> ## Documentation Index
> Fetch the complete documentation index at: https://docs.percify.io/llms.txt
> Use this file to discover all available pages before exploring further.
# How long does a Percify video take to generate?
> A Percify talking avatar usually takes 1 to 3 minutes, and each Replicate a short clip about 1 to 3 minutes. Podcasts and long jobs show an estimate.
It depends on what you're making. In the Percify app, a talking avatar made in chat usually takes 1 to 3 minutes, and each clip of a **Replicate a short** remake takes around 1 to 3 minutes. Longer jobs, like podcasts and multi-scene videos, show their own estimate. Many of them keep running if you leave the page.
## How long does each feature take?
These are the estimates the app itself shows.
| What you're making | Typical time shown in the app | Can you leave the page? |
| ------------------------------------------- | ------------------------------------------------------- | ------------------------------------------------------------- |
| Talking avatar video in chat | Usually 1 to 3 minutes | Yes: "You can keep chatting." |
| Talking video from a photo (Clone Yourself) | No time shown | Yes: "We'll notify you when it's ready." |
| Replicate a short: analyzing the video | Usually 1 to 3 minutes | Not stated |
| Replicate a short: generating the replica | About 1 to 3 minutes per clip | Yes. You get an in-app notification, **Your video is ready**. |
| Replicate a short: share cuts | Usually 1 to 4 minutes after the replica is done | They appear on their own |
| Flows with several scenes | Around 1 to 3 minutes per scene | Yes: "Safe to leave." |
| Podcast Studio | "Ready in about N to M minutes", shown before you start | Yes. You're notified when it's done. |
| Basketball highlights | "A 30 minute game takes a few minutes." | Yes |
| Realtime video | The screen says "Video in seconds, not minutes." | Not stated |
| Playground | Depends on the model | Yes, see below |
### Why do podcasts take longer?
Podcast Studio's estimate grows with length: roughly 10 to 30 minutes of waiting for every minute of podcast. A 15-second podcast shows about 3 to 8 minutes, and the 3-minute maximum shows about 30 to 90 minutes. The exact range appears in the studio before you generate.
### Why does a replica take longer than one clip?
A replica is made one clip at a time, so the total wait grows with the number of shots. Once the first clip is done, the screen estimates the time left.
## What if a generation is taking too long?
In the **Playground**, the app waits 4 minutes for an image or other output, or 10 minutes for a video. If the job is still running after that, a **Still rendering** message tells you it's taking longer than usual. Your result appears when it's ready, and you can leave and come back.
If a Playground job still hasn't finished after 15 minutes, plus 5 seconds for every second of audio or video you gave it, it's stopped with the error **Generation timed out** and your credits are refunded. See [Refunds and failed generations](/billing/refunds).
## How can I get results faster?
* **Choose Fast for talking videos.** On the talking video screen, **Fast** is described as "quicker render, lower cost", and **High** as "sharper render, slower".
* **Keep it short.** Podcasts and replicas take longer the more seconds or clips they have.
* **Leave long jobs running.** Replicas, flows, podcasts and highlights finish without you, so you can do something else meanwhile.
## FAQ
No. Jobs aren't sped up or moved ahead because of your plan. Paid plans add credits.
Not for the features marked above. They run on Percify's side and show up when they're done. Playground results also appear once they finish.
Most features refund the credits automatically. See [Generation failed](/troubleshooting/generation-failed) for common causes.
## Related
Remake a short video, clip by clip.
Two-person AI podcasts, and their pricing.
What to do when a job doesn't finish.
What each generation costs.