Hotel lobby AI video6 original lobby scenes

Support

Frequently Asked Questions

The support docs in one place: how credits and refunds work, which quality to pick, what to do when a generation fails, and how photo privacy is handled.

Quality & cost per export

Photo selection guide →

What does each quality cost, and which should I pick?

Every export produces the same 15-second scene and soundtrack — only resolution, credit cost and file size differ. 480p (20 credits) is enough to judge faces, motion and scene. 720p (45) is the practical default, sharp on phones and laptops. 1080p (80) is for big displays or post-editing. Unsure? Start at 480p — it costs less than half of 720p, and you can re-export higher once the result is right.

Does higher quality change the motion, or fix a blurry photo?

No on both counts. The scene, motion and audio are identical across qualities — only the pixel count changes. And resolution cannot recover detail that was never in your photo: a blurry source produces a sharper blurry video at 1080p, not a better one. If sharpness is the problem, fix the source photo first — see the photo selection guide.

How do credits work? Do they expire?

Credits are the only currency — no subscription, no monthly fee. The cost is charged when you press generate (20 credits for 480p, 45 for 720p, 80 for 1080p), and the generator always shows the price before you commit. Credits never expire: buy a pack now and spend it next year, the balance will still be there.

What happens to my credits when a generation fails?

They come back automatically — no ticket needed. If the safety review rejects the request or rendering fails, the full amount returns to your balance, usually within a minute. One exception: canceling a queued task yourself is treated as a stop request, so those reserved credits are not returned.

Failed or stuck generations

Fix face drift →

My task failed instantly — or mid-render. Why?

Instant failures are almost always the safety review rejecting the prompt or a photo before rendering — change the prompt or swap the photo and retry; the rejected attempt already refunded itself. Mid-render failures usually mean the model choked on the inputs (a very low-resolution or corrupted photo, an extreme aspect ratio) or a temporary provider overload — check your photos, regenerate, and if it fails again immediately, wait a few minutes.

My task has been processing for 15 minutes. Is it stuck?

Maybe, maybe not — queues are real. Rendering a 15-second video typically takes a few minutes, and busy queues can push it longer; wait up to about 30 minutes before treating it as stuck. If a task eventually times out on our side, it is marked failed and the credits come back automatically.

Photo privacy & safety

Privacy policy →

What happens to my photos? Do you train AI models on them?

Your photos travel over an encrypted HTTPS connection to our servers and then to the AI providers that render your video — they need the images to generate anything at all. We do not use your photos to train models; generation goes to third-party providers under API terms that prohibit training on customer-submitted content. The full legal picture — what we collect, how long, and your rights — is in the privacy policy.

Who is allowed to appear in my video?

Only people who have agreed to be depicted: yourself, friends and family who are fine with it. Strangers, ex-partners, coworkers or public figures are off-limits — realistic photos of real people are also the most likely to trip the automated safety review, which rejects the task before rendering and refunds the credits. One hard rule: the service is for adults only, so never upload photos of minors, even as a parent or guardian.

Still stuck?

Contact support and include your task ID — we can trace exactly what happened.