Skip to content
PipeListedHosted

LTX-2 video with generated audio

Generate a video with accompanying generated audio from a text description. Describe subjects, appearance, actions over time, camera movement, and desired sounds, dialogue, or singing. No uploaded image or audio is required. The selected output combines the decoded imagery and audio into one video; MP4 is the default container.

Workflow details

FamilyLTX-VideoResolution1024x1024Estimated VRAMNot yet characterizedPublisherSweet Tea AutomationUpdatedOct 26Bgrade

Runtime profile

Publisher
Sweet Tea Automation
Status
Listed
Version
Checking published version…
Updated
Oct 26, 2026

Hover nodes for Pipe controls

Core

Primary Pipe surface

Scene and sound

CLIPTextEncode

Negative prompt

CLIPTextEncode

Output size

EmptyImage

Video length

PrimitiveInt

Video output

SweetTeaPreviewVideo

Output

Extended

Additional surfaced controls

Video timing

PrimitiveFloat

Audio timing

PrimitiveInt

Initial camera guidance

LoraLoaderModelOnly

Refinement camera guidance

LoraLoaderModelOnly

Initial randomness

RandomNoise

Initial guidance

CFGGuider

Initial sampling

LTXVScheduler

Initial sampler

KSamplerSelect

Refinement randomness

RandomNoise

Refinement guidance

CFGGuider

Refinement sampler

KSamplerSelect

Video precision

CreateVideo

Distilled refinement adapter

LoraLoaderModelOnly

Refinement noise schedule

ManualSigmas

What you need

Model files
26.1 GiB known subtotal

Remaining hardware or download-size estimates are not available yet.

How this recipe works

smaller decode tiles

failure · stage is decode

  • Set tile_size to 256 on node 109.

Smaller tiles can change seams and decode work; inspect the output. Only the VAE decode stage changes. Model loading and denoising memory are unaffected.

Guides for this recipe

Output

  • Sweet Tea Preview Video · video

Requirements

Models

5

Node packs

1

Sources

1 source
Sweet TeaSource