Loading
Vous avez un workflow ComfyUI ? Obtenez une documentation lisible directement dans votre navigateur.Rendre un workflow lisible.Essayer maintenant ↗
This is version 4 for LTX Video 0.9.8.
Profil d'exécution
Description de la source
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:
Commentaire généré par l’IA
Explication générée par l’IA à partir des détails de la source et de la configuration. Les suggestions sont clairement signalées.
Use LTXvideo Image to Video refined in ComfyUI to turn an input image and prompt into video, generating around 480p before upscaling it to 720p.
It includes a keyframe workflow and a humanized Multiprompt workflow; both include internal upscaling with LTX nodes.
The keyframe option is based on LTX Video’s keyframe example, while the Multiprompt option follows the workflow on LTX Video’s repository page.
Drag an image into an image node, adjust the prompt, and press queue.
Suggestion
The stated minimum is 12 GB of VRAM and at least 32 GB of system RAM; the workflow was generated with 16 GB of VRAM.
It requires these node packs: ComfyUI-Image-Filters, comfyui-kjnodes, ComfyUI-LTXVideo, and comfyui-videohelpersuite.
Before running it, verify that these listed model files are available: ltxv-13b-0.9.8-distilled-fp8.safetensors, ltxv-spatial-upscaler-0.9.8.safetensors, and t5xxl_fp8_e4m3fn_scaled.safetensors.
Suggestion
This is version 4 for LTX Video 0.9.8.
The description says LTX-Video 0.9.8 supports up to 60 seconds of video and is compatible with official IC-LoRAs.
Test your chosen input image before relying on it; the description reports 19 unusable results from 20 attempts with one image.
Suggestion · non vérifié
In the reported examples, the Multiprompt workflow took about twice as long as the keyframe workflow, and its upscaling reached an out-of-memory error after about an hour.
Faut-il modifier cela ?
Connectez-vous pour envoyer une demande de modification.
Sources
1 sourceExtraits de sources
1 extraitSource 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.
Remaining hardware or download-size estimates are not available yet.
Exigences
ltxv-13b-0.9.8-distilled-fp8.safetensors
Non résoluCheckpoint · Unknown
ltxv-spatial-upscaler-0.9.8.safetensors
Non résoluUpscaler · Unknown
Text encoder · 4.80 GB · SAFETENSORS · fp8 · pruned · Unknown
Pack de nœud · Registry
Pack de nœud · Registry
ComfyUI-LTXVideo
TrouvéPack de nœud · Registry
comfyui-videohelpersuite
TrouvéPack de nœud · Registry
