Convert Vertical Video to Horizontal Format Without Cropping in ComfyUI
You filmed or generated a vertical video. Now you need it for YouTube, a presentation, or a widescreen display. Cropping cuts the top and bottom. Pillarboxing black bars look unfinished.
There's a better option.
Vertical video in. Full cinematic horizontal frame out, with AI-generated content filling the sides.
Run it now on Floyo!
Why This Workflow Is Different
The standard approach is a crop or black bars on the sides. Either you lose content, or the video looks like it was made for a phone and forced onto a screen.
This workflow outpaints instead. Wan 2.1 VACE generates new content to the left and right of the original frame to fill the widescreen canvas naturally. The original vertical video stays centered and intact. The AI extends the scene around it.
no cropping, no pillarboxing
AI-generated scene extension on both sides of the frame
original motion and timing preserved throughout
subject stays centered in the horizontal frame
prompt-guided to match the scene context
How It Works
Your vertical video is loaded and frames are extracted with FPS and dimensions preserved. The workflow calculates how much horizontal space needs to be generated on either side to reach 16:9 format (1280×720).
ImagePadForOutpaint creates masks for the empty side regions. Wan 2.1 VACE 14B with the FusionX LoRA fills those masked areas with temporally consistent video content that matches the original scene across every frame. The result is assembled back at the original FPS and saved as a clean horizontal file.
Key Inputs
Your Video
Any vertical MP4. The workflow extracts frames and calculates the required padding automatically.
Works well with:
talking head or interview footage with centered subjects
portrait or character video with readable environment context
AI-generated vertical clips needing horizontal delivery
product demo footage with a clear central subject
Works less well with:
footage where the subject fills the full vertical frame edge-to-edge — less context for the model to extend from
fast lateral motion where side extension becomes inconsistent
very low-resolution source video
Positive Prompt
Describe the scene so the model generates appropriate content in the extended side regions. Match the environment in your original footage.
Examples:
"indoor studio, clean background, professional lighting, consistent environment""outdoor urban street, natural daylight, city background""nature scene, forest background, soft natural light, wide landscape""product photography, white background, studio lighting"
Negative Prompt
Prevent artifacts in the generated regions.
"artifacts, blurry edges, inconsistent lighting, distorted background, watermarks, black bars"
What This Is Great For
YouTube and streaming: convert TikTok, Reels, or AI-generated vertical content into widescreen format for YouTube upload without losing the original shot.
AI video post-processing: AI video generators often output in portrait format for social platforms. Run the output through this workflow when you need the same clip for widescreen delivery.
Presentations and displays: vertical content on a widescreen display looks wrong. Convert it properly before presenting or displaying on horizontal screens.
Content repurposing: produce one master vertical clip and convert to both formats in the same session. Covers every platform from one generation.
Marketing and ad teams: repurpose vertical social ad assets into horizontal format for YouTube pre-roll, display ads, and widescreen placements.
What to Watch Out For
Subjects that fill the full vertical frame leave the model less to work from when extending sideways. If there's no visible background in the original, the outpainting has to invent more. Leave some environmental context visible in your source clip for cleaner results.
Fast lateral camera movement creates consistency issues in the extended regions. Stable, slower footage converts more cleanly.
Match your prompt to the actual scene. A mismatched prompt produces inconsistent side generation. If your video shows a white studio background, say so in the prompt.
Very long clips take proportionally longer to process. Test on a short segment before running the full clip.