Home
Creators
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register Favorites Keys Review DMs Logout
Community
Telegram ThePornDude
Home Creators Posts Import Register Favorites Logout

Z-IMAGE BASE THE NEW AI KING FINALLY HERE! < 8GB VRAM REQUIRED! V2 (Patreon)

Published:
2026-01-31 20:31:44
Edited:
2026-02-08 05:41:25
Imported:
2026-06
Tags:

Downloads

Content

UPDATE 7 february 2026: Hey everyone! I just released a new video update for the V2 Workflow that you can watch here: https://youtu.be/PAh870We7tIIn this update I added a ControlNet workflow for better pose and composition control, a Detailer Daemon pass to crank the micro details even higher, AND THE 8 STEPS LORA that allows you to generate images in 8 steps instead of 25!

Just use the Z-IMAGE_BASE_ULTRA-MODELS-NODES_INSTALL-V2.bat installer, put it in your comfyui folder, run it to install the missing nodes and files, then update comfyui to the latest version with the update comfyui bat file inside the update folder then download the the updated workflow from this post: https://www.patreon.com/posts/z-image-base-new-149607616drop it into your ComfyUI folder and you’re ready to go 😉

Hey everyone! I've created a 1-click installer for THE NEW TEXT TO IMAGE AI KING BASE MODEL called “Z-IMAGE”. The BIG difference compared to Z-IMAGE-TURBO is that this is the BASE model, that has amazing  image image diversity and styles, meaning that real finetuning and training is now possible, so the community can finally cook up new LoRAs, styles, characters and insane custom versions while running with LESS THAN 8GB OF VRAM! JUST INSANE! 

You can check out the video right here: https://youtu.be/9EeCWOJvLOw

The workflow: https://www.patreon.com/posts/149607616

The installer of course automates the entire install process for comfyUI and the models downloads!

LOCAL INSTALL!!! RECOMMENDED TO START FROM SCRATCH

Download the Z-IMAGE_BASE_ULTRA-COMFYUI-MANAGER_AUTO_INSTALL-V2.bat

Run the bat file

Wait for the install to finish

The webui will launch, load the Z-IMAGE_BASE_ULTRA_WORKFLOW-V2.json file and there you go😎

DON'T FORGET TO GO UPDATE COMFYUI BY GOING INSIDE THE UPDATE FOLDER AND RUNNING THE COMFYUI UPDATE BAT FILE!!!!

IF YOU ALREADY HAVE COMFYUI INSTALLED USING A PREVIOUS INSTALLER:

Download the Z-IMAGE_BASE_ULTRA-MODELS-NODES_INSTALL-V2.bat and place it in the ComfyUI_windows_portable\ComfyUI folder

Run the bat file

Wait for the install to finish

???

Profit 😎

IF YOU GET AN ERROR WHEN LAUNCHING COMFYUI AFTER USING THIS INSTALLER SAYING "AssertionError: Torch not compiled with CUDA enabled" it's ok don't panic.

Just go inside the python embedded folder, click on the folder path, type cmd press enter, this will bring the command prompt window and there type:

python.exe -m pip uninstall torch

python.exe -s -m pip install --upgrade torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

--->replace cu128 by your cuda version (if you don't know your cuda version just open a new cmd window and type nvcc --version it will then tell you at the end Cuda compilation tools, release 12.8, V12.8.93 or something different like 12.6, in that case just use cu126 instead of cu128)

IF YOU ARE USING RUNPOD:

Create an account if you haven't already: Runpod 

Click on Pod (on the left side) then click deploy

Choose a GPU with at least 24gb of VRAM (4090 is best), click here for the comfyui template (aitrepreneur/comfyui), then edit the template and choose 80gb for both the container and volume disk, then deploy on demand

Go to my pods, wait for everything to finish and then click "connect", then "Connect to HTTP SERVICE port 3000" and click on the manager to update comfyui and restart it

Then go to the port 8888, go inside the comfyui folder and then drag and drop the Z-IMAGE_BASE_ULTRA-AUTO_INSTALL-RUNPOD-V2.sh file on the left side of the UI then click on the "Terminal" icon on the right side on the UI

Copy and paste these two lines then press enter:

chmod +x Z-IMAGE_BASE_ULTRA-AUTO_INSTALL-RUNPOD-V2.sh

./Z-IMAGE_BASE_ULTRA-AUTO_INSTALL-RUNPOD-V2.sh

Wait for everything to be installed

Go back to the port 3000 and click on the manager and restart it

Load the Z-IMAGE_BASE_ULTRA_WORKFLOW-V2.json file and there you go😎

Also use the command:

tail -f comfyui.log

inside the workspace/logs to get the real time logs so that you always know when the install is done. It will then ask you to restart, so click on that restart button, wait for the webui to reboot, click refresh once it's done and there you go.

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

Omar Salleh

Excellent work, as always - thank you! Just a quick question - May I ask why you split each new thing across multiple posts, with workflow in one and installer files in another? It would be far easier to follow for new folks if it was all in one place.

nando Pluister

Your work is always amazing, and this is the first time I got an error. I have tried to update the guff node, but that did not fix the issue. I get a GGUF magic invalid when comfy tries to load the base model (z-image-Q6_k.gguf File "C:\Users\****\Documents\comfyui\ComfyUI_windows_portable\python_embeded\Lib\site-packages\gguf\gguf_reader.py", line 138, in __init__ raise ValueError('GGUF magic invalid') ValueError: GGUF magic invalid

nando Pluister

downloading the q6 model from hugging face fixed the issue, maybe the download is corrupt?

Darkcide

When running the error "CLIPLoaderGGUF Unexpected text model architecture type in GGUF file: 'qwen3'" pops up, any insight to what might cause this? I'm stumped

Christopher Carnes

I do believe you are right. I had the same issue and that fixed it for me. Thanks!

Goblin

Is there a way to change the output filepath?

Charlie T.

AMAZING WORFLOW !!!!

Darren Keathley

I have not been able to install stuff lately, If you see this and ComfyUI did not start try updating your Nvidia Drivers to the latest. If you get a c10.dll error you need to install vc redist that you can find: https://aka.ms/vc14/vc_redist.x64.exe C:\Users\User\Downloads\ComfyUI_windows_portable>pause Press any key to continue . . .

George Poulopoulos

Hello sir, when i try image to image it never use the photo i upload, it create one conpletly new. am i doing something wrong?

Francesc

Same problem. Thanks nando. Your solution also worked for me.

SavageActual
>>201630963

yeah i get this issue too, no idea how to fix this EDIT: reinstalled and it worked *shrug*

Anime Downloader

In 2026, Windows Defender is the only antivirus you need. Any other paid antivirus are just bloatware. So if Windows Defender is not flagging it, it's safe. Beside, you can open the bat files with a text editor, and examine the code yourself. Better yet, copy and paste all the codes into ChatGPT and ask it to scrutinize its safety if reading code is not in your domain.

Kevin

Suggestion to add these into the first few lines, to avoid any directory issues. I was using the nodes install since I already done the full one-click installer before. Cheers! @echo off cd /d "%~dp0" setlocal EnableExtensions EnableDelayedExpansion chcp 65001 >nul

Ken L. Black

Bless you, sir....you come through once again! I may actually try making a LoRA for the first time. Will your previously-done run-down using Z-Image Turbo work as well for Z-Image? (I'm also more interested in a an, ummmm, NSFW man-woman POSE LoRA rather than the usual character LoRAs you do. Not sure if the instructions are still the same.)

Paul Mazurkiewicz

Does anyone have a good reference for Zimage base prompting that has a list of the styles that can be called out explicitly?

Paul Mazurkiewicz

I just want to thank Aitrepreneur for this incredible workflow. Dude, you are the best and deserve the support of this community. This is a real banger. Best workflow in the world right now for Zimage base. The idea of using turbo to refine it is just brilliant. Keep up the amazing work!!

Albion22

yes the model doesn't download correctly. Mine was only 1KB. Download it from huggin face and replace it. Then it will work

Albion22

check the file type at the clip loader. It shouldn't be qwen but lumina2. If it still doesn't work make a clean installation of comfyui because sometimes when you try to reuse older ones it messes things up.

Andre Saddler

why do these install to system32 folder and Prompt execution failed Prompt outputs failed validation: VAELoader: - Value not in list: vae_name: 'qwen_image_vae.safetensors' not in ['ae.safetensors', 'wan_2.1_vae.safetensors', 'taesd', 'taesdxl', 'taesd3', 'taef1', 'pixel_space'] CLIPLoaderGGUF: - Value not in list: clip_name: 'Qwen2.5-VL-7B-Instruct-UD-Q4_K_S.gguf' not in ['Qwen3-4B-UD-Q6_K_XL.gguf'] UnetLoaderGGUF: - Value not in list: unet_name: 'Qwen_Image_Distill-Q4_K_S.gguf' not in ['z_image-Q6_K.gguf', 'z_image_turbo-Q6_K.gguf']

Darkcide

I finally figured out what the issue was, I had to do a full reinstall. There was something with the custom nodes that was messed up

neil

I installed this but when i click run it says prompt has no outputs

Ken L. Black

Andre, while waiting for Aitrepreneur to answer you, you may want to try moving the entire ComfyUI_windows_portable folder from System32 to anywhere else on your drive. I have mine in the root C: drive. The joy of the ComfyUI_windows_portable version of ComfyUI is that it is just that: portable. You can move it anywhere you want without it affecting anything else. It's POSSIBLE that since System32 is a, well...system folder, that this may be the cause of the error messages. (You might also try right-clicking the .bat file to start ComfyUI and then RUN AS ADMINISTRATOR.) But really, I would just encourage you to move it elsewhere on your hard drive.

Alexreal

Z-IMAGE_BASE_ULTRA-COMFYUI-MANAGER_AUTO_INSTALL It doesn't work during installation and returns an error.

Dirk

You have to activate a least one group

Andre Saddler

already re-coded the bat to avoid system directories

Dirk

So, for me it technically works, but the anatomy is really really bad. A closeup of a hand results in molten uneven fingers, that's much worse than z-image turbo (or any other model I tried so far). Strange...

Hpz937

z_image-Q6_K doesn't exist in HF repo error: 'Entry not found'

llee8820

How would I add Z image base to AI toolkit? Rerun the auto install? I want to train Lora with base in AI toolkit

James Stewart

Looking forward to next vids/updates with Controlnet & LoRA training. I assume you can take the images created and use in the LTX-2 workflow if you want a video created?

lostinthewash

Sorry if I missed this, but how does one use the Z-Image Inpainting flow in the master workflow provided? Are you not able to select or paint the area of interest with that flow? Is it purely text prompt based?

Replicant

Is it me or does the nodes installer not install the base model. I only see Turbo.

Silb3erbart 1

doesnt work for me. Any ideas? (even reinstalled from scratch) ----------------- UnetLoaderGGUF GGUF magic invalid # ComfyUI Error Report ## Error Details - **Node ID:** 559 - **Node Type:** UnetLoaderGGUF - **Exception Type:** ValueError - **Exception Message:** GGUF magic invalid ## Stack Trace ``` File "C:\Daten_C\progz\Comfy_AIT\ComfyUI_windows_portable\ComfyUI\execution.py", line 518, in execute output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Daten_C\progz\Comfy_AIT\ComfyUI_windows_portable\ComfyUI\execution.py", line 329, in get_output_data return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Daten_C\progz\Comfy_AIT\ComfyUI_windows_portable\ComfyUI\execution.py", line 303, in _async_map_node_over_list await process_inputs(input_dict, i)

Aldier Green

Instaled it from scratch, twice, but it keeps giving me this error. I guess I'll have to download the models? where, how? Is it something else completely different? I don't know much about this kind of thing... :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: Prompt outputs failed validation: UnetLoaderGGUF: - Value not in list: unet_name: 'z_image_turbo-Q8_0.gguf' not in ['z_image-Q6_K.gguf', 'z_image_turbo-Q6_K.gguf'] UnetLoaderGGUF: - Value not in list: unet_name: 'z_image-Q8_0.gguf' not in ['z_image-Q6_K.gguf', 'z_image_turbo-Q6_K.gguf']

Tark Morris

Im confused it works fine and everything but i've been downloading z-image loras from CivitAI and putting them in the lora folder... when generating an image with a lora it messes up?

Mark Sykes

I am just getting GGUF magic invalid no matter what I try

Blackened

To the people having trouble with GGUF magic invalid, you'll need to download the regular z_image-Q6_K.gguf manually, it does not exist on aitrepreneur's huggin face repo. THere is Q4, Q5 and Q8 there, but Q6 is not. Either grab another model or find it on another huggin face repo. Get https://huggingface.co/Aitrepreneur/FLX/blob/main/z_image-Q5_K_S.gguf or https://huggingface.co/Aitrepreneur/FLX/blob/main/z_image-Q8_0.gguf and put it in your installation folder/comfyui/models/unet

JAB

The Z-IMAGE ULTRA COMBO IMAGE TO IMAGE part isn't working for me. It completely ignored the input image and just creates a new pic from the prompt. Anyone else have this problem?

JAB

Denoise is set to 1.0 in the workflow, Just noticed. Set it back to 0.6. All good

Swatman

I noticed that the Z-IMAGE ULTRA COMBO TEXT TO IMAGE does not save the prompt in the image. Is there another way to save the prompt?

Werner Römling

is there a way to include an AMD version?

BigOwl

You are probably using Z-Image Turbo loras, which are incompatible with Z-Image Base.

Dmoe2600

it says i'm missing nodes FastFilmGrain and FastLapLacianSharpen, It's on Rempod

Juan Zombie

Yes, works better with Base model than Turbo model

Rick Doman

When i use Z-Image combo controlnet and depth anything like the video, it makes a black and white depth image, but it never creates a new image based on my prompt?. it just stops at the depth image

AR reality

Do you have a link for the Q6 model on HF?

Rodger Hurlburt

With the base model you need to only use lora's that were trained on base. Any lora trained on base will also work with turbo

Rodger Hurlburt

These never work for me because I install my comfyui portable in drive I I don't want this cluttering up my C drive

Complex Avocado

I need my subscription canceled please help. Ive resubscribed on a new account and i’d like to keep subscribing however, i lost my access to my old account 2 months ago, I cant find another way to reach out to you personally. Previous account was Hannaali**995@gmail.com card ending in 3083

Neo Hebrea

how do I install it on Mac

Matteo Moramarco

As far as the use of nsfw loras is concerned, both for turbo and less per base, is there any sense in using it only with loras for turbo?

Bob

please update your wan i2V workflows. comfyui is updated and all your workflow mostly missing nodes or juts dont have no more right nodes. but all the workflows i find on internet is way too complicated! please make new updated version and as simple as you always make.

Noel Holmes

For the Runpod step, after the script runs it says to "update comfyui" - does anyone know if this means specifically just update comfui or also "update all"?

Justin

Hey is there a way to turn a video clip into an animation with any of the video in this group? Ideally being able to turn longer from into animation

Yukon Cornelius

Does anyone else find that the images this produces seem to be inferior in many ways to Zit turbo? idk. its weird. after playing around quite a bit it just seems like the images always turn out grainy and strange, even when you turn the grain off, or tweak it... idk.. just seems less... "perfect" . i guess in some ways that makes it realistic, but in other ways it just feels like the images are coming out "dirty" idk.

Whistler

I think the combo workflow gives pretty good images without loras and stuff. But the base model is suppose to give lower quality images, it is in the dev description. But it is more advanced in what images it can do and it is more open to be build on as a base so to say.

J R

Perfect worked well!

XxWalla_WallaxX

If your having trouble and getting errors with FastFilmGrain and FastLaplacianSharpen I have found a fix. First go to the folder using the path: "ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-vrgamedevgirl" -then open the requirements.txt file with notepad -delete all text inside and replace with: "kornia librosa imageio torchcodec" -then save and close the file. -Then go to the folder path: "ComfyUI_windows_portable\ComfyUI\custom_nodes" and open the windows command prompt by typing cmd in the file explorer address bar. -Once open type "..\..\python_embeded\python.exe -m pip install -r comfyui-vrgamedevgirl\requirements.txt" and hit enter to run the command. Problem should be fixed after that.