ComfyUI Auto Installer with Torch 2.8, CUDA 12.9, FaceID, IP-Adapter, InsightFace, Reactor, Triton, DeepSpeed, Flash Attention, Sage Attention, xFormers, Including RTX 5000 Series, Automatic Installers for Windows, RunPod, Massed Compute, Linux

The Explorographer

Everything was working great then it wasn't. I followed your tutorial and did a complete re-install with the latest installers... Now I get this in RED TEXT all the time and failed/poor results when generating. == SwarmUI logs 2025-10-29 14:28 == 14:28:53.513 [Error] Internal error in async task (unlabeled): System.Net.WebSockets.WebSocketException (0x80004005): The remote party closed the WebSocket connection without completing the close handshake. at System.Net.WebSockets.ManagedWebSocket.ThrowEOFUnexpected() at System.Net.WebSockets.ManagedWebSocket.EnsureBufferContainsAsync(Int32 minimumRequiredBytes, CancellationToken cancellationToken) at System.Runtime.CompilerServices.PoolingAsyncValueTaskMethodBuilder`1.StateMachineBox`1.System.Threading.Tasks.Sources.IValueTaskSource.GetResult(Int16 token) at System.Net.WebSockets.ManagedWebSocket.ReceiveAsyncPrivate[TResult](Memory`1 payloadBuffer, CancellationToken cancellationToken) at System.Runtime.CompilerServices.PoolingAsyncValueTaskMethodBuilder`1.StateMachineBox`1.System.Threading.Tasks.Sources.IValueTaskSource.GetResult(Int16 token) at System.Threading.Tasks.ValueTask`1.ValueTaskSourceAsTask.<>c.<.cctor>b__4_0(Object state) --- End of stack trace from previous location --- at SwarmUI.Utils.Utilities.ReceiveData(WebSocket socket, Int32 maxBytes, CancellationToken limit) in D:\SwarmUI\SwarmUI\src\Utils\Utilities.cs:line 309 at SwarmUI.WebAPI.T2IAPI.<>c__DisplayClass1_0.<b__1>d.MoveNext() in D:\SwarmUI\SwarmUI\src\WebAPI\T2IAPI.cs:line 103 --- End of stack trace from previous location --- at SwarmUI.Utils.Utilities.<>c__DisplayClass54_0.<b__0>d.MoveNext() in D:\SwarmUI\SwarmUI\src\Utils\Utilities.cs:line 642 14:29:22.616 [Error] [WebAPI] Error handling API request '/API/GenerateText2ImageWS' for user 'local': Internal exception: System.Net.WebSockets.WebSocketException (0x80004005): The WebSocket is in an invalid state ('Aborted') for this operation. Valid states are: 'Open, CloseReceived' at System.Net.WebSockets.WebSocketValidate.ThrowIfInvalidState(WebSocketState currentState, Boolean isDisposed, WebSocketState[] validStates) at System.Net.WebSockets.ManagedWebSocket.SendAsync(ReadOnlyMemory`1 buffer, WebSocketMessageType messageType, WebSocketMessageFlags messageFlags, CancellationToken cancellationToken) --- End of stack trace from previous location --- at SwarmUI.Utils.Utilities.SendJson(WebSocket socket, JObject obj, TimeSpan maxDuration) in D:\SwarmUI\SwarmUI\src\Utils\Utilities.cs:line 271 at SwarmUI.WebAPI.T2IAPI.GenerateText2ImageWS(WebSocket socket, Session session, Int32 images, JObject rawInput) in D:\SwarmUI\SwarmUI\src\WebAPI\T2IAPI.cs:line 137 at SwarmUI.WebAPI.API.HandleAsyncRequest(HttpContext context) in D:\SwarmUI\SwarmUI\src\WebAPI\API.cs:line 148 14:29:22.619 [Error] [WebAPI] Error handling API request '/API/GenerateText2ImageWS' for user 'local': Internal exception while handling prior exception closure: System.Net.WebSockets.WebSocketException (0x80004005): The WebSocket is in an invalid state ('Aborted') for this operation. Valid states are: 'Open, CloseReceived, CloseSent' at System.Net.WebSockets.WebSocketValidate.ThrowIfInvalidState(WebSocketState currentState, Boolean isDisposed, WebSocketState[] validStates) at System.Net.WebSockets.ManagedWebSocket.CloseAsync(WebSocketCloseStatus closeStatus, String statusDescription, CancellationToken cancellationToken) --- End of stack trace from previous location --- at SwarmUI.WebAPI.API.HandleAsyncRequest(HttpContext context) in D:\SwarmUI\SwarmUI\src\WebAPI\API.cs:line 187 == SwarmUI logs 2025-10-29 14:52 == 14:52:22.084 [Info] Generated an image in 65.97 sec (prep) and 38.78 min (gen) == SwarmUI logs 2025-10-29 15:23 == 15:23:13.155 [Info] User local requested 1 image with model 'Wan2.2-I2V-A14B-Moe-Distill-Lightx2v-Low_fp8_scaled.safetensors'... 15:28:45.192 [Error] Internal error in async task (unlabeled): System.Net.WebSockets.WebSocketException (0x80004005): The remote party closed the WebSocket connection without completing the close handshake. at System.Net.WebSockets.ManagedWebSocket.ThrowEOFUnexpected() at System.Net.WebSockets.ManagedWebSocket.EnsureBufferContainsAsync(Int32 minimumRequiredBytes, CancellationToken cancellationToken) at System.Runtime.CompilerServices.PoolingAsyncValueTaskMethodBuilder`1.StateMachineBox`1.System.Threading.Tasks.Sources.IValueTaskSource.GetResult(Int16 token) at System.Net.WebSockets.ManagedWebSocket.ReceiveAsyncPrivate[TResult](Memory`1 payloadBuffer, CancellationToken cancellationToken) at System.Runtime.CompilerServices.PoolingAsyncValueTaskMethodBuilder`1.StateMachineBox`1.System.Threading.Tasks.Sources.IValueTaskSource.GetResult(Int16 token) at System.Threading.Tasks.ValueTask`1.ValueTaskSourceAsTask.<>c.<.cctor>b__4_0(Object state) --- End of stack trace from previous location --- at SwarmUI.Utils.Utilities.ReceiveData(WebSocket socket, Int32 maxBytes, CancellationToken limit) in D:\SwarmUI\SwarmUI\src\Utils\Utilities.cs:line 309 at SwarmUI.WebAPI.T2IAPI.<>c__DisplayClass1_0.<b__1>d.MoveNext() in D:\SwarmUI\SwarmUI\src\WebAPI\T2IAPI.cs:line 103 --- End of stack trace from previous location --- at SwarmUI.Utils.Utilities.<>c__DisplayClass54_0.<b__0>d.MoveNext() in D:\SwarmUI\SwarmUI\src\Utils\Utilities.cs:line 642 15:29:13.185 [Error] [WebAPI] Error handling API request '/API/GenerateText2ImageWS' for user 'local': Internal exception: System.Net.WebSockets.WebSocketException (0x80004005): The WebSocket is in an invalid state ('Aborted') for this operation. Valid states are: 'Open, CloseReceived' at System.Net.WebSockets.ManagedWebSocket.SendAsync(ReadOnlyMemory`1 buffer, WebSocketMessageType messageType, WebSocketMessageFlags messageFlags, CancellationToken cancellationToken) --- End of stack trace from previous location --- at SwarmUI.Utils.Utilities.SendJson(WebSocket socket, JObject obj, TimeSpan maxDuration) in D:\SwarmUI\SwarmUI\src\Utils\Utilities.cs:line 271 at SwarmUI.WebAPI.T2IAPI.GenerateText2ImageWS(WebSocket socket, Session session, Int32 images, JObject rawInput) in D:\SwarmUI\SwarmUI\src\WebAPI\T2IAPI.cs:line 137 at SwarmUI.WebAPI.API.HandleAsyncRequest(HttpContext context) in D:\SwarmUI\SwarmUI\src\WebAPI\API.cs:line 148 15:29:13.187 [Error] [WebAPI] Error handling API request '/API/GenerateText2ImageWS' for user 'local': Internal exception while handling prior exception closure: System.Net.WebSockets.WebSocketException (0x80004005): The WebSocket is in an invalid state ('Aborted') for this operation. Valid states are: 'Open, CloseReceived, CloseSent' at System.Net.WebSockets.WebSocketValidate.ThrowIfInvalidState(WebSocketState currentState, Boolean isDisposed, WebSocketState[] validStates) at System.Net.WebSockets.ManagedWebSocket.CloseAsync(WebSocketCloseStatus closeStatus, String statusDescription, CancellationToken cancellationToken) --- End of stack trace from previous location --- at SwarmUI.WebAPI.API.HandleAsyncRequest(HttpContext context) in D:\SwarmUI\SwarmUI\src\WebAPI\API.cs:line 187 15:30:46.336 [Info] Generated an image in 0.47 sec (prep) and 7.54 min (gen) 15:33:19.660 [Info] Creating new session 'local' for 192.168.1.244 (via REMOVED IP) Any idea?

Furkan Gözükara

hi this version is all text code. you can open all files with notepad and look what is inside. it is false negative of your system. here virustotal results : https://www.virustotal.com/gui/file/45a34db562e37e1982ced56ed257bcb2f635462aa8ab87dee7dbd9d118179ea4

shen oracle

你的这个版本我顺利安装成功了，谢谢你！但我怀疑这个版本有挖矿的木马，经常无故退出，如果可能请您认真检查一下。

Furkan Gözükara

where did you install? i tested twice on different platforms it didnt happen. default install?

Paul Pfab

i installed the newest v59 (on python 3.10). after loading the custom nodes (via your install batchs) it stops with an error about SOXR. if that happens just downgrade soxr from 1.0.0 to 0.5.0. open your venv (./scripts/activate) and do: pip install soxr==0.5.0. hope that helps...

Furkan Gözükara

Ye it is ok you can ignore this message. Just use and see if any errors

simon DINEEN

WARNING: You did a git pull without building. Will now build for you... The system cannot find the path specified. error: The source specified has already been added to the list of available package sources. Provide a unique source. Determining projects to restore... Restored E:\SWARMUI\SwarmUI\src\SwarmUI.csproj (in 203 ms). SwarmUI -> E:\SWARMUI\SwarmUI\src\bin\live_release\SwarmUI.dll

Furkan Gözükara

yep on my list and they will be easiest to use and best ones hopefully

Mark Sutherland

Awesome as always, any chance of a one click installer for wan animate 2.2 and the new Omni which looks pretty good, thanks again for your hard work

Furkan Gözükara

it is generating. please follow requirements tutorial first : https://youtu.be/DrhUHnYfwC0

Furkan Gözükara

these generates a venv based on your python version you installed. please follow this requirements tutorial. every venv is isolated : https://youtu.be/DrhUHnYfwC0

Taiga

windows machine: "please select your python version"; the one that I want or the one that I have? Are these installers like "portable" installations. I ask because, for your stuff, I created a separate user login specifically using 3.10.. which is somewhat inconvenient in my actual work flow. When I've chosen 3.12 (my main production user); those installs don't work. Just asking for clarity.

Baekdoosixt

Hello , great job as always . Is it normal that the v56 installation is not creating a venv folder ?

Furkan Gözükara

the problem is that your comfyui is not properly doing block swap. add this to your swarmui backend args --reserve-vram 4 and try. if still same try 8 gb. follow vram usage on nvitop. this happens sometimes, i explained in several tutorials e.g. : 6:57 https://youtu.be/adF9X9E0Chs

Tomasz Pasieka

Hello, hi. I have a 4090 card plus 64 GB of RAM, and if I try to generate any Queen photos from our settings, from the Queen model, it can take up to 10 minutes. Even though I see everything generating on the graphics card, not on the CPU, it still takes such an insane amount of time to generate photos. The photos come out beautiful, but it takes 9-10 minutes even with Lora, eight steps, and so on. Can anyone help me with what's causing this problem? Prompt: A kinetic sculpture of a galloping horse, made of interlocking metal rods and panels in brushed gunmetal and oxidized copper. Suspended in motion, with rotating parts casting mechanical shadows on a concrete pedestal.⧉, Model: Qwen-Image-Edit-Plus-2509-Q8_0.gguf, Seed: 546794641, Steps: 12, CFG Scale: 1, Aspect Ratio: 1:1 (Square), Width: 1328, Height: 1328, Sampler: Euler Ancestral (Randomizing), Scheduler: Simple, Preferred DType: Default (16 bit), Automatic VAE: true, LoRAs: 0: Qwen-Image-Lightning-8steps-V2.0 : 0.7,, date: 2025-09-30, prep_time: 3.03 sec, generation_time: 9.15 min, Swarm Version: 0.9.7.0

Furkan Gözükara

Bu senin bir custom node ben yüklemiyorum. Bunu geliştirene yaz ComfyUI bazı şeyleri bozdu sanırım, bunu geliştiren düzeltmesi lazım

Cemil Hacimahmutoglu

hocam tüm denemelerime rağmen (2 kez temiz kurulum yaptım) ilk install.bat dosyasından sonrasında aşağıdaki hatayı alıyorum. bu hatayı yok sayp tüm düğümleri hatasız yükelemem rağmen sorunsuz kullandığım workflow hata veriyor. Sorun ilk yüklemede başlayan hatamı acaba sorun nerde. [ComfyUI-Manager] default cache updated: https://api.comfy.org/nodes FETCH DATA from: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json [DONE] [ComfyUI-Manager] broken item:{'author': 'fr0nky0ng', 'title': 'ComfyUI-Face-Comparator', 'description': 'This is a node to detect the similarity between two faces', 'install_type': 'git-clone', 'reference': 'https://github.com/fr0nky0ng/ComfyUI-Face-Comparator'} [ComfyUI-Manager] All startup tasks have been completed.

Furkan Gözükara

no it is not because of that. it is because of custom nodes developers : https://github.com/kijai/ComfyUI-KJNodes/pull/386 but i will compile with cu129 now. however as i said i tested it works fine with cu128 too started compile will update installers hopefully

W4v3r1d3r

In your latest 13 of September update, comfyui installer you use cuda12.9 but I see in the installer your sageattention is using cuda12.8. I tried to run a workflow for wan2.2 and got stuck in Patching comfy attention to use sageattn. Is that an error or should it work?

Neil Rhodes

after installing fresh ComfiUI and it complained that Xformers was not installed and moving on to installing sawrmUI i get this error The system cannot find the path specified. WARNING: You did a git pull without building. Will now build for you... The system cannot find the path specified. 'dotnet' is not recognized as an internal or external command, operable program or batch file. 'dotnet' is not recognized as an internal or external command, operable program or batch file. The system cannot find the path specified. The system cannot find the path specified. Press any key to continue . . . I cannot install SwarmUI there is not Dotnet folder as you told me to delete it

Furkan Gözükara

delete venv folder and use latest installer and run install. it should fix

Neil Rhodes

i have my current Comfui ui inside A:\SwarmUI_Model_Downloader_v49\SwarmUI\dlbackend\comfy\ i just broke it by doing the wrong thing how do i use this new installer to fix it? do i use your installer here A:\SwarmUI_Model_Downloader_v49\SwarmUI\dlbackend ( i have the last version of Swarm updater 71 before your latest one spotted today)

Furkan Gözükara

just update comfyui and it will be updated to latest. zip file versions are only necessary when some files inside zip file is changed

Furkan Gözükara

you dont need to use swarmui our comfyui installer supports everything as a base. then use your workflow

djbuzz

I don't want to use SwarmUI. Everything else is great. Advice: you should not push so much for us to use SwarmUI.

Taiga

where can I see which version of this that I am using?

Furkan Gözükara

your installation wrong. please follow tutorial : https://youtu.be/R02kPf9Y3_w here my logs : https://gist.github.com/FurkanGozukara/6e5de82d06ba0632c990bcea02833f50 if you upgrade gold tier i can connect your pc and show my used template : pytorch:2.2.0-py3.10-cuda12.1.1-devel-ubuntu22.04

hosung jang

Hi, i'm using your latest "Comfy_UI_V43" installer on Runpod Pytorch 2.2.0 runpod/pytorch:2.2.0-py3.10-cuda12.1.1-devel-ubuntu22.04. I do 1) "export HF_HOME="/workspace" chmod +x RunPod_Install.sh ./RunPod_Install.sh" and then 2) apt update apt install psmisc fuser -k 3000/tcp cd /workspace/ComfyUI/venv source bin/activate cd /workspace/ComfyUI python main.py --listen 0.0.0.0 --port 3000 --use-sage-attention and it does not seem to be working anymore. Here's the error log ```oot@166972e93996:/workspace# apt update apt install psmisc fuser -k 3000/tcp cd /workspace/ComfyUI/venv source bin/activate cd /workspace/ComfyUI python main.py --listen 0.0.0.0 --port 3000 --use-sage-attention Hit:1 http://security.ubuntu.com/ubuntu jammy-security InRelease Hit:2 https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64 InRelease Hit:3 http://archive.ubuntu.com/ubuntu jammy InRelease Hit:4 https://ppa.launchpadcontent.net/deadsnakes/ppa/ubuntu jammy InRelease Hit:5 http://archive.ubuntu.com/ubuntu jammy-updates InRelease Hit:6 http://archive.ubuntu.com/ubuntu jammy-backports InRelease Reading package lists... Done Building dependency tree... Done Reading state information... Done 138 packages can be upgraded. Run 'apt list --upgradable' to see them. Reading package lists... Done Building dependency tree... Done Reading state information... Done psmisc is already the newest version (23.4-2build3). The following packages were automatically installed and are no longer required: fontconfig libaom3 libasound2 libasound2-data libass9 libavc1394-0 libavcodec58 libavutil56 libbluray2 libcairo-gobject2 libcairo2 libchromaprint1 libcodec2-1.0 libdatrie1 libdav1d5 libfribidi0 libgdk-pixbuf-2.0-0 libgdk-pixbuf2.0-common libgme0 libgraphite2-3 libgsm1 libharfbuzz0b libmfx1 libmp3lame0 libmpg123-0 libnuma1 libogg0 libopenjp2-7 libopus0 libpango-1.0-0 libpangocairo-1.0-0 libpangoft2-1.0-0 libpixman-1-0 libraw1394-11 librsvg2-2 libshine3 libslang2 libsnappy1v5 libsoxr0 libspeex1 libswresample3 libthai-data libthai0 libtheora0 libtwolame0 libudfread0 libusb-1.0-0 libva-drm2 libva-x11-2 libva2 libvdpau1 libvorbis0a libvorbisenc2 libvpx7 libwebpmux3 libx264-163 libx265-199 libxcb-render0 libxrender1 libxvidcore4 libzvbi-common libzvbi0 ocl-icd-libopencl1 xkb-data Use 'apt autoremove' to remove them. 0 upgraded, 0 newly installed, 0 to remove and 138 not upgraded. [START] Security scan WARNING: No metadata found in ./venv/lib/python3.10/site-packages [DONE] Security scan ## ComfyUI-Manager: installing dependencies. (GitPython) WARNING: Error parsing dependencies of torch: [Errno 2] No such file or directory: '/workspace/ComfyUI/venv/lib/python3.10/site-packages/torch-2.7.1.dist-info/METADATA' ## ComfyUI-Manager: installing dependencies done. ** ComfyUI startup time: 2025-07-31 17:46:36.982 ** Platform: Linux ** Python version: 3.10.12 (main, Nov 20 2023, 15:14:05) [GCC 11.4.0] ** Python executable: /workspace/ComfyUI/venv/bin/python ** ComfyUI Path: /workspace/ComfyUI ** ComfyUI Base Folder 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 WARNING: No metadata found in ./venv/lib/python3.10/site-packages WARNING: No metadata found in ./venv/lib/python3.10/site-packages [ComfyUI-Manager] PyTorch is not installed Prestartup times for custom nodes: 49.4 seconds: /workspace/ComfyUI/custom_nodes/ComfyUI-Manager Traceback (most recent call last): File "/workspace/ComfyUI/main.py", line 145, in import comfy.utils File "/workspace/ComfyUI/comfy/utils.py", line 20, in import torch ModuleNotFoundError: No module named 'torch'```. I tried with 4090, 5090 and both are throwing the same errors. I have been using this same steps for the last few months and it was working great until recently.

Furkan Gözükara

you can set comfyui model folder path or swarmui folder path both have options

Winnie

Hi Furkan, I followed your guide and all works fine! (ComfyUI + SwarmUI). I tested SwarmUI and works as intended. Now I would like to use standalone ComfyUI and I am wondering how I can set it up to just use one "models" folder (e.g. redirect to SwarmUI models folder or viceversa). I would like to try to avoid having the models duplicated into my ComfyUI installation folder and SwarmUI folder. Thanks!!

Furkan Gözükara

pod could be broken sadly.

Jesen

I get to about 9:10 in the instructions in the video, let it run installing backend, but it gives this error next: [Error] Self-Start ComfyUI-0 on port 7823 failed. AutoRestart ignored as this was an initial launch failure , then trying again it just repeats the error, on port 7824,7825, etc.

Furkan Gözükara

same edit bat file and add that as args python.exe -s main.py --windows-standalone-build --use-sage-attention --reserve-vram 3

michele carlone

Hi Furkan. How do I add the --reserve-vram 3 parameter directly in ComfiUI and not through the SwarmUI backend? Some workflows saturate my RTX5090 to 100%, even with Q8 models.

Furkan Gözükara

you can look Windows_Install.bat (open with any text editor) and you will see how it is installed. you need to activate venv and run pip installer according to your python version

LoloFakess

Is there a way to download sageattention only?

Furkan Gözükara

these are not standard nodes what did you install extra?

Muhammad Nisar

Hello Dear Sir i used your 1 click installer bat file and my comfyui working very well but today i notice i am getting an error which is below WARNING: some comfy_api_nodes/ nodes did not import correctly. This may be because they are missing some dependencies. IMPORT FAILED: nodes_ideogram.py IMPORT FAILED: nodes_openai.py IMPORT FAILED: nodes_minimax.py IMPORT FAILED: nodes_veo2.py IMPORT FAILED: nodes_kling.py IMPORT FAILED: nodes_bfl.py IMPORT FAILED: nodes_luma.py IMPORT FAILED: nodes_recraft.py IMPORT FAILED: nodes_pixverse.py IMPORT FAILED: nodes_stability.py IMPORT FAILED: nodes_pika.py IMPORT FAILED: nodes_runway.py IMPORT FAILED: nodes_tripo.py IMPORT FAILED: nodes_moonvalley.py IMPORT FAILED: nodes_rodin.py IMPORT FAILED: nodes_gemini.py This issue might be caused by new missing dependencies added the last time you updated ComfyUI. Please do a: pip install -r requirements.txt so please guide me how to fix this error. and also guide me how i can update my comfyui and how to update all everything with single click i mean all nodes and comfyui also. one more thing i want to know did i need to install CUDA now or its included in your installer file? Thank you.

Furkan Gözükara

i dont think he plans to add. but you can use workflows in comfyui tab

Chris

Hi Furkan, will Multitalk work directly in SwarmUI in the future?

Furkan Gözükara

this wont install cuda. you have to first follow this tutorial : https://youtu.be/DrhUHnYfwC0 once you followed above tutorial this will generate venv and install torch 2.7, xformers, flash attention, sage attention, deepspeed, insightface, triton, onnxruntime-gpu, GGUF, accelerate, diffusers

Muhammad Nisar

First of all Thank you for all of your sharing. I just want to confirm before using this Auto Installation. let me share my PC specifications so you just confirm me that can i use this Auto Installation. Windows11 RTX5070 12GB so i just want to know did this package has all the dependencies like Cuda, PyTorch, Xformers, Insiteface etc and did these all work with my PC?

Furkan Gözükara

comfyui wont impact your swarmui. you can use latest zip file update bat file. also you can update swarmui too.

JS

I attempted to install Flux Kontext via your tutorial, I made the mistake of updating Swarm before updating CUI to v36 not realizing that there is no comfyupdate.bat included the in CUI v27 install (you may want to clarify that CUI v27 has no update.bat when or if you update the Flux Kontext tutorial to save others from the same issue). I am not sure if I wrecked my entire Swarm install by trying to follow the Flux Kontext tutorial but now I get an error that I presume pertains to not having CUIv36 installed previous to updating Swarm. I guess I have to remove remove CUIv27 and replace it with CUIv36? How do I do that without causing more damage than I already have?

Furkan Gözükara

thank you so much

Furkan Gözükara

if everything working right now dont do anything. maybe they removed recently i am not sure

JS

You mentioned missing diffusers, dose that mean that previous installations of your 1 click installer for swarm are not functioning properly? If the previous install is not functioning will using the updater with the one click installer I already have installed fix the diffuser issue? Being comfy is installed separately I am not sure how to completely remove it to use the new version.

Rudbby

It's great to wake to a new drop from you bro! Works like a charm, thanks for your hard work.

Furkan Gözükara

i would recommend to open an issue here looks like SwarmUI dev caused error : https://github.com/mcmonkeyprojects/SwarmUI/issues

Ec Jep

Do you have an idea why this error is popping up today with my swarmui installation? I've tried an older installation that doesn't use the separate comfyui and it also shows the same error: [ComfyUI-0/STDERR] Checkpoint files will always be loaded safely. [ComfyUI-0/STDERR] Total VRAM 24564 MB, total RAM 65426 MB [ComfyUI-0/STDERR] pytorch version: 2.7.0+cu128 [ComfyUI-0/STDERR] xformers version: 0.0.30+3abeaa9e.d20250424 [ComfyUI-0/STDERR] Enabled fp16 accumulation. [ComfyUI-0/STDERR] Set vram state to: NORMAL_VRAM [ComfyUI-0/STDERR] Device: cuda:0 NVIDIA GeForce RTX 4090 : cudaMallocAsync [ComfyUI-0/STDERR] Using sage attention [ComfyUI-0/STDERR] Traceback (most recent call last): [ComfyUI-0/STDERR] File "D:\SD\comfyui_v28\ComfyUI\main.py", line 130, in [ComfyUI-0/STDERR] import execution [ComfyUI-0/STDERR] File "D:\SD\comfyui_v28\ComfyUI\execution.py", line 13, in [ComfyUI-0/STDERR] import nodes [ComfyUI-0/STDERR] File "D:\SD\comfyui_v28\ComfyUI\nodes.py", line 41, in [ComfyUI-0/STDERR] from comfy_config import config_parser [ComfyUI-0/STDERR] File "D:\SD\comfyui_v28\ComfyUI\comfy_config\config_parser.py", line 5, in [ComfyUI-0/STDERR] from pydantic_settings import PydanticBaseSettingsSource, TomlConfigSettingsSource [ComfyUI-0/STDERR] ModuleNotFoundError: No module named 'pydantic_settings'

Furkan Gözükara

here AMD instructions : https://github.com/comfyanonymous/ComfyUI?tab=readme-ov-file#directml-amd-cards-on-windows and here : https://github.com/comfyanonymous/ComfyUI?tab=readme-ov-file#for-amd-cards-not-officially-supported-by-rocm

AgioTAAA

Are there alternatives for AMD GPUs? I'm new to all of this and been wanting to use Wan 2.1 on my RX 7800XT

Furkan Gözükara

i would say delete that folder and try again : C:\Program Files\dotnet

Furkan Gözükara

you have to install it manually : https://youtu.be/fTzlQ0tjxj0

Furkan Gözükara

sadly i dont know NPU

Rahul Gautam

Hi...How can i use my laptops NPU into the picture...its around 18gb...will it be of any help?

Taiga

I followed your python/c++ etc video all the way through. I ran the latest (v28) install and this is the error when trying to run: The system cannot find the path specified. python.exe: can't open file 'D:\\Comfy_UI_V28\\ComfyUI\\main.py': [Errno 2] No such file or directory ~~~~~ there is essentially nothing in comfyui directory.

Michael

Hi, upon running "windows_install_swarmUI" I am getting the error "C:\Program Files\dotnet\sdk\8.0.410\Microsoft.Common.CurrentVersion.targets(5034,5): error MSB3094: "DestinationFiles" refers to 1 item(s), and "SourceFiles" refers to 14 item(s). They must have the same number of items."

Furkan Gözükara

i will make a new video for this hopefully. just to show comfyui install and swarmui install

Lucy

Runpod instruction file is different from your video tutorial. So I dont know how to start comfy ui

Furkan Gözükara

we suport 3.10, 3.11 and 3.12. 3.13 wont work. have one of those installed. i prefer 3.10.11

Taiga

I'm on Python 3.13. should I select python 3.12 in install option?

Furkan Gözükara

what is your gpu and python version?

tokyoplastic

Furkan Gözükara

hi you need to set CFG to 1 for both models

Iván López

Also, previsualization is not working. Even though it is selected in the backend config. Thank you!

Andres Eduardo Damonte

Hi, can you help me? I don't know what I'm doing wrong. Look at the images — see how they turn out with FLux dev and with HiDream. https://postimg.cc/gallery/c3MCSGN

Iván López

Hi Furkan! Thanks for the tutorial. I have been using Swarm and Wan2.1 from your installers. What is the difference between those and these? I dont see any updater. Is this better or the ones you had already that are updating?

Furkan Gözükara

well i tested in swarmui and even flux image generation got speed up from sage attention.

Pablo Montero

Ey Furkan, everyone is finding difficult to speed up WAN with Triton and Sageattetion. I tryed to do so and the installation looked okey, but couldn't make it work at the end. It would be awesome if you could confirm that this optimization works properly in ComfyUI, not only in Swarm. Thank you!

Furkan Gözükara

in swarm ui we have to use --use-sageattention command. i am not sure for comfyui. so research that. probably same

Lucas D'Erman

the installation seemed to have worked well. but when selecting the sageattention node in comfyUI, should we leave it as 'auto' or 'fp16_cuda'/'fp16_triton'? If I select triton or cuda, my comfy ui crashes, if I select 'auto', then it works but the speed is not that much different than sageattention 1? I'm using a 5090 btw

Furkan Gözükara

just replied you in details from private message

Hipno

Hey Sir! Are you able to make video on building character set images from only 1 image? Like mickmumpitz did on his YouTube channel?

Furkan Gözükara

yes will do hopefully sorry for delay

Furkan Gözükara

patreon has dm. just send you one

michele carlone

Hi Furcan, but in patreon there is no more direct message box? I should ask you for advice in private.

JC

Hey Doc, please can you update your SwarmUI to include these latest Comfy updates?

Furkan Gözükara

you are welcome. i tested that in swarmui worked perfect

Sonivas Sx

thank you for always being everywhere and trying to make everything work !!

Furkan Gözükara

just updated and now you can install sage attention on windows

Furkan Gözükara

it arrived today. right now i am testing on wan 2.1 . however sage attention reduces quality more. so i can add it to installers but it is a choice. let me update installers

Sonivas Sx

can we have the option of installing Sage Attention rather than Flash attention? sage is said to be two times faster. i also installed the 3.12 version and when i run "pip show triton" in the venv file it says triton not found

Furkan Gözükara

no idea :d i dont use comfyui on daily bases :D

Sonivas Sx

removed the comfyui-mtb apparently the nodes with noteplus dont work anymore, also when i update comfyui from manager, does everything gets updated even the dependencies and everything needed to run smoothly ?

Furkan Gözükara

hi probably extension requires some older libraries or maybe some other specific things. that is what makes comfyui hard to work.

Sonivas Sx

i followed your instructions and the used this workflow https://civitai.com/models/1129063/flux-modular-wf had to download some nodes and when i tried to open a new tab i got this error "TypeError: Cannot read properties of undefined (reading 'uuid') TypeError: Cannot read properties of undefined (reading 'uuid') at onRemoved (http://127.0.0.1:8188/extensions/comfy_mtb/note_plus.js:604:37) at NotePlus.onRemoved (http://127.0.0.1:8188/assets/index-N7tf9g2I.js:172235:23) at LGraph2.clear (http://127.0.0.1:8188/assets/index-N7tf9g2I.js:90543:26) at app.graph.clear (http://127.0.0.1:8188/extensions/ComfyUI-Manager/comfyui-manager.js:1450:15) at LGraph2.configure (http://127.0.0.1:8188/assets/index-N7tf9g2I.js:91540:25) at LGraph$1.configure (http://127.0.0.1:8188/assets/index-N7tf9g2I.js:222619:26) at LGraph.configure (http://127.0.0.1:8188/extensions/ComfyUI-Custom-Scripts/js/reroutePrimitive.js:14:29) at ComfyApp.loadGraphData (http://127.0.0.1:8188/assets/index-N7tf9g2I.js:222858:18) at async app.loadGraphData (http://127.0.0.1:8188/extensions/ComfyUI-Manager/components-manager.js:774:9) at async Object.loadBlankWorkflow (http://127.0.0.1:8188/assets/index-N7tf9g2I.js:220176:5) This may be due to the following script: /extensions/comfy_mtb/note_plus.js"

Furkan Gözükara

you can use any of installers. which Python you have installed to your PC?

耀良 梁

I am currently using Comfy_UI_V12. I want to experience Triton (3.2+) and DeepSpeed ​​(0.16.4), insightface (0.7.3), onnxruntime-gpu and Flash Attention (2.7.4 - I compiled for all 3 Python versions) into our VENV Windows. I downloaded Comfy_UI_V16. Which file can I install?

Furkan Gözükara

great

Chris

I did another fresh reinstall and now it seems to work :)

Chris

You mean this video? https://youtu.be/DrhUHnYfwC0 I have it already installed according to this video

Furkan Gözükara

make another install and make sure you have installed pre compiled backend as in video

Chris

Did fresh install, still this error: G:\swarm2\SwarmUI\dlbackend\comfy\python_embeded>python.exe pip install -U --pre triton-windows python.exe: can't open file 'G:\\swarm2\\SwarmUI\\dlbackend\\comfy\\python_embeded\\pip': [Errno 2] No such file or directory G:\swarm2\SwarmUI\dlbackend\comfy\python_embeded>python.exe pip install huggingface_hub ipywidgets hf_transfer python.exe: can't open file 'G:\\swarm2\\SwarmUI\\dlbackend\\comfy\\python_embeded\\pip': [Errno 2] No such file or directory G:\swarm2\SwarmUI\dlbackend\comfy\python_embeded>python.exe pip install moviepy python.exe: can't open file 'G:\\swarm2\\SwarmUI\\dlbackend\\comfy\\python_embeded\\pip': [Errno 2] No such file or directory

Chris

You, Sir, are a legend!

Furkan Gözükara

Updated : https://www.patreon.com/posts/114517862 Read changelogs :)

Chris

Nice! Will wait! Thank you :)

Furkan Gözükara

i dont have but let me update our SwarmUI installer so you will have and you can see how to do :D

Chris

Hi Furkan, do you have a tutorial on how to install triton for acceleration in swarmUI in windows 10 / 11?

Furkan Gözükara

massed compute has Jupyter Notebook installed try it and see if works. since i am not sure how exactly you want to use i cant comment in more details. but on massed compute you can upload and download very fast from google drive

matias correa

Hey, I'm a new supporter. Thanks for all the tutorials! I've been using Google Colab to run ComfyUI with my Google Drive storage (since I don’t have enough storage on my notebook). Is it possible to run the official Jupyter Notebook from comfyanonymous on Massed Compute? My main goal right now is to run ComfyUI on Massed Compute with cloud storage, so I can have full control to upload or download files like models, custom nodes, etc., directly to the cloud storage. I’m open to using any other cloud storage option besides Google Drive. Thanks!

Furkan Gözükara

personally i dont feel we need ComfyUI . but some people still preferring it.

baka

Is there any point of using ComfyUI when we have SwarmUI? SwarmUI uses Comfy as Backend and is 100 times more intuitive, especially for beginners!

Furkan Gözükara

yes it is huge. i should also start working on it as you said

v33ts

Learned all my lora and fine tuning from you. I hope you start doing comfyui workflows. I know you prefer other generators but comfyui is really powerful and the way you explain things, I could see lots of content

Furkan Gözükara

currently it auto installs Insightface on massed compute. just follow Massed_Compute_Instructions_READ.txt

C

Hi, I'm a long time supporter, thanks for all the work. I use Massed Compute following your tutorials. I am trying to install Insightface on your SECourses Massed Compute Image. What do I need to do exactly? Do I need to install any python or any tools into the Massed Compute image? How do I install Insightface? Thank you

Furkan Gözükara

for kaggle we have this swarmui you can use it : https://www.patreon.com/posts/106650931

Ava Frigg

can I use this in kaggle?

Furkan Gözükara

sure let me know

JamZam WamBam

Ok. I'll try it on the C drive and get back to you. Thanks!

Furkan Gözükara

just made a fresh install and has no issues . try on C drive and let me know [START] Security scan [DONE] Security scan ## ComfyUI-Manager: installing dependencies. (GitPython) [notice] A new release of pip is available: 24.0 -> 24.3.1 [notice] To update, run: python.exe -m pip install --upgrade pip ## ComfyUI-Manager: installing dependencies done. ** ComfyUI startup time: 2024-12-16 00:28:27.631171 ** Platform: Windows ** Python version: 3.11.9 (tags/v3.11.9:de54cf5, Apr 2 2024, 10:12:12) [MSC v.1938 64 bit (AMD64)] ** Python executable: C:\Users\Furkan\Downloads\Comfy_UI_V9\ComfyUI\venv\Scripts\python.exe ** ComfyUI Path: C:\Users\Furkan\Downloads\Comfy_UI_V9\ComfyUI ** Log path: C:\Users\Furkan\Downloads\Comfy_UI_V9\ComfyUI\comfyui.log [notice] A new release of pip is available: 24.0 -> 24.3.1 [notice] To update, run: python.exe -m pip install --upgrade pip [notice] A new release of pip is available: 24.0 -> 24.3.1 [notice] To update, run: python.exe -m pip install --upgrade pip Prestartup times for custom nodes: 20.2 seconds: C:\Users\Furkan\Downloads\Comfy_UI_V9\ComfyUI\custom_nodes\ComfyUI-Manager Total VRAM 24564 MB, total RAM 65366 MB pytorch version: 2.5.0+cu124 xformers version: 0.0.28.post2 Set vram state to: NORMAL_VRAM Device: cuda:0 NVIDIA GeForce RTX 3090 Ti : cudaMallocAsync Using xformers cross attention ****** User settings have been changed to be stored on the server instead of browser storage. ****** ****** For multi-user setups add the --multi-user CLI argument to enable multiple user profiles. ****** [Prompt Server] web root: C:\Users\Furkan\Downloads\Comfy_UI_V9\ComfyUI\web ### Loading: ComfyUI-Manager (V2.55.5) ### ComfyUI Version: v0.3.7-33-g6d1a3f7 | Released on '2024-12-15' Import times for custom nodes: 0.0 seconds: C:\Users\Furkan\Downloads\Comfy_UI_V9\ComfyUI\custom_nodes\websocket_image_save.py 0.0 seconds: C:\Users\Furkan\Downloads\Comfy_UI_V9\ComfyUI\custom_nodes\ComfyUI_IPAdapter_plus 0.4 seconds: C:\Users\Furkan\Downloads\Comfy_UI_V9\ComfyUI\custom_nodes\ComfyUI-Manager Starting server To see the GUI go to: http://127.0.0.1:8188

JamZam WamBam

When I run the Windows_Run_GPU.bat file I get this error :( [START] Security scan [DONE] Security scan ## ComfyUI-Manager: installing dependencies done. ** ComfyUI startup time: 2024-12-15 16:35:16.668262 ** Platform: Windows ** Python version: 3.11.9 (tags/v3.11.9:de54cf5, Apr 2 2024, 10:12:12) [MSC v.1938 64 bit (AMD64)] ** Python executable: A:\ComfyV9\ComfyUI\venv\Scripts\python.exe ** ComfyUI Path: A:\ComfyV9\ComfyUI ** Log path: A:\ComfyV9\ComfyUI\comfyui.log [notice] A new release of pip is available: 24.0 -> 24.3.1 [notice] To update, run: python.exe -m pip install --upgrade pip [notice] A new release of pip is available: 24.0 -> 24.3.1 [notice] To update, run: python.exe -m pip install --upgrade pip Prestartup times for custom nodes: 2.1 seconds: A:\ComfyV9\ComfyUI\custom_nodes\ComfyUI-Manager Traceback (most recent call last): File "A:\ComfyV9\ComfyUI\main.py", line 130, in import comfy.utils File "A:\ComfyV9\ComfyUI\comfy\utils.py", line 20, in import torch File "A:\ComfyV9\ComfyUI\venv\Lib\site-packages\torch\__init__.py", line 899, in raise ImportError( ImportError: Failed to load PyTorch C extensions: It appears that PyTorch has loaded the `torch/_C` folder of the PyTorch repository rather than the C extensions which are expected in the `torch._C` namespace. This can occur when using the `install` workflow. e.g. $ python setup.py install && python -c "import torch" This error can generally be solved using the `develop` workflow $ python setup.py develop && python -c "import torch" # This should succeed or by running Python from a different directory. Press any key to continue . . .

Furkan Gözükara

usually they work best with google chrome

Koudkunstje

I've already got comfyui running. I've followed the instructions on the comfui github page except i've created and activated a venv for it. Although the UI isn't that responsive in Firefox and Linux, kinda laggy.

Furkan Gözükara

Hello. Use Massed Compute for linux. remember you need to install python cuda cudnn git c++ tools yourself on linux. python 3.10, cuda 12.4, cudnn 8.9.7, c++ tools that is proper for tthese. i have a tutorial for windows here : https://youtu.be/DrhUHnYfwC0

Koudkunstje

I switched from windows to linux. The title says installers for windows, runpod, massed compute, linux... I don't see installer for linux?

Furkan Gözükara

hi it is so straight forward. make folders like this c:/comfy1 , c:/comfy2 etc, and extract zip file inside each one of them and use Windows_Install.bat so each one will be separate unique one and you can use a different workflow extensions nodes etc on each one without breaking each other

Curious Party

is there a basic tutorial or video that explains how to go through this process? I am eager to set this up locally on both an a6000 and a 3080ti. which files in the .zip will I need to run/how do I run them? thanks!

Furkan Gözükara

well if i needed yes it would make. my installer generates unique venv so you can use it.

Aixsponza

So, do you have parallel installs for different workflows?

Aixsponza

😵‍💫

Furkan Gözükara

Nope. One of the biggest problem of ComfyUI . You need to have separate installations so each one will have their own VENV

Aixsponza

Is there a best practice to avoid compatibility issues of libraries between lots of installed custom nodes?

Furkan Gözükara

great

Taylor

This worked. Thanks for the update.

Furkan Gözükara

yes i am aware. if you need more help for swarmui, his developer suggesting to join their server, he told me that, here their server : https://discord.gg/smdGbypw

Brandon

Also that video has no audio if you are not aware. Im using on windows so the same thong applies. I would copy the path of main.py in my comfui folder to the backend in swarmui. It will load all my settings, custom nodes, manager?

Furkan Gözükara

it is harder if you are not expert dont do it . but you can use your comfyui as backend in swarmui. watch this tutorial : https://youtu.be/1TBqf3AMcBI we use existing comfyui as backend in swarmui in kaggle

Brandon

I know you said comfy can be replace in swarm but want to confirm with this posting if i Can delete the comfy ui files in swarm ui and extract the installer and install in the folder will swarm ui recognize it or would i need to change settings in the backend on swarm? Is it possible to add in updates on swarm tutorial and this one ?

Furkan Gözükara

every xformer requires a specific torch you need to activate venv and do pip install e.g. pip3 install torch==2.4.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124 --upgrade pip install xformers==0.0.27.post2

Rob Con

How to update torch and xformers manualy?

Taylor

Okay, will try. Thank you!

Furkan Gözükara

sorry please install v6

Furkan Gözükara

try v5 just installed no issues. install into c:/Comfy_UI_V5 folder

Taylor

Yes, correct

Furkan Gözükara

are you on windows?

Taylor

I'm running into an error: numpy isn't found. How can I fix this? I've installed all the correct Python/Cuda/etc versions following your video

Furkan Gözükara

awesome

druhl

Works flawlessly!

Furkan Gözükara

thanks i agree. fresh install is best to test

reaper557

This is great! ComfyUI is tons of fun, but I always seem to break something when I do clean install. Your auto installers are life savers. :D

Furkan Gözükara

Awesome let me know

Nenad Kuzmanovic

Great. I will test it today on A6000 on MassedCompute.

Furkan Gözükara

added auto install Insightface with v2 zip file

Nenad Kuzmanovic

Nice.. I will test it, and Insightface will be problem with some workflows, IPAdapter may not working without it.

Furkan Gözükara

ComfyUI Manager already included in the auto installer. it installs

Nenad Kuzmanovic

For the start, can you add just ComfyUI manager to this auto instal? Cause you can download and instal everything with it.. Just check if it's working.. No need for any aditional instals, cause when you load a workflow, it will tell you what are you missing and through Manager you can download models and instal extensions... Besides that, InsightFace is kinda tricky to instal properly, and if you are planing to update instaler, those are mandatory: ComfyUI Manager and InsightFace... I'm 24/7 on ComfyUI btw.. Amazing software

Furkan Gözükara

there are so many models. i may add optionally them. i will check later

Rabadan

Very good! It would be even better if there was an automatic way to download all the models needed by the ipAdapter (there are several), normally described in https://github.com/cubiq/ComfyUI_IPAdapter_plus?tab=readme-ov-file#installation. Because for me, I still get error messages without the models. Is possible?