ComfyUI Node
LTX-2.3 Img/Audio to Video ⚡
Prompts, init latent and noise masks for LTX-2.3 T2V/I2V/A2V/IA2V. Feed into the LTX-2.3 KSampler.
LTX-2.3 Img/Audio to Video ⚡
- clip
- vae
- audio_vae
- image
- reference_audio
- positive
- negative
- latent
◄prompt—►
◄negative_prompt►
◄width768►
◄height512►
◄length121►
◄frame_rate24.00►
◄batch_size1►
◄image_strength0.70►
◄length_from_audiotrue►
Category🤖 CCTech/LTX-2.3
Inputs (14)
| Name | Type | Default | Description |
|---|---|---|---|
| clip | CLIP | — | |
| vae | VAE | The loader's video_vae output. | |
| audio_vae | VAE | The loader's audio_vae output. | |
| prompt | STRING | Describe the scene and its motion. A caption, not an instruction. | |
| negative_prompt | STRING | — | |
| width | INT | 76864–16384 | — |
| height | INT | 51264–16384 | — |
| length | INT | 1219–16384 | Frames; 8k+1 tiles exactly (9, 97, 121...). Ignored when length_from_audio is on. |
| frame_rate | FLOAT | 24.001–120 | 24 is the LTX-2 convention. Match this in CreateVideo or playback drifts. |
| batch_size | INT | 11–4096 | — |
| imageopt | IMAGE | First frame. Resized and CENTER-CROPPED to width x height here - do not scale it upstream. | |
| reference_audioopt | AUDIO | — | |
| image_strengthopt | FLOAT | 0.700–1 | i2v only. How much of the init image to keep. 0.7 is the official value; 1.0 locks the first frames hard. |
| length_from_audioopt | BOOLEAN | true | With reference_audio: size the video to the clip. |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| positive | CONDITIONING | — |
| negative | CONDITIONING | — |
| latent | LATENT | — |