Video Guide

The Lomio video pipeline takes an approved storyboard and drives it all the way to rendered video output, with a multi-stage process you can monitor and control at each step.

How the video pipeline works

Video generation in Lomio is a multi-stage automated pipeline, not a single button click. Each stage produces intermediate assets that are reviewed or auto-approved before passing to the next stage. This design gives you visibility and control at each step while letting the AI handle the heavy lifting.

The pipeline runs asynchronously — you do not need to stay on the page.

Pipeline stages

Stage 1 — Frame prompt generation

Starting from an approved storyboard, the pipeline generates an optimized image-generation prompt for every frame. These prompts incorporate:

• Shot type and camera angle from the storyboard.
• Character visual descriptors and LoRA trigger words for each character in frame.
• Scene environment, lighting, and mood descriptors.
• Style tags consistent with the project's visual direction.
• Negative prompt elements to avoid common artefacts.

Stage 2 — Frame image generation

Each frame prompt is sent to the image generation model. Character LoRAs are applied automatically. You can review generated frames individually in the Frame Review panel — approve, reject, or request a variant for any frame before proceeding.

Stage 3 — Video clip generation

Approved frames are converted to short video clips using a video generation model. Each clip adds motion to the static frame — camera movement, character motion, environmental animation. Motion parameters (intensity, direction) can be adjusted per frame in the pipeline settings.

Stage 4 — Audio assembly

If voice-over or music is configured for the project, this stage assembles the audio track — synthesized voice-over timed to the storyboard dialogue, background music from the project's audio library, and sound effects. Audio assembly is optional; you can skip it if you prefer to add audio in post-production externally.

Stage 5 — Final render and stitch

All clips are stitched together in storyboard order, transitions are applied, and the final video file is rendered. The completed video is stored in the Video Library and a download link is available immediately.

Starting a video pipeline run

1. Open an episode that has an Approved storyboard.
2. Click the Video tab and then Start Pipeline.
3. Configure pipeline settings:

   • Aspect ratio — see the table below.
   • Output resolution — 720p, 1080p, or 4K (if your plan supports it).
   • Frame review mode — Manual (you approve each frame) or Auto-approve (frames above a quality threshold are approved automatically).
   • Audio — enable/disable voice-over and music.
4. Review the estimated credit cost and click Confirm & Run.

Aspect ratios and resolutions

The default aspect ratio for a new pipeline run is inherited from the project's platform setting. You can override it per run.

Monitoring pipeline progress

The Video tab shows a real-time status indicator for each pipeline stage. You can also monitor the overall run status from the project dashboard. Stages run sequentially — a stage only starts after the previous one completes or all frames in that stage are approved.

Pipeline runs can be paused between stages (for manual review) andresumed when you are ready. They cannot be rewound — to regenerate earlier-stage output, cancel the run and start a new one.

The video library

Every completed pipeline run produces one or more video files stored in the project's Video Library. From the library you can:

• Preview any video in the browser player.
• Download in the original rendered format (MP4, H.264).
• Compare two runs side-by-side using the comparison view.
• Tag and annotate videos for client review.
• Archive older runs to keep the library organized.

Video files are stored on Cloudflare R2 and served via CDN. Download links are permanent for the lifetime of your workspace; they are revoked if the workspace is deleted.

Credit costs for video generation

Video generation is the most credit-intensive operation in Lomio. Credit consumption depends on:

• Number of frames (proportional to episode length and storyboard density).
• Output resolution (4K costs more than 1080p).
• Whether audio assembly is enabled.
• The video generation model tier selected.

An estimated credit cost is displayed before you confirm a pipeline run. You can reduce costs by lowering resolution for draft runs and reserving high-resolution renders for final delivery. See the Billing FAQ for plan-level credit details.

Re-running a single frame

If a specific frame in a completed run does not meet your quality bar, you can re-run just that frame without restarting the entire pipeline. Open the frame in the Video Library, click Regenerate Frame, optionally adjust the frame prompt, and submit. The new frame replaces the previous one in the assembled video on next render.