Text to video AI
Type a prompt, choose a resolution and length, and generate a Wan 3.0 clip. You see the token cost before you send. If the job fails, the tokens come back in full.
How it works
- Write your prompt.
- Choose the resolution and length.
- Check the token estimate on the send button.
- Generate, then download.
Write a prompt that works
A good prompt has four parts, in this order:
- Subject: who or what is in the frame.
- Action: what happens.
- Camera: how the camera moves, or that it stays still.
- Light and mood: where the light comes from and how the scene feels.
Example 1 "A street food cart at a night market, steam rising from the pan, the handheld camera slowly follows the cook, warm light from string bulbs. Sizzling sound and distant crowd chatter."
“A street food cart at a night market, steam rising from the pan, the handheld camera slowly follows the cook, warm light from string bulbs....”
Example 2 "Cinematic aerial shot gliding slowly forward over terraced rice fields in Bali at sunrise, mist drifting between palm trees, golden light on the water-filled terraces, soft birdsong and a gentle breeze."
“Cinematic aerial shot gliding slowly forward over terraced rice fields in Bali at sunrise, mist drifting between palm trees, golden light on the...”
Keep the prompt focused: two to four sentences with one main action. Many actions in one prompt often produce confusing motion.
Prompts work best in English. Indonesian prompts are not tested yet.
Before and after
Before: "a dog on a beach"
After: "A golden retriever runs across a beach at sunset, waves splashing, slow tracking shot from the side, soft orange backlight."
The full prompt gives the model a subject, an action, a camera move and light to work with, so there is less to guess.
Resolution and duration
| Resolution | Use it for | Tokens per second (Standard) |
|---|---|---|
| 480p | Drafts and tests | 2 (Rp1,200) |
| 720p | Social posts | 4 (Rp2,400) |
| 1080p | Final cuts | 8 (Rp4,800) |
A 5-second 720p Standard clip costs 20 tokens (Rp12,000). See the full pricing.
Token estimate and refunds
The estimate appears on the send button before you start. If a job fails, the tokens return to your balance in full, automatically.
Questions
How long can a text to video clip be?
From 2 to 30 seconds per clip.
Why did my video fail?
Common reasons are a prompt or content that breaks our Content Policy (the job is rejected) and temporary service errors. Tokens for a failed job return in full.
Do I need tokens for text to video?
Yes. Every clip uses tokens, so top up first. See pricing.