Weird video out put with ltx 13b model

[macanhhuy]

Run any example workflow in comfyui. Only get weird output

[X1mu5]

Maybe it is because of scheduler. It gives me the same with Karras, but with linear it is OK.

[pachieh]

I got the same. I think the last time I had this issue, I had to ensure I had the latest nightly build of Pytorch running.

Check out this thread... https://github.com/Lightricks/ComfyUI-LTXVideo/issues/43#issuecomment-2555502825

On MacOS, M3 Max w/36GB RAM, Pytorch 2.6, ComfyUI.

[macgopuff]

I tried, torch 2.5.1 cuda 1124 and torch 2.7 cuda 128, same issue

[pachieh]

You try the nightly builds?

[ReinerBforartists]

Just one more voice to confirm the problem. The example workflows for the ComfyUI LTX Video nodes still uses 0.9.6, and renders fine. But with 0.9.7 i see the above result.

Toyed around with the sampler and the scheduler, no luck. The same workflow works with version 0.9.6 just fine.

Console output:

` H:\comfyui\ComfyUI_windows_portable>.\python_embeded\python.exe -s ComfyUI\main.py --windows-standalone-build [START] Security scan [DONE] Security scan

ComfyUI-Manager: installing dependencies done.

** ComfyUI startup time: 2025-05-14 12:31:26.609 ** Platform: Windows ** Python version: 3.12.9 (tags/v3.12.9:fdb8142, Feb 4 2025, 15:27:58) [MSC v.1942 64 bit (AMD64)] ** Python executable: H:\comfyui\ComfyUI_windows_portable\python_embeded\python.exe ** ComfyUI Path: H:\comfyui\ComfyUI_windows_portable\ComfyUI ** ComfyUI Base Folder Path: H:\comfyui\ComfyUI_windows_portable\ComfyUI ** User directory: H:\comfyui\ComfyUI_windows_portable\ComfyUI\user ** ComfyUI-Manager config path: H:\comfyui\ComfyUI_windows_portable\ComfyUI\user\default\ComfyUI-Manager\config.ini ** Log path: H:\comfyui\ComfyUI_windows_portable\ComfyUI\user\comfyui.log

Prestartup times for custom nodes: 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\rgthree-comfy 2.3 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Manager

Checkpoint files will always be loaded safely. Total VRAM 16380 MB, total RAM 32687 MB pytorch version: 2.6.0+cu126 Set vram state to: NORMAL_VRAM Device: cuda:0 NVIDIA GeForce RTX 4060 Ti : cudaMallocAsync Using pytorch attention Python version: 3.12.9 (tags/v3.12.9:fdb8142, Feb 4 2025, 15:27:58) [MSC v.1942 64 bit (AMD64)] ComfyUI version: 0.3.34 ComfyUI frontend version: 1.19.9 [Prompt Server] web root: H:\comfyui\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfyui_frontend_package\static NumExpr defaulting to 16 threads. [H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfy-mtb] | INFO -> loaded 103 nodes successfuly [H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfy-mtb] | INFO -> Some nodes (5) could not be loaded. This can be ignored, but go to http://127.0.0.1:8188/mtb if you want more information. [Crystools INFO] Crystools version: 1.22.1 [Crystools INFO] CPU: AMD Ryzen 7 5800X3D 8-Core Processor - Arch: AMD64 - OS: Windows 10 [Crystools INFO] Pynvml (Nvidia) initialized. [Crystools INFO] GPU/s: [Crystools INFO] 0) NVIDIA GeForce RTX 4060 Ti [Crystools INFO] NVIDIA Driver: 572.60 Total VRAM 16380 MB, total RAM 32687 MB pytorch version: 2.6.0+cu126 Set vram state to: NORMAL_VRAM Device: cuda:0 NVIDIA GeForce RTX 4060 Ti : cudaMallocAsync

Loading: ComfyUI-Manager (V3.32.2)

[ComfyUI-Manager] network_mode: public

ComfyUI Revision: 3461 [158419f3] *DETACHED | Released on '2025-05-12'

Python version is above 3.10, patching the collections module. H:\comfyui\ComfyUI_windows_portable\python_embeded\Lib\site-packages\transformers\models\auto\image_processing_auto.py:604: FutureWarning: The image_processor_class argument is deprecated and will be removed in v4.42. Please use slow_image_processor_class, or fast_image_processor_class instead warnings.warn( [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/github-stats.json [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/extension-node-map.json [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/model-list.json [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/alter-list.json [H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux] | INFO -> Using ckpts path: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux\ckpts [H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux] | INFO -> Using symlinks: False [H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux] | INFO -> Using ort providers: ['CUDAExecutionProvider', 'DirectMLExecutionProvider', 'OpenVINOExecutionProvider', 'ROCMExecutionProvider', 'CPUExecutionProvider', 'CoreMLExecutionProvider'] DWPose: Onnxruntime with acceleration providers detected FizzleDorf Custom Nodes: Loaded Efficiency Nodes: Attempting to add Control Net options to the 'HiRes-Fix Script' Node (comfyui_controlnet_aux add-on)...Success! 2025-05-14 12:31:33.395 | INFO | cozy_comfyui.node:loader:121 - JOV_CAPTURE 4 nodes loaded 2025-05-14 12:31:33.395 | INFO | cozy_comfyui.node:loader:121 - JOV_CAPTURE 4 nodes loaded 2025-05-14 12:31:33.397 | INFO | cozy_comfyui.node:loader:121 - JOV_CAPTURE 4 nodes loaded 2025-05-14 12:31:33.397 | INFO | cozy_comfyui.node:loader:121 - JOV_CAPTURE 4 nodes loaded

[rgthree-comfy] Loaded 42 exciting nodes. 🎉

WAS Node Suite: OpenCV Python FFMPEG support is enabled WAS Node Suite Warning: ffmpeg_bin_path is not set in H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\was-node-suite-comfyui\was_suite_config.json config file. Will attempt to use system ffmpeg binaries if available. WAS Node Suite: Finished. Loaded 220 nodes successfully.

    "Success is not the key to happiness. Happiness is the key to success." - Albert Schweitzer

Import times for custom nodes: 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\websocket_image_save.py 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\cg-use-everywhere 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\lora-info 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_ipadapter_plus 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_fizznodes 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_essentials 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfy-image-saver 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-custom-scripts 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\mikey_nodes 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_UltimateSDUpscale 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-frame-interpolation 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-videodircombiner 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-various 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\derfuu_comfyui_moddednodes 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-dream-video-batches 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-advanced-controlnet 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\rgthree-comfy 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Florence2 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-kjnodes 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\efficiency-nodes-comfyui 0.0 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux 0.1 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Crystools 0.1 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-LTXVideo 0.1 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-videohelpersuite 0.1 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-fluxpromptenhancer 0.1 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-AnimateDiff-Evolved 0.1 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\jovi_capture 0.2 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-detail-daemon 0.3 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Manager 0.4 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_caption_this 0.4 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfy-mtb 0.9 seconds: H:\comfyui\ComfyUI_windows_portable\ComfyUI\custom_nodes\was-node-suite-comfyui

Starting server

To see the GUI go to: http://127.0.0.1:8188 FETCH ComfyRegistry Data: 5/85 F`

[CoderRamy]

i have the same issue

[ReinerBforartists]

Just to report back, Affected is the fp8 version of the weight. The full dev version produces good results.

But when i load the fp8 version ...

I have even redownloaded the fp8 one, because there was a reupload before three days, but it has the same size, and produces the same result.

[michaellightricks]

please download the latest version of fp8 model, latest model can now run natively in ComfyUI without using q8_kernels and Q8 Patch node.

[michaellightricks]

Make sure to update to latest comfy and ComfyUI-LTXVideo as well

[ameen-roayan]

@michaellightricks Everything is updated and I also still get the same issues, and running the FP8 patched causes comfyui to crash with no errors and just show prompt paused and close.

4090/windows/Python version: 3.11.6/pytorch version: 2.7.0+cu128/xformers version: 0.0.30

Latest update just moments ago:

• with the Q8 Patch regardless of enable/disable states, it crashes.
• Bypassing the Q8 Patch this is what happens

image immediately becomes super cloudy, just the first frame works :

https://github.com/user-attachments/assets/a5a34a86-47d1-425a-8cfc-9ce2e100d9e8

and this is following the workflow : https://github.com/Lightricks/ComfyUI-LTXVideo/blob/master/example_workflows/13b-distilled/ltxv-13b-dist-i2v-base-fp8.json

Tried everything in the other threads, including trying all the current available variants of the fp8 model.

the e4m3fn: https://github.com/user-attachments/assets/8ff87dbe-aaa5-4c75-a8a6-2123daabcaa4

the regular: https://github.com/user-attachments/assets/78943d17-46a7-41fc-9ad8-d5fd9e1464cf

the dev: https://github.com/user-attachments/assets/9c2a6486-d711-4532-969e-e57b845f310e

[ReinerBforartists]

I get an error message now after updating all. Same default workflow as above. And the fresh downloaded ltxv-13b-0.9.7-dev-fp8.safetensors

As a note, this default workflow still loads with version 0.9.6 first.

[ameen-roayan]

@michaellightricks Halp