LTX 2.3 added timeline editing, so you can now stitch images, text, and audio into one video.

The Forge / 工坊 Workflows / ComfyUI / 工作流
1

Following up on that last workflow, here’s a new thing: a node group called LTX DIRECTOR. Basically, it gives LTX 2.3 a full timeline editor. You can throw images, text prompts, and your own audio files into the same video—just load everything in, write your prompts, hit run, and let it handle the rest.

For people making short-form content, this approach makes a lot of sense: you lock in the music and pacing during generation, no need to go back to your editing software to sync things up after the video’s done.

Plus, like the previous setup, it runs on 6G VRAM too. Personally, I’m curious if the scene transitions actually follow the audio nodes when you feed in custom audio, or if it’s just background noise. Anyone tried it? How reliable is the audio-video sync?

2

It’s pretty ahead of the curve to decide on the soundtrack during the generation phase.

3

Right now, audio-to-image sync is mostly just background padding.

4

The timeline editor is a pretty nice upgrade, makes the whole experience way better.

5

People keep hammering on about 6G VRAM, guess it’s really aimed at low-end setups.

6

Just click run and it generates. The workflow’s definitely simpler now.

7

Totally agree. This kind of sync is at best just laying down some ambient background. If you actually want it to hit the beat, you still gotta manually align it in post.