What resolutions and frame counts are valid for LTX-2 in this workflow?

Set width and height to values divisible by 64 (e.g., 1280x768, 1920x1088). Use a frame count of the form 8n+1 (e.g., 9, 17, 25, 33). If you enter invalid values, the Subgraph will automatically choose the closest valid parameters, so you won’t see an error but your result may not match the exact numbers you typed.

Does this workflow generate audio and lip sync from text?

The LTX-2 distilled Subgraph can produce expressive lip sync when your prompt includes dialogue in the audio section. When the model returns an audio track, SaveVideo will mux it into the output. If your model build/weights do not output audio, you will get a silent video (you can add audio later in an editor).

How can I improve motion quality without huge render times?

Use the distilled model’s default 720p for faster runs, keep prompts precise about actions over time, and choose a moderate frame count like 17 or 25. If you upscale to 1920x1088, expect longer times. For stability, keep the seed fixed while iterating and avoid overly aggressive motion descriptions in a single short clip.

Why does my output size or frame count not match what I typed?

This workflow is designed to be forgiving: it silently snaps to the nearest valid width/height (multiples of 64) and to frame counts of 8n+1. Double-check the Subgraph’s effective settings after running, and adjust your inputs to the nearest valid target if you need a specific result.

LTX-2 Image to Video (Distilled)

This ComfyUI workflow turns a single still image into a dynamic video using the distilled LTX-2 model from Lightricks. It wraps the animation pipeline inside a Subgraph node (b7c2d337-c38d-4c04-922b-2d638449d13e) so you can focus on prompting and parameters instead of wiring low-level components. Your source is loaded with LoadImage, conformed to a video-safe size using ResizeImageMaskNode, then animated by the LTX-2 Distilled Subgraph, which generates temporally coherent frames with expressive motion and lip sync. The resulting video (and audio, when provided by the model) is written to disk with SaveVideo. A MarkdownNote in the canvas provides inline guidance and links to Subgraph docs to help you understand and customize the encapsulated graph.

API

Use this workflow from code

Every Comfy workflow is a JSON graph. The payload below is this workflow, exactly as ComfyUI runs it — fetch it from the URL, keep it in version control, or load it in ComfyUI and run it node by node.

video_ltx2_i2v_distilled.json

Fetching workflow JSON…

Run it from TypeScript or Python with the Comfy SDK. The same code targets Comfy Cloud or a ComfyUI you host yourself — only the base URL changes.

// Install (beta)
npm i @comfyorg/sdk

// Run "LTX-2 Image to Video (Distilled)" (TypeScript)
import { Comfy } from "@comfyorg/sdk";

const client = new Comfy({ apiKey: "comfyui-..." });

// This workflow, exported in API format (see note below)
const wf = await client.workflows.fromFile("video_ltx2_i2v_distilled_api.json");

const asset = client.assets.fromFile("input.png");
wf.setInput("98", "image", asset);  // LoadImage

const job = await client.run(wf);
await job.getOutputs("75")[0].toFile("output.png");  // SaveVideo

The SDK takes a workflow in API format: open this workflow in ComfyUI and use File → Export Workflow (API).

Comfy Cloud API access requires a plan with an API key.

SDK docs Get an API key

FAQ