1-Click CogVideoX1.5-5B-I2V Image to Video Installers For Windows, RunPod, Massed Compute - SOTA Open Source Model
Added 2025-10-15 12:01:00 +0000 UTC1-Click to install and use CogVideoX-5B SOTA image to guided video generation model on your PC locally or on RunPod and Massed Compute
Patreon exclusive posts index to find our scripts easily, Patreon scripts updates history to see which updates arrived to which scripts and amazing Patreon special generative scripts list that you can use in any of your task.
Join discord to get help, chat, discuss and also tell me your discord username to get your special rank : SECourses Discord
Please also Star, Watch and Fork our Stable Diffusion & Generative AI GitHub repository and join our Reddit subreddit and follow me on LinkedIn (my real profile)
=======
Latest zip file : CogVideoX_v5.zip
CPU offloading has to be used until Diffusers library fixes it
If your model downloads fail or stuck, edit (use any text editor) Windows_Start_APP.bat and remove set HF_HUB_ENABLE_HF_TRANSFER=1
This is used to Speed up download but causes error at some systems randomly
I recommend Wan 2.2 Image to Video nowadays instead of this model : https://youtu.be/3BFDcO2Ysu4
15 October 2025 Update
Since we keep our apps working I have updated it to use latest Torch 2.8, CUDA 12.9, and other libraries
Now fully supports RTX 5000 series as well as older GPUs like RTX 2000 3000 4000 series etc and also cloud GPUs like A100, B200 etc
I recommend a fresh installation
Also now it will download models from Hugging Face into CogVideo\models folder in your installation folder
I recommend Wan 2.2 Image to Video nowadays instead of this model : https://youtu.be/3BFDcO2Ysu4
25 December 2024 Update
Full YouTube tutorial published : https://youtu.be/5UCkMzP2VLE
If you also upvote and leave a comment to this Reddit Thread I appreciate
Windows Requirements
Python 3.10, FFmpeg, CUDA 12.9, C++ Tools, MSVC, Git
If it doesn't work make sure to follow below tutorial step by step and install everything exactly as shown in this below tutorial
General Information
1-Click installers made for RunPod, Massed Compute and Windows
This is a Gradio APP developed by me from scratch
DeepSpeed, torchao and Triton all automatically installed and working
All generated videos will be saved inside outputs folder
Official Repo : https://github.com/THUDM/CogVideo
Used model : https://huggingface.co/THUDM/CogVideoX1.5-5B-I2V
Supports 5 and 10 second videos at 16 FPS = 81, 161 frames but you can use any amount of frame number as well like 41, 25, etc (+1 coming from input image i think)
Supports all resolutions divisible to 16 but recommend are between 768 and 1360
Image to video works
Supports almost all resolutions that are divisible to 16
I have tested the following resolutions and here their VRAM requirements - may work on lower VRAM GPUs too but slower
512x288 - 41 frames : 7700 MB , 576x320 - 41 frames : 7900 MB
576x320 - 81 frames : 8850 MB , 704x384 - 81 frames : 8950 MB
768x432 - 81 frames : 10600 MB , 896x496 - 81 frames : 12050 MB
896x496 - 81 frames : 12050 MB , 960x528 - 81 frames : 12850 MB
1024x576 - 81 frames : 13900 MB , 1280x720- 81 frames : 17950 MB
1360x768 - 81 frames : 19000 MB
How To Install
Use Windows_Install.bat file
When you first time generating a video it will download CogVideo files into Hugging Face cache folder
Massed Compute (Recommend Cloud) :
Please register via this link : https://vm.massedcompute.com/signup?linkId=lp_034338&sourceId=secourses&tenantId=massed-compute
We have a special coupon for all GPUs : SECourses
If you want to learn more about GPUs and prices read this link : https://www.patreon.com/posts/126671823
Select RTX A6000 or Better GPU - like L40S or A6000 ADA or A100 or H100 or best is RTX 6000 PRO
Then select our image SECourses from Creator dropdown
Then follow Massed_Compute_Instructions_READ.txt
Same as my any other Massed Compute installer script
Example tutorial for learn how to install and use Massed Compute
(Starts at 12:58) : https://youtu.be/KW-MHmoNcqo?si=G1WbG-Qw4ujWvOtG&t=778
RunPod (Cloud):
Please register via this link : https://get.runpod.io/955rkuppqv4h
Then follow Runpod_Instructions_READ.txt
Same as my any other RunPod installer script
Use the template written in Runpod_Instructions_READ.txt file
Example tutorial for learn how to install and use RunPod
(starts at 22:03) : https://youtu.be/KW-MHmoNcqo?si=QN8X8Sjn13ZYu-EU&t=1323
Demo and VRAM
Download attached vram_usages .zip and demo_material .zip for more info
Comments
Wan 2.2 of course
Furkan Gözükara
2025-10-29 08:30:52 +0000 UTCCompared to Wan 2.2 and CogVideo, which model generates video better?
Hoàng Giang Sơn Trương
2025-10-29 08:04:07 +0000 UTCyes installer was outdated. just updated with v5 and fixed please read changelogs at the top
Furkan Gözükara
2025-10-15 14:36:06 +0000 UTCHello, I followed this video as well as the other video to install the other programs needed to run Open Source programs. I got this error when I tried to run CogVideo. I am totally new to these things so maybe I missed a step. Any advice is appreciated , thank you! Skipping import of cpp extensions due to incompatible torch version 2.5.0+cu124 for torchao version 0.14.0 Please see GitHub issue #2919 for more info Traceback (most recent call last): File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\utils\import_utils.py", line 990, in _get_module return importlib.import_module("." + module_name, self.__name__) File "C:\Python310\lib\importlib\__init__.py", line 126, in import_module return _bootstrap._gcd_import(name[level:], package, level) File "", line 1050, in _gcd_import File "", line 1027, in _find_and_load File "", line 992, in _find_and_load_unlocked File "", line 241, in _call_with_frames_removed File "", line 1050, in _gcd_import File "", line 1027, in _find_and_load File "", line 1006, in _find_and_load_unlocked File "", line 688, in _load_unlocked File "", line 883, in exec_module File "", line 241, in _call_with_frames_removed File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\models\autoencoders\__init__.py", line 1, in from .autoencoder_asym_kl import AsymmetricAutoencoderKL File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\models\autoencoders\autoencoder_asym_kl.py", line 22, in from ..modeling_utils import ModelMixin File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\models\modeling_utils.py", line 41, in from ..quantizers import DiffusersAutoQuantizer, DiffusersQuantizer File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\quantizers\__init__.py", line 16, in from .auto import DiffusersAutoQuantizer File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\quantizers\auto.py", line 22, in from .bitsandbytes import BnB4BitDiffusersQuantizer, BnB8BitDiffusersQuantizer File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\quantizers\bitsandbytes\__init__.py", line 1, in from .bnb_quantizer import BnB4BitDiffusersQuantizer, BnB8BitDiffusersQuantizer File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\quantizers\bitsandbytes\bnb_quantizer.py", line 22, in from ..base import DiffusersQuantizer File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\quantizers\base.py", line 24, in from .quantization_config import QuantizationConfigMixin File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\quantizers\quantization_config.py", line 55, in from torchao.quantization.quant_primitives import MappingType File "C:\Cogvideo\CogVideo\venv\lib\site-packages\torchao\__init__.py", line 100, in from torchao.quantization import ( File "C:\Cogvideo\CogVideo\venv\lib\site-packages\torchao\quantization\__init__.py", line 6, in from .autoquant import ( File "C:\Cogvideo\CogVideo\venv\lib\site-packages\torchao\quantization\autoquant.py", line 11, in from torchao.dtypes import ( File "C:\Cogvideo\CogVideo\venv\lib\site-packages\torchao\dtypes\__init__.py", line 1, in from . import affine_quantized_tensor_ops File "C:\Cogvideo\CogVideo\venv\lib\site-packages\torchao\dtypes\affine_quantized_tensor_ops.py", line 11, in from torchao.dtypes.affine_quantized_tensor import ( File "C:\Cogvideo\CogVideo\venv\lib\site-packages\torchao\dtypes\affine_quantized_tensor.py", line 17, in from torchao.quantization.quant_primitives import ( File "C:\Cogvideo\CogVideo\venv\lib\site-packages\torchao\quantization\quant_primitives.py", line 175, in torch.int1: (-(2**0), 2**0 - 1), File "C:\Cogvideo\CogVideo\venv\lib\site-packages\torch\__init__.py", line 2563, in __getattr__ raise AttributeError(f"module '{__name__}' has no attribute '{name}'") AttributeError: module 'torch' has no attribute 'int1' The above exception was the direct cause of the following exception: Traceback (most recent call last): File "C:\Cogvideo\CogVideo\app.py", line 5, in from diffusers import AutoencoderKLCogVideoX, CogVideoXTransformer3DModel, CogVideoXImageToVideoPipeline File "", line 1075, in _handle_fromlist File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\utils\import_utils.py", line 981, in __getattr__ value = getattr(module, name) File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\utils\import_utils.py", line 980, in __getattr__ module = self._get_module(self._class_to_module[name]) File "C:\Cogvideo\CogVideo\venv\lib\site-packages\diffusers\utils\import_utils.py", line 992, in _get_module raise RuntimeError( RuntimeError: Failed to import diffusers.models.autoencoders.autoencoder_kl_cogvideox because of the following error (look up to see its traceback): module 'torch' has no attribute 'int1'
Dennis Alcordo
2025-10-14 21:34:12 +0000 UTCHunyuan i think supports but i havent yet. also wait me to publish Wan 2.1 tutorial it is even better
Furkan Gözükara
2025-02-25 21:12:18 +0000 UTCThis is the best one of the video generators I have used so far. Do you know of any video generator that support Loras?
Hockey
2025-02-25 13:48:44 +0000 UTCare you using gradio live? sometimes it can happen. you need to download from outputs folder. the video should be generated.
Furkan Gözükara
2025-01-08 00:34:44 +0000 UTCHello. Runpod setup worked but generating just gives a generic "connection errored out" message.
Skyler
2025-01-07 22:06:19 +0000 UTChello. i dont know which library requesting sadly :(
Furkan Gözükara
2025-01-02 20:21:21 +0000 UTCHi all works well when I am connected to internet. When I want to generate video while beeing offline CogVideo is giving me error not possible to establish connection after pressing generate video. Is there way to generate video while beeing offline ? All models are downloaded so no need to be connected in my point of view. Thanks
Zd Ba
2025-01-02 15:40:27 +0000 UTCit is not frozen. it is trying to load. how much RAM you have? also your harddisk is very slow? or perhaps corrupted? did you enable virtual RAM?
Furkan Gözükara
2024-12-29 23:38:27 +0000 UTCThanks for your hard work. I have an issue when I've installed Cog on Win 11. After launching the Gradio app and generation my first image, it freezes during installing the models required: * Running on local URL: http://127.0.0.1:7860 To create a public link, set `share=True` in `launch()`. Downloading shards: 0%| | 0/4 [00:00
David
2024-12-29 22:54:49 +0000 UTChi please follow this video every step exactly : https://youtu.be/DrhUHnYfwC0
Furkan Gözükara
2024-12-29 21:21:56 +0000 UTCHi i have this warning WARNING: Failed to find MSVC. INFORMACIÓN: no se pudo encontrar ningún archivo para los patrones dados. * Running on local URL: http://127.0.0.1:7860 how can i fix them
FabiMath
2024-12-29 20:53:12 +0000 UTCGreat installer thank you for the work. success at first try but in miniconda for P3.11 and P3.10 switching
MEGA AMIGA
2024-12-28 12:55:09 +0000 UTCMMAudio is separate app here : https://www.patreon.com/posts/117990364 will make a tutorial hopefully soon
Furkan Gözükara
2024-12-25 14:30:23 +0000 UTCI am using the latest 1.5-5B-I2V Image to Video app, I thought it is including the MMAudio. Am I misunderstand?
Sugar Coat VFX Design
2024-12-25 13:19:19 +0000 UTC2 : ye that is weird. are you using text to sound or video to sound? did you try our demo images videos?
Furkan Gözükara
2024-12-25 12:37:39 +0000 UTC1: Got it. It will be great if you can implement later - batch generation based on a folder of images. 2: yes, all silent. Is it related to the prompt?
Sugar Coat VFX Design
2024-12-25 11:59:49 +0000 UTC1 : we cant batch process yet. but you can generate more than 1 video of the input image if you mean that with different seeds. it is at the bottom of the input image 2 : MMAudio generated audio in my all tests. all videos of yours were silent?
Furkan Gözükara
2024-12-25 10:10:05 +0000 UTCThank you for the great app~!I have two questions: 1. How to enable batch processing? 2. Any control of MMAudio? I did few test the generated video is silent. Merry X'mas~!
Sugar Coat VFX Design
2024-12-25 05:31:11 +0000 UTCgreat ty
Furkan Gözükara
2024-12-23 23:18:47 +0000 UTCI did fix the issues thank you. I wrote all about it in the other comment string. Thanks for help. Great service! A worthwhile Patreon subscription.
DanO..
2024-12-23 22:38:07 +0000 UTChi did you fix your issues? if you upgrade gold tier i can connect your pc and setup for you and you can learn as well
Furkan Gözükara
2024-12-23 13:33:14 +0000 UTCWhat I'm getting is horrible. If you have any tips or pointers please let us know at some time. It appears you need to describe EVERYTHING or it becomes a mess quickly. Thanks again for helping me run it. I really appreciate it.
DanO..
2024-12-21 00:30:54 +0000 UTCThat fixed it! Self inflicted but in my defense I can't be the only person who operates like I do. I never update anything until something doesn't work. :) I've been on personal computers since the late 80s and learned the hard way not to update things until you have to. So, like normal, I used your 1-click install. It didn't work and THEN I set about updating CUDA and Python. Normally, that would then fix it. But since they were necessary in the install process it didn't install correctly. I'm letting you know all this because it will certainly come up again. Thanks.
DanO..
2024-12-21 00:02:22 +0000 UTCit auto creates the venv and install everything. did you follow this video step by step before installing? : https://youtu.be/DrhUHnYfwC0 make a fresh install and email me install logs : monstermmorpg@gmail.com
Furkan Gözükara
2024-12-20 21:34:39 +0000 UTCCorrect. It created the "CogVideo" directory and all subdirectories. Also, update bat says it's up to date. Should I create a venv in CogVideo? Something like that. Is the other thing a Hugging Face call?
DanO..
2024-12-20 21:33:16 +0000 UTCyou are executing the install bat file first right?
Furkan Gözükara
2024-12-20 21:29:40 +0000 UTCGreat trouble shooting advice! "The system cannot find the path specified." Is the error.
DanO..
2024-12-20 21:08:52 +0000 UTChi open a cmd in that folder. then type bat file name and enter. that way cmd wont close so we can see the error
Furkan Gözükara
2024-12-20 21:03:58 +0000 UTCHi. I've been trying to get this to work. When I try to run it a CMD window pops up for a fraction of a second and disappears. I'm using Python 3.11 and CUDA 12.4. Any ideas? Windows 10.
DanO..
2024-12-20 20:31:47 +0000 UTCThanks Doc
steven losch
2024-12-12 16:45:13 +0000 UTCUpdated to next level version : https://www.patreon.com/posts/112848192
Furkan Gözükara
2024-12-12 16:21:38 +0000 UTCchecking right now
Furkan Gözükara
2024-12-11 22:00:15 +0000 UTCRunning the app I get: Traceback (most recent call last): File "/workspace/ImageToVideoAI/app.py", line 27, in import moviepy.editor as mp ModuleNotFoundError: No module named 'moviepy.editor' After commenting out that line the app will launch. But now after generating a video in the Gradio interface the gif display of the video shows an error: Traceback (most recent call last): File "/workspace/ImageToVideoAI/venv/lib/python3.10/site-packages/gradio/queueing.py", line 624, in process_events response = await route_utils.call_process_api( File "/workspace/ImageToVideoAI/venv/lib/python3.10/site-packages/gradio/route_utils.py", line 323, in call_process_api output = await app.get_blocks().process_api( File "/workspace/ImageToVideoAI/venv/lib/python3.10/site-packages/gradio/blocks.py", line 2043, in process_api result = await self.call_function( File "/workspace/ImageToVideoAI/venv/lib/python3.10/site-packages/gradio/blocks.py", line 1590, in call_function prediction = await anyio.to_thread.run_sync( # type: ignore File "/workspace/ImageToVideoAI/venv/lib/python3.10/site-packages/anyio/to_thread.py", line 56, in run_sync return await get_async_backend().run_sync_in_worker_thread( File "/workspace/ImageToVideoAI/venv/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 2505, in run_sync_in_worker_thread return await future File "/workspace/ImageToVideoAI/venv/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 1005, in run result = context.run(func, args) File "/workspace/ImageToVideoAI/venv/lib/python3.10/site-packages/gradio/utils.py", line 865, in wrapper response = f(args, kwargs) File "/workspace/ImageToVideoAI/venv/lib/python3.10/site-packages/gradio/utils.py", line 865, in wrapper response = f(*args, kwargs) File "/workspace/ImageToVideoAI/app.py", line 381, in generate clip = mp.VideoFileClip(video_path) NameError: name 'mp' is not defined
steven losch
2024-12-11 21:38:15 +0000 UTCWell I am following developments. Nothing better yet. There is 1 new text to video model but it is heavy. Also text to video not very useful we need image to video and video to video
Furkan Gözükara
2024-10-26 15:36:19 +0000 UTCAny updates on your reported issues? Think they are still marked as open, but have they been fixed? CogVideo is dead easy to use, but it's a pity that it is limited to 6s and 8fps. Do you know of anything comparable that is a little more flexible in this?
Bjoern
2024-10-26 13:49:55 +0000 UTCyou were right just made update and fixed
Furkan Gözükara
2024-09-28 12:34:33 +0000 UTCyou were right just made update and fixed
Furkan Gözükara
2024-09-28 12:34:29 +0000 UTCyou were right just made update and fixed
Furkan Gözükara
2024-09-28 12:34:26 +0000 UTCSame issue here, the problem occurs on startup, before even getting access to the gradio.
GeekZolda
2024-09-28 06:57:12 +0000 UTCmake sure image is fully loaded into gradio interface
Furkan Gözükara
2024-09-27 18:17:39 +0000 UTCTypeError: expected str, bytes or os.PathLike object, not NoneType
shen oracle
2024-09-27 15:42:30 +0000 UTCThanks. I'll do that.
cool1
2024-09-27 14:35:35 +0000 UTCWindows Requirements Python 3.10, FFmpeg, CUDA 12.4, C++ tools and Git If it doesn't work make sure to follow below tutorial step by step and install everything exactly as shown in this below tutorial https://youtu.be/-NjNy7afOQ0 More up-to-date info about CUDA 12.4 and C++ tools > https://www.patreon.com/posts/111553210 CUDA 11.8 still should work but if fails make default 12.4 as shown in above video and install
Furkan Gözükara
2024-09-27 14:33:20 +0000 UTCOkay I'll do that. I have RTX3090 that has cuda. I don't know if it's setting the right paths for cuda. I started installing into the 2nd folder. AVG again said a warning (it was about pip) then it shows "python.exe, app unable to start correctly" then the installation keeps continuing. I'll try reinstalling again or using the update version bat file.
cool1
2024-09-27 14:28:41 +0000 UTCplease install into a new folder. i just made a fresh install into a new folder i have no issues. also do you have cuda set in your system?
Furkan Gözükara
2024-09-27 14:24:50 +0000 UTCI tried deleting the installed files/folders and clicking the install again then running the start windows bat file and it says: ... backends = _discover_backends() File "G:\CogVideoX-SECourses-version\ImageToVideoAI\venv\lib\site-packages\triton\backends\__init__.py", line 43, in _discover_backends compiler = _load_module(name, os.path.join(root, name, 'compiler.py')) File "G:\CogVideoX-SECourses-version\ImageToVideoAI\venv\lib\site-packages\triton\backends\__init__.py", line 12, in _load_module spec.loader.exec_module(module) File "G:\CogVideoX-SECourses-version\ImageToVideoAI\venv\lib\site-packages\triton\backends\nvidia\compiler.py", line 3, in from triton.backends.nvidia.driver import CudaUtils File "G:\CogVideoX-SECourses-version\ImageToVideoAI\venv\lib\site-packages\triton\backends\nvidia\driver.py", line 18, in include_dir += [os.path.join(os.environ.get("CUDA_PATH"), "include")] File "C:\Program Files\Python310\lib\ntpath.py", line 104, in join path = os.fspath(path) TypeError: expected str, bytes or os.PathLike object, not NoneType Press any key to continue . . . I don't know if it's anything to do with antivirus that puts a warning up when installing one file then after a certain number of seconds says it looks okay.
cool1
2024-09-27 14:21:03 +0000 UTCdo reinstall you never need to set python path
Furkan Gözükara
2024-09-27 14:01:58 +0000 UTCI'd installed the CogStudio like I said above and that could work. When I tried installing this version (on Windows 10 Pro), since it might be faster etc, with the .bat file and then running it with Windows_Start_APP.bat I get: "ModuleNotFoundError: No module named 'triton' Traceback (most recent call last): File "G:\CogVideoX-SECourses-version\ImageToVideoAI\app.py", line 15, in from pipelines.pipeline_cogvideox import ( File "G:\CogVideoX-SECourses-version\ImageToVideoAI\pipelines\pipeline_cogvideox.py", line 24, in from diffusers.loaders import CogVideoXLoraLoaderMixin ImportError: cannot import name 'CogVideoXLoraLoaderMixin' from 'diffusers.loaders' (C:\Users\myusername\AppData\Roaming\Python\Python310\site-packages\diffusers\loaders\__init__.py) Press any key to continue . . " Is it anything to do with the Python version (that's looking at Python 310. I have 3.11 in another folder)? Do I need to set a path to a later version of Python?
cool1
2024-09-27 13:51:01 +0000 UTCimage2video doesnt have 2b it is only 5b and currently app only uses 5b. for 24 gb gpus i am working on optimization if i can achieve :)
Furkan Gözükara
2024-09-27 09:25:40 +0000 UTCI used a CogStudio install instructions and installed that. Is this basically the same version (or does it use more or less vram/different speed?). I have 24GB GPU. I tried image2video in that and it sometimes produced good video but sometimes it didn't work that good or produced a mostly still image and it took quite a long time, minutes to work (with I think the GPU sounding a lot, but I think someone also says it offloads to CPU if needed). But the image2video option didn't have the selection of the model (I think the prompt to video had the model selection including 5B and a lower one I think). Is it the same with this and if so how can you tell what model it's using in image2video (ie. if you want to make sure it's using the best model available, like the 5B one)?
cool1
2024-09-27 09:16:03 +0000 UTC
