Example with generated characters — these people don't exist
Deeeeeelai!
A viral video from a gas station at sunset: one person dances, another films a selfie at arm's length, a third gets out of a car. Upload three photos — and the characters become yours.
- Dances at the start
- Films the selfie
- Gets out of the car
Sign up and top up your balance — credits are charged only for a finished video.
💡 Photos of celebrities or movie stills? The AI may reject them — make a "character sheet" first
Upload the photo to Qwen Image 3.0 with the prompt below — you'll get a stylized illustration of the same person from 4 angles. Upload that to the trend: generations go through much more often this way, especially if one of the scenes failed before.
Using the person in this reference photo, generate a single square image containing a 2x2 character reference sheet (4 cells in one image): Top-left: FRONT VIEW (face straight at camera, neutral expression). Top-right: SIDE PROFILE (90° left profile view). Bottom-left: THREE-QUARTER VIEW (3/4 angle, slightly turned). Bottom-right: FULL BODY (head to toe, same person, same clothing). CRITICAL RULES: - Keep the EXACT same person: same face shape, hair color/style, skin tone, clothing, body proportions. - Each cell should have a clean, simple background (light gray or white). - Professional character sheet layout — clear separation between cells. - Detailed digital art / illustration style — clearly NOT a photograph, but highly detailed and recognizable as the same person. - Stylize like a high-quality video game character or concept art. - Square output (1:1 aspect ratio). - High quality, detailed rendering. - Do NOT add any text labels, captions, or titles on the image. Output exactly ONE square image.
- Prompt and motion are already set up — nothing to write
- Charged only for a finished video; credits come back if it fails
- Files are kept for 10 days — download what you want to keep.
How it works
Upload three photos: who dances at the start, who films the selfie and who gets out of the car. The scenario, camera motion and prompt are already set up: the AI recreates the video shot by shot with your characters, and we add the trend's music ourselves.
Full-length photos with one person in the frame, a clearly visible face and the outfit you want to wear in the video work best: the AI takes the face, hairstyle and clothes from the photo. If the photo shows only the face or the upper body, it will draw the missing clothes itself. The finished video is vertical, 20 seconds long — ready for Reels, Shorts and Stories.
FAQ
How long does the video take?
Usually 10–15 minutes. The result appears in "My generations", and if your account is linked to the bot, it also arrives in Telegram or MAX.
Can I upload the same photo several times?
Yes. If you want to be all three characters at once, upload your photo to every slot.
What if the generation fails?
Credits are charged only for a finished video. If the generation fails, the reserved credits return to your balance automatically.
Can I upload photos of other people?
Only with their consent. Photos are used only for this generation and are deleted together with the result.