Generation jobs and recovery
How Lovision tracks AI tasks and recovers interrupted or failed generations.
How generation jobs work
When you start an AI generation, Lovision creates a generation job and sends it to the AI model. Jobs run asynchronously — you can continue working on the canvas while the AI processes the request. The status of each active job is shown in the generation status area.
What happens when a job is interrupted
Jobs can be interrupted by:
- A browser refresh or tab close while the job is running
- A network disconnect
- A timeout on a long-running task
When you reopen the project, Lovision checks if any jobs were in progress and tries to recover the result:
- If the AI finished the job before the interruption, Lovision applies the result to the project automatically.
- If the job is still running, Lovision waits and shows the status.
- If the job failed, Lovision shows the error state. Start a new generation after adjusting the prompt, references, or model.
Recovering a job manually
If Lovision does not automatically apply a result after reopening:
- Check the visible generation status for pending or failed work.
- For pending jobs, wait for the status to update — recovery may take a moment.
- For failed jobs, adjust the prompt, reference files, or model.
- Submit a new generation from the prompt input.
Tips
- Keep the project open for at least 30–60 seconds after starting a large generation — closing it immediately interrupts the handoff.
- Video and high-resolution image generations take longer and are more likely to hit timeouts. Check back in a few minutes rather than waiting on the page.
- If the same prompt consistently fails, try a shorter prompt, fewer output variations, or a different model.