Unlike traditional creation (which matches your script to stock footage), the AIGC pipeline generates original visuals for each scene from scratch — a storyboard image first, then an animated motion video clip. This gives you fully custom, on-brand content that doesn't exist in any footage library.

How the AIGC Pipeline Works

The AIGC pipeline splits your video into scenes and produces visuals in two stages:

1. Storyboard image generation — For each scene, the AI generates a still image based on your script, chosen visual style, object and environment descriptions, and optional reference images.
2. Motion video generation — Each storyboard image is animated into a short motion video clip.

Both stages run asynchronously. Track per-scene progress via the scene status API or receive webhooks on completion.

💡 Tip — You can let the pipeline run end-to-end automatically, or take manual control: generate storyboards first, review them scene-by-scene, regenerate any frames you don't like, and then trigger motion video generation only when ready.

Choose a Visual Style

Before creating an AIGC project, browse the available visual styles. The style you choose controls the look and feel of every generated storyboard image.

• List all AIGC styles — see GET /openapi/v1/project/list-aigc-style

Create an AIGC Video Project

Start the pipeline by creating a project. Provide your script, pick a visual style, and decide whether motion videos should be generated automatically.

• Create an AIGC video project — POST /openapi/v1/project/generate-aigc-video — returns a project UUID.
• Create an AIGC video and auto-export to clip — POST /openapi/v1/project/generate-aigc-video-to-clip — returns both a project and a clip; the clip is populated and exported automatically when the pipeline finishes.

Two control modes

The aigc_config object controls how far the pipeline runs without you:

Use Manual mode when you want full control over each storyboard frame before committing to motion video generation (which is the more time- and credit-intensive step).

Key Workflow Steps

Automatic mode

If you used generate-aigc-video-to-clip, step 3 is handled for you — the clip is created and exported automatically, and you receive both project_generation and project_export webhooks.

Manual mode

Scene-Level Operations

The AIGC pipeline exposes two scene-level endpoints that let you generate or regenerate storyboard images and motion videos independently. Both endpoints are credit pre-checked before any generation starts.

Generate storyboard images

POST /openapi/v1/project/{projectUuid}/scene/generate-image

Generate or regenerate storyboard images for one or more scenes (up to 20 per request). You may override the prompt, aspect ratio, and provide reference images per scene.

{
  "scenes": [
    {
      "scene_id": "1531339307812433920",
      "prompt": "Cinematic, documentary-inspired close-up of a sea turtle...",
      "aspect_ratio": "landscape",
      "ref_images": [
        { "entity_uuid": "1531339304335355904", "image_url": "https://..." }
      ]
    }
  ],
  "force_override_video": false
}

Behavior: always (re)generates images for the requested scenes — existing storyboard images are overwritten. Scenes that already have a motion video are rejected unless force_override_video=true.

Generate motion videos

POST /openapi/v1/project/{projectUuid}/scene/generate-motion-video

Generate or regenerate motion videos for one or more scenes (up to 20 per request). Each scene reuses its storyboard image as the source frame.

{
  "scenes": [
    {
      "scene_id": "1531339307812433920",
      "prompt": "Gentle camera push-in...",
      "aspect_ratio": "landscape",
      "motion_video_mode": "first_frame_to_video",
      "ref_images": [ ... ]
    }
  ],
  "completion_webhook": "https://example.com/callback"
}

Scene optional fields:

Behavior: (re)generates motion videos for all requested scenes — existing motion videos are overwritten by default. There is no skip logic; every scene in scenes is processed.

Credit Pre-Check

Both scene-level generation endpoints perform a credit pre-check before any generation starts. If the workspace credit balance is insufficient, the request is rejected immediately with an error — no partial generation occurs.

Notes:

• The motion-video estimate uses the same catalog and floor as the create-video pre-check.
• The estimate is a pre-check only; the actual credit consumption happens in the async job. If the pre-check passes, the job will proceed.
• A request with zero processable scenes (all scenes failed validation) skips the credit check and returns immediately.

Status Tracking

Which API to poll

• Get Project Info — tracks project generation status. If you are not regenerating scenes, this is all you need.
• Scene AIGC Info (GET /openapi/v1/project/{projectUuid}/scene-aigc-info) — per-scene status, for when you regenerate scenes.

Project generation status

Poll the Get Project Info API and track two fields:

How they relate:

• If auto_generate_motion_video is false, the task ends when task_status reaches completed (motion_status stays skipped / not_required).
• If auto_generate_motion_video is true, motion_status only starts after task_status is completed — so you need to track both, in that order.

Per-scene status

Each scene reports two items — storyboard and motion video — each with its own status:

The response also includes:

• transcript — the scene voiceover transcript as a structured JSON object (items / sentences / originalVoiceover).
• assetLink / thumbnailLink — S3 presigned URLs for each asset (valid for ~25 hours).
• prompt / genModel (currently fixed to veo_3.1) / aspectRatio / enableAudio — generation parameters.
• refImages[] — reference images used for the generation (presigned URLs).

Handling Failures

Both storyboard and motion video generation can occasionally fail for individual scenes. Visla provides dedicated scene-level endpoints to retry without re-running the whole pipeline.

Regenerate a storyboard image

If a scene's storyboard image failed (or you want a new frame), call the unified image endpoint:

• POST /openapi/v1/project/{projectUuid}/scene/generate-image
  • Provide a custom prompt and/or ref_images to steer the result.
  • Existing images for the requested scenes are overwritten.

Regenerate a motion video

If a scene's motion video failed (or you want to re-animate an existing frame), call the unified motion video endpoint:

• POST /openapi/v1/project/{projectUuid}/scene/generate-motion-video
  • All requested scenes are (re)generated; existing motion videos are overwritten.
  • If you want to regenerate the source frame as well, call scene/generate-image first.

Both retry endpoints re-run the credit pre-check before generating.

Webhook Notifications

AIGC generation is asynchronous. Register a completion_webhook URL when you create the project (or call a generate endpoint) to be notified when each stage finishes.

Quick Reference