUI --recursive Navigate into the newly created ComfyUI_PuLID folder. Install dependencies using: python -m pip install -r requirements.txt Restart ComfyUI and check if PuLID is working correctly. Step-by-Step Guide: 1. Open Command Prompt Press Win + R, type cmd, and hit Enter. 2. Navigate to the custom_nodes directory Run the command: cd "ComfyUI"\custom_nodes Replace "ComfyUI" with your actual ComfyUI installation path (i.e. C:\ComfyUI) 3. Remove the Existing PuLID Installation Run the command: rmdir /S /Q ComfyUI_PuLID 4. Reinstall PuLID Run the command: git clone https://github.com/cubiq/PuLID_ComfyUI --recursive 5. Navigate into the New PuLID Folder Run the command: cd ComfyUI_PuLID 6. Install Dependencies Run the command: python -m pip install -r requirements.txt 7. Restart ComfyUI Close and reopen ComfyUI. Check if PuLID is now working correctly.

Paul H

installing pulid is very painful... still can't solve the missing nodes.

B. Elliott Baker

What software are you using to create the ControlNet pose?

Adam Bossy

Is it possible to make normal instruction from scratch? To point all needed versions of custom nodes? What the point of creating new and new pipelines while no one is possible to be installed from your instructions? I regret that I've subscribed for your patreon. Just missed a lot of time and a little money. This "advanced", "unique" instructions are just SCAM

Andrey Stille

I'm encountering the same extreme slowdown on my rtx4090 with 24GB VRAM. Looks like it'll take about 4 hours to complete this portion of the render.

Joel o' St. Pete

I installed last week and got it to work after a bit of fiddling. I noticed that if there was any image rotation to the main full-body image that the computed rotations could go pretty bad, so I added nodes to resize, rotate and crop to address that. In text above, you mention adding the ability to load a separate face image, but I didn't see that image load node; it was something I tried to add last week, but I was thinking it should happen before the primary rotations in the multi-view setup (didn't work well there), though perhaps it should have injected in the 'select your image' area before the emotions group (that area was initially broken btw with comfy throwing an error). Where's the best place to insert the separate face? Finally you mention fluxgym for taking the next step to lora; I still do more with SDXL/Pony versions, is there a good writeup for moving to lora with those models (while remaining on the local hardware)? Thanks!

James

I have rtx4060 ti with 16GB VRAM, but the sampler custom advanced after the apply pulid section is extremely slow. Any fix?

Relaxing Music

Great but in the previous and even in this one I just can't get past the face comparison. I've followed the new guide to the letter (I believe). # ComfyUI Error Report ## Error Details - **Node ID:** 1374 - **Node Type:** PulidFluxInsightFaceLoader - **Exception Type:** AssertionError - **Exception Message:**

axa

thats cool, but the same custom nodes failed to import, for example ComfyUI_UltimateSDUpscale

Jeremia

Awesome 😎💯

rubz