This version uses the newest ltx-2.3-22b-distilled-1.1.safetensors instead of the ltx-2.3-22b-dev-fp8.safetensors . It's heavier on VRAM but will get better results in almost the same generation time as version 1.0....
Profil d'exécution
Description de la source
LtxApp360 - i2v/t2v CUSTOM AUDIO DRIVEN 60 SECONDS VIDEO WITH 3 PROMPTS
3-Prompts / 60 Seconds Total
Each prompt = 20 seconds.
(WARNING: This workflow is super cool but a bit slow)
DIFFERENCE BETWEEN 1.0 and 1.1:
Version 1.0 uses this model: ltx-2.3-22b-dev-fp8 (28gig)
Version 1.1 uses this newer heavier model: ltx-2.3-22b-distilled-1.1 (45gig)
Tested and Working well on a 5080 with Vanilla Comfyui. (on purpose)
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 this v1.1 workflow to create a 60-second image-to-video or text-to-video sequence from three prompts and a required reference audio track; it can lip-sync the video to that audio and produces three 20-second segments.
It saves the segments in order as Ltx360_Cut1, Ltx360_Cut2, and Ltx360_Final.
Choose i2v or t2v, enter three prompts, and attach a reference audio track.
For talking or singing scenes, try describing the subject as talking or singing and transcribing the audio; this may improve results.
Suggestion · non vérifié
This v1.1 workflow is published by WanApp and uses LTXV 2.3 as its base model.
The workflow includes a simplified APP mode intended for convenient use by beginners.
The required model files are gemma_3_12B_it_fp4_mixed.safetensors, ltx-2.3-22b-dev-fp8.safetensors, ltx-2.3-22b-distilled-lora-384.safetensors, and ltx-2.3-spatial-upscaler-x2-1.1.safetensors.
The required node packs are comfyui-easy-use, comfyui-kjnodes, and comfyui-videohelpersuite.
Toggle between t2v and i2v, and between high and low quality.
The simplified mode removes manual width and height controls.
The version notes give 45 gig for v1.1 and 28 gig for v1.0; use those figures when checking available VRAM.
Suggestion · non vérifié
The description names ltx-2.3-22b-distilled-1.1 for v1.1, while the required files include ltx-2.3-22b-dev-fp8.safetensors; verify which model is selected before running.
Suggestion · non vérifié
Verify that all required model files and node packs are available in your ComfyUI setup.
Suggestion · non vérifié
Faut-il modifier cela ?
Connectez-vous pour envoyer une demande de modification.
Sources
1 sourceExtraits de sources
1 extraitSource context: 291 downloads · Type Workflows · Base model LTXV 2.3
The prompts video gets saved in this order so you can stop generating if Cut 1 is bad.
Ltx360_Cut1 / Ltx360_Cut2 / Ltx360_Final
Using (MANDATORY) a reference audio, This Workflow will lip-sync to the audio you provide.
For best possible results you should prompt that the subject is talking.
And if you transcribe what is said in your audio input, results might be even better.
My APP MODEs are Designed for convenience not flexibility, as some of these Workflows if not all from RUNEXX are complex for Beginners, and the models he used have to be dug up instead of auto installed with the LTX2.3 template built-in Comfyui
that most people should start with to get.... Comfy. :)
PROMPT EXAMPLE:
The Punk guitarist on the right sings with perfect lip-sync to the attached audio.
The viking guitarist on the left headbang while playing the main guitar part of the attached audio perfect sync.
The woman drummer in the middle plays looks very angry and play the drums of the attached audio in perfect sync
The whole background is realistic and on fire.
MODIFICATIONS TO ORIGINAL RUNEXX Workflow:
to look around for models if you just install those from the LTX2.3 i2v comfyui official template.
Made a simplified APP Mode for Easy Usage and Beginners.
Removed Tael Mini Vae and ltx preview override that are not useful in APP Mode ( also removes the need to go fish for that tiny VAE model )
Removed the need/freedom to manually set the width and height.
Removed the prompt enhancer for so many reasons I stopped counting
Post-Scriptum:
I make these workflow for myself and share them here out of my hearts desire.
Don't ask for support. I already work tech support and you probably can't afford me ;-)
This is a very fun APP
OPTIONS:
Toggle : t2v / i2v
Toggle : High / Low quality
LtxApp360 Theme Song Lyrics:
[INTRO]
This Workflow is using Custom audddioooo
And it will lip-sync if you prompt it
to the song of your choice!
OH YEAH!
{CHORUS]
For best possible results
you should prompt
that the subject is talking or singing!.
[Guitar Solo]
OH YEAH!
Estimation des besoins VRAM
32+ Go estimés
44,0 GB de fichiers de modèle. Total des fichiers du modèle + 25 % de surcharge de chargement + 2 Go de tampon d'exécution, arrondi à l'unité supérieure.
Exigences
Exigences 7Text encoder · 8.80 GB · Hugging Face · Comfy-Org/ltx-2
VAE · 27.1 GB · Hugging Face · Lightricks/LTX-2.3-fp8
LoRA · 7.08 GB · Hugging Face · Lightricks/LTX-2.3
Upscaler · 950 MB · SAFETENSORS · bf16 · Hugging Face · Lightricks/LTX-2.3
Pack de nœud · Registry
Pack de nœud · Registry
Pack de nœud · Registry