LTX-2.5 ICLoRA Outpaint for Video Outpainting

ABOUT THE WORKFLOW

Extend a Video Beyond the Frame
Upload a video and set a larger canvas size. The model fills the new area to match the existing scene, preserving the original region and its audio. Add an optional prompt to describe what the outpainted area should look like.

Model

• LTX-2.5 22B Distilled by Lightricks. An open-source video model that generates video and audio in one pass. Strong at fast inference, outpainting, and inpainting with IC-LoRA guidance.

• IC-LoRA (In/Outpainting). A LoRA adapter that guides the model to fill masked regions while keeping consistency with the original footage.

HOW IT WORKS

Step 1. Upload your video
The video you want to extend. This sets the length, frame rate, and audio track.
Works great with: landscapes · action shots · drone footage · interiors

Step 2. Set the target canvas size
Enter a wider or taller resolution than the original. The difference becomes the outpainted area.

Step 3. Add a prompt (optional)
Describe what the extended area should look like. Without a prompt, the model extends the scene on its own.

Step 4. Add a start image (optional)
Turn on "use image input" and upload a separate image to condition the first frame. Leave it off for most runs.

Step 5. Hit run and download
The workflow runs a two-stage diffusion pass, blends the result with the original, and reassembles the video with audio.
Ready for: Premiere · DaVinci Resolve · After Effects · any NLE

First time? Leave every setting as-is. The defaults (1920×1088 · seed 43 · no prompt) are the right starting point for almost everyone.

RECOMMENDED SETTINGS

Quick-start guide. Find the goal that matches yours and copy the settings.

• Standard outpaint (most people) — 1920×1088 · no prompt · seed 43. The right starting point for almost everyone.

• Want a wider frame — Increase Target Width beyond your video's native width. Keep the height the same if you only need horizontal extension.

• Want a taller frame — Increase Target Height. Useful for converting landscape footage to a squarer or portrait crop.

• Outpaint a known location — Add a prompt describing the surrounding scene. Turn on "Enhance positive prompt" to let the model expand your description.

• Green traces at the edges — Increase the blend dilation value inside the workflow. A different seed can also help.

• Result looks soft or wrong — Add a short prompt describing the outpainted area before changing any other setting. Keep the expansion moderate; small extensions blend better than doubling the frame.

Prompt: Leave it empty for your first run. The model extends the scene on its own. When you add a prompt, describe the outpainted region the way you would describe a scene, not an edit. "Wide open sky with soft clouds, sunlight from the left" works. "Add more sky" does not.

LEARN

📹 Videos

• Intro to Floyo

• ComfyUI 101 Free Course ft. Sebastian Kamph

• Floyo 101 for Team Collaboration

✨ Quick links

• Help docs

• LoRA Training guide

• Discover more workflows

USE CASES

🎬 Filmmakers & Editors
Extend a shot to fit a wider aspect ratio without reshooting. Convert 16:9 footage to ultrawide or IMAX-style framing in post.

🎮 Game Developers
Generate extended environment footage for in-game cinematics or skybox animations. Upload a reference clip and widen the frame for panoramic backgrounds.

🥽 VR & Immersive Creators
Expand video footage beyond the original field of view for 180° or wider VR experiences. Pair with a panorama workflow for full 360° coverage.

📱 Social Media & Content Creators
Reframe horizontal footage into square or vertical formats by extending the top and bottom, keeping the original content intact instead of cropping it.

WHAT WORKS BEST / WHAT TO AVOID

✅ Works great

• Outdoor landscapes and wide establishing shots

• Moderate canvas extensions (20%–50% larger)

• Well-lit footage with even exposure across the frame

• Footage with clear environmental context

⚠️ May produce softer results

• Doubling the canvas size in one run

• Tight close-ups with shallow depth of field

• Heavy vignetting or uneven lighting at the edges

• Fast camera movement across the frame edge

FAQ

What is video outpainting and how does it work?
Video outpainting extends a video beyond its original frame boundaries. Instead of cropping or stretching, the model generates new content that matches the existing scene in lighting, color, and motion. This workflow uses LTX-2.5 with an IC-LoRA adapter trained for outpainting, running a two-stage diffusion process with Laplacian pyramid blending to produce a clean transition between the original footage and the generated area.

What is LTX-2.5 and what makes it good for outpainting?
LTX-2.5 is a 22-billion parameter open-source video model by Lightricks. It generates video and audio together in a single pass, and supports image conditioning, inpainting, and outpainting through IC-LoRA adapters. Its distilled transformer architecture keeps inference fast, typically completing a clip in under 10 seconds for standard lengths.

Does the outpainted video keep the original audio?
Yes. The workflow extracts the audio from your input video, encodes it as a frozen reference, and reassembles it with the final output. The original audio track is preserved in the exported file.

What resolution should I set for the target canvas?
The default is 1920×1088, which works well for extending standard HD footage. Width and height must be multiples of 64. Keep the expansion moderate for the best results. Doubling the canvas area roughly doubles the generation time and cost.

Can I use this to change aspect ratios, like 16:9 to 9:16?
Yes. Set the target width and height to the aspect ratio you need and the model fills in the new area. This is useful for reformatting horizontal footage for vertical platforms without losing the original content to a crop.

Is the output licensed for commercial use?
LTX-2.5 is released under the LTX-2.x Community License. It is free for entities with under $10M annual revenue. Larger entities need a paid license from Lightricks. Check the license terms for your specific use case.

How to run LTX-2.5 video outpainting online?
You can run LTX-2.5 video outpainting online through Floyo. No installation, no setup, no API key to wire up. Open the workflow in your browser, upload your video, and hit run. Free to try.

WHY FLOYO?

Floyo is the only platform with team collaboration for ComfyUI in the browser. You run workflows with no install. You share run history, assets, and models across your team. You pay only when you generate. Floyo supports open-source and closed-source models.

A designer runs an edit and likes the result. A teammate opens that exact run from shared history and keeps going. No file handoffs. No version confusion.

For studios and enterprise teams, Floyo adds private workspaces, pooled resources, and a team usage dashboard. Other ComfyUI cloud tools run for one person at a time. Floyo runs for the whole team, with transparent per-generation costs.

Ready to try it?
Upload your video, set the canvas size, and run it. The settings are already set.

→ Launch Workflow, Free

Questions? Watch the free course or check the FAQ above.