Building an AI rap video workflow with scene previews and recoverable jobs

Описание и применение

Веб‑приложение RapVideo добавило шаг предварительного просмотра сцены, атомарное резервирование кредитов и функцию восстановления задач, чтобы избежать потери средств и неполных видео из‑за тайм‑аутов.

Задача

Requests to AI video providers can time out after the provider accepts a job, leaving users unaware of completion and potentially costing credits; generated scenes may have incorrect faces, requiring a review step before animation.

Результат

The updated workflow introduces a scene preview step to catch face blending errors, reserves credits atomically with task creation to avoid orphaned jobs, and uses a shared reconciliation function to recover from timeouts and ensure tasks settle correctly.

Рабочий процесс

• Submit portraits and settings
• Generate scene image
• User reviews scene image
• Approve or reject scene
• Generate video with audio using approved scene
• Store MP4 and task history
• Reconcile pending tasks via polling, webhook, or scheduled recovery

Доказательства

A request can time out after the provider has already accepted the job. The video might then finish after the user closes the tab.

Источник

The user goes through this sequence: portrait(s) + scene choices ↓ scene image generation ↓ user checks the scene ↓ video generation with audio ↓ stored MP4 + task history

Источник

Before submission, the application reserves the task and debits credits in one atomic database operation.

Источник