Skip to main content

Video Generation

Summary: XBRUSH's video generation creates a video from a text prompt, optionally guided by reference images, videos, or audio. Select an AI engine, set the aspect ratio, length, and resolution, describe the motion with a prompt, and a video is automatically generated.


What is Video Generation?​

Video generation is XBRUSH's feature that uses an AI engine to create a video with natural motion from a text prompt and, optionally, reference images. With the default engine (Seedance 2.5) you can generate a video even without an image. The credit cost depends on the engine and settings (resolution, length, etc.) and is shown on the Generate button.


Video Generation Overview​

Video › Generate tab with Seedance 2.5, Reference images 0/30, Reference videos 0/10, Reference audio 0/10, 16:9, 720p, 5s, Draft 480p, Audio, Settings, and the Generate button

Video › Generate in the Direct panel. Sub-tabs: Generate · LARP · AI Edit · Upscale. The prompt line shows an example ("e.g. @Image 1 and @Image 2 are taking a selfie together"), followed by the option chips Seedance 2.5 · Reference images 0/30 · Reference videos 0/10 · Reference audio 0/10 · 16:9 · 720p · 5s · Draft 480p (off) · Audio (on) · Settings, and the Generate button with its credit cost.

  • In the Workspace input box, switch to Direct mode, select the Video tab, and open the Generate sub-tab. (Other sub-tabs: LARP, AI Edit, and Upscale.)
  • You can select the AI engine used to generate the video. Seedance 2.5 is used by default.
  • A start image is not required with the default engine — leave the reference images empty and the video is generated from the prompt alone.
  • You can add up to 30 Reference images, 10 Reference videos, and 10 Reference audio files.
  • Choose the aspect ratio (Auto, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16; default 16:9) and the resolution (480p / 720p / 1080p; default 720p) from their chips.
  • Length: the 5s chip sets the video length. The default is 5 seconds; the available range depends on the engine.
  • Draft 480p generates a quick low-resolution draft. It is off by default.
  • Audio (on by default) creates audio along with the video. Background music is left out by default because it may be blocked by copyright checks; in Chat mode you can ask the agent to add it (e.g. "add background music too").
  • Settings holds additional video options.
  • You can add a text prompt to describe the motion.
  • With the default settings (Seedance 2.5 · 16:9 · 720p · 5 seconds · audio on), the Generate button shows about ◆ 454. The exact cost changes with the engine and settings.

Generate a Video from a Generated Image​

Image ⋮ menu with the Use submenu open, showing Upscale, Remove BG, Animate, Edit in Canvas, and Filter

Hover over a generated image, click its ⋮, and choose Use › Animate to turn it into a video. The image detail view has the same Animate button.

  • Add an image and click the Run button.
  • The video generates immediately without a prompt, producing a 5-second video by default.
  • Processing time depends on the engine and settings; when it is done, a result card appears in the session feed.

Generate a Video from an Uploaded Image​

  • Add an image with + Reference images (optional), adjust the settings, and click the Generate button.
  • The default video length is 5 seconds.
  • Processing time depends on the engine and settings; when it is done, a result card appears in the session feed.

A generated video result card in the session feed

A finished video result card in the session feed, with the agent's note above it and the label · number · avatar · ▾ · time · ⋮ row below.

Writing a Prompt​

  • You can enter text to describe the motion of the image for video generation.
  • When you attach reference files, refer to them in the prompt with @: attached files are numbered in the order you added them, such as @Image 1 and @Image 2. For example, "@Image 1 and @Image 2 are taking a selfie together" tells the engine which reference plays which role. This is especially useful when you combine several references.

Generation Settings​

  • Set the aspect ratio, video length, resolution, Draft 480p, and Audio from the option chips; more options are under Settings. The available options vary by engine.

Generation History​

See Credit Usage History.


Video LARP​

Video › LARP tab with Mode Swap the lead | Restyle, Source video 0/1, Character photos 0/3, Quality Best, Matches source · 720p, Keep original audio, + sheet 45 · if needed, and the Make video button

Video › LARP: Mode (Swap the lead | Restyle) · Source video 0/1 · Character photos 0/3 · Quality Best · Matches source · 720p · Keep original audio · + sheet 45 · if needed, the notice "Use only videos and photos of yourself or people who agreed.", and the Make video button.

LARP (formerly Recast) remakes a source video with your character as the lead (Swap the lead) or in a different style (Restyle), keeping the original shots and motion. Add the source video and up to 3 character photos, optionally describe changes, and click Make video. Use only videos and photos of yourself or of people who have agreed. For the full steps, see Video Tab — LARP.

Video AI Edit​

Video › AI Edit tab with Source Video · required, Gemini Omni 1.1 Flash, Keep original, and the AI Edit button

Video › AI Edit: Source Video · required · Gemini Omni 1.1 Flash · Keep original, and the AI Edit button.

AI Edit changes an existing video from a written instruction. Upload the video with + Source Video · required, describe what to change ("Describe what to change."), and click AI Edit. Keep original keeps the source video's original audio.


Video Upscale​

Video upscaling improves the resolution and visual quality of an already-generated video.

Video › Upscale tab with Source Video · required, the RealESRGAN model, the 2× scale button, and the Proceed button

Video › Upscale: Source Video · required · RealESRGAN · 2×, and the Proceed button.

  1. In Direct mode, select the Video tab and open the Upscale sub-tab.
  2. Upload the video with + Source Video · required.
  3. Check the AI model (default RealESRGAN) and the scale (default 2×); click either chip to choose another model or scale.
  4. Click Proceed. The upscaled video is processed and saved as a new file alongside the original.

Notes:

  • The credit cost is shown on the Proceed button.
  • Processing time varies depending on video length and resolution.
  • The original video is preserved and not replaced.

Next Steps

  1. Audio — Background music, lip-sync, and other audio
  2. Editor — Create presentations using your videos
  3. Social Media — Publish results to your social accounts