Vidu Q4 Video Generator

1. 01

   Step 1: Choose a mode and add your images

   Pick Image to Video when you already have the exact first frame you want, such as a product photo, a portrait or a finished illustration. Pick Reference to Video when you want Vidu Q4 to build a new scene from separate pieces: a character, an outfit, a prop and a location.

   • One clean, well-lit image works better than a busy collage. Crop out borders and text before uploading.
   • Use more reference images only when they add new information. Up to 12 is worth it for recurring characters in a series: front, side and three-quarter views of a face, the full outfit, key props and the set.
   • For a product ad, 3–6 references are usually enough: the product from two or three angles, the logo, and the background mood you want.
2. 02

   Step 2: Write the prompt and add voices

   Describe what happens over time rather than what the picture looks like, because the images already show that. Cover the action, the camera move and the mood, in that order. In Reference to Video, refer to inputs by position, for example "the character from reference image 2".

   • Voice references are MP3 files of 3–12 seconds. Use one clean recording per speaking character, without music or background noise.
   • With two or three voices, name who speaks to whom in the prompt, such as "reference image 1 says hello to reference image 3 using reference audio 2".
   • Image to Video clips always come with native audio. In Reference to Video, leave Native audio on for ambient sound and dialogue, or switch it off for silent clips you will score yourself.
3. 03

   Step 3: Set length and resolution, then generate

   Duration runs from 3 to 16 seconds. Short clips render faster and are easier to get right, so draft at 3–5 seconds and extend once the motion works. The credit cost updates as you change settings, before you generate anything.

   • Draft at 540p: it is the cheapest way to test a prompt, and the motion you see carries over to higher resolutions.
   • Choose 1080p for social media and most web use. Choose 2K or 4K, which also add 10-bit color, only for final renders that need large screens, color grading or cropping.
   • Keep the seed when you like a result and only want to tweak the prompt; leave it empty to explore new variations.

New to the model itself? Read the introduction to Vidu Q4 on the home page first.