Have a ComfyUI workflow? Get a readable Workflow Doc in your browser.Make a workflow readable.Try it now ↗
This is version 4 for LTX Video 0.9.8.
Runtime profile
Source description
This is version 4 for LTX Video 0.9.8.
Two workflows for the LTX Video 0.9.8 release. It comes with nice new features. Long shot generation in LTXV-13B! LTX-Video now supports up to 60 seconds of video. And is compatible with the official IC-LoRAs.
There are also new distilled models. 13B distilled model ltxv-13b-0.9.8-distilled. And 2B distilled model ltxv-2b-0.9.8-distilled Both models are distilled from the same base model ltxv-13b-0.9.8-dev and are compatible for use together in the same multiscale pipeline. Improved prompt understanding and detail generation Includes corresponding FP8 weights and workflows. Release of a new detailer model LTX-Video-ICLoRA-detailer-13B-0.9.8
Lightrix LTX Video repository:
AI-generated commentary
AI-generated explanation based on source and configuration details. Suggestions are clearly labeled.
Use the LTXvideo Image to Video refined workflow in ComfyUI to turn an input image and prompt into video. It generates roughly 480p video and can upscale the result to 720p.
It includes two variants: a keyframe workflow and an official Multiprompt workflow. Both include LTX-based upscaling.
Drag an image into the image nodes, adjust the prompt, and press queue.
The video path starts with roughly 480p generation and includes an upscaling step to 720p.
The stated minimum is 12 GB of VRAM; the workflow was generated with 16 GB of VRAM, and the notes say not to use fewer than 32 GB of system RAM.
The listed node packs are ComfyUI-LTXVideo, ComfyUI-Image-Filters, comfyui-kjnodes, and comfyui-videohelpersuite.
The listed model files are ltxv-13b-0.9.8-distilled-fp8.safetensors, ltxv-spatial-upscaler-0.9.8.safetensors, and t5xxl_fp8_e4m3fn_scaled.safetensors.
This is version 4 for LTX Video 0.9.8.
The LTX Video 0.9.8 release includes long-shot generation up to 60 seconds and compatibility with official IC-LoRAs.
ltxv-13b-0.9.8-distilled and ltxv-2b-0.9.8-distilled are compatible for use together in the same multiscale pipeline.
Before use, verify that the listed node packs and model files are available in your ComfyUI setup.
Suggestion · not verified
Input-image results may be inconsistent: 19 useless results were reported from 20 attempts with one image.
Suggestion · not verified
Does this need to be edited?
Sign in to send an edit request.
Sources
1 sourceSource excerpts
1 excerptSource context: 1926 downloads · Type Workflows · Base model LTXV
Howto
Drag an image into the image nodes, adjust the prompt, and press queue. The rest should fit.
Description
An image to video ComfyUI workflow with LTX Video. Better said, two of it. A pimped version that is based at the keyframe workflow that comes with the LTX Video node. It can be found in the examples folder of the node. This version is more compact though, better readable, with comments and downloadlinks.
And the official Multiprompt workflow from the LTX Video repository page in a humanized form. More compact, better readable, with comments and downloadlinks.
Both workflows contains an internal upscaling with LTX nodes. Means you first generate a 480p video, or something similar, and then you upscale it to 720p.
Time
The example video from the keyframe workflow rendered in around 4 minutes for the low version. Upscaling it took around 20 minutes. Which is not bad for a 720p video with lots of details then. The example video with the multiprompt took around as double as long. And i was not able to let the upscaling finish. I ran into an OOM after an hour or so. LTX Video has become a bit slower.
Requirements
This workflow was generated with 16 gb vram. Minimum requirement is 12 gb vram. And you should not have fewer than 32 gb system ram.
Note
LTX Video quality has dramatically increased with this version. And the new features are great. However, it has also become very picky with the input image. It can work. But it does not have to. I tried one image for twenty times, and got 19 useless results.
Second note, the adress for the other workflows has changed. They are now here:
Youtube Video
There is also a youtube explanation video available for this workflows.
Estimated VRAM requirement
Estimate unavailable
Model file sizes are not complete enough to calculate this requirement yet.
Requirements
7 requirementsltxv-13b-0.9.8-distilled-fp8.safetensors
Not resolvedCheckpoint · Unknown
ltxv-spatial-upscaler-0.9.8.safetensors
Not resolvedUpscaler · Unknown
t5xxl_fp8_e4m3fn_scaled.safetensors
ConflictText encoder · Unknown
Node pack · Registry
Node pack · Registry
Node pack · Registry
Node pack · Registry
