A47 MediaHelp CenterOpen A47
Browse help topics
Help CenterVideos

Videos

Generate, refine, and export videos.

Why is my generation taking longer?

What affects wait times and what to do while a creation is processing.

Generation time depends on the media type, model, duration, resolution, and current provider capacity.

  • Keep the page open while the creation is processing; its status updates in the results area.
  • Avoid submitting the same request again, because each request uses usage.
  • If a creation fails, the app records the result and applies any available refund automatically.

View video creations

How do I generate an AI video?

Create a video from a prompt or reference image, then review the result.

  • Open Create, select Video, and describe the scene, subject, movement, lighting, and camera direction you want.
  • Add a reference image when you need visual consistency, then choose the available duration and resolution options.
  • Review the price shown in the form before generating.

Your result stays in the studio while it processes and appears in Library when complete.

  • Seedance 2.5 (BytePlus) — 4–30s, 480p / 720p / 1080p. multimodal references (images, video, audio), native audio
  • Seedance 2.0 (BytePlus) — 4–15s, 720p / 1080p / 4k. multimodal references (images, video, audio), native audio
  • Gemini Omni Flash (Google) — 10s, 720p. ~10s clip with native audio, 720p
  • HappyHorse 1.1 T2V (DashScope) — 2–15s, 720p / 1080p. HappyHorse 1.1, #1 ranked, native audio, cinematic quality
  • HappyHorse 1.1 I2V (DashScope) — 2–15s, 720p / 1080p. HappyHorse 1.1, animate any image with native audio
  • HappyHorse 1.1 R2V (DashScope) — 2–15s, 720p / 1080p. HappyHorse 1.1, multi-reference character-consistent generation
  • Wan 3.0 T2V (DashScope) — 2–30s, 720p / 1080p. text to video with native audio, 5000 char prompts
  • Wan 3.0 I2V (DashScope) — 2–30s, 720p / 1080p. image to video, animate any image
  • Wan 3.0 R2V (DashScope) — 2–30s, 720p / 1080p. reference to video, character-consistent generation
  • Kling 3.0 (Kling) — 3–15s, 720p / 1080p / 4k. first+last frame, optional native audio, 720p/1080p/4K
  • Kling 3.0 Turbo (Kling) — 3–15s, 720p / 1080p. faster generation, always-on native audio; no first/last frame
  • Kling 3.0 Omni (Kling) — 3–15s, 720p / 1080p / 4k. scene and style reference images, first+last frame
  • Kling Motion Control (Kling) — 3–30s, 720p / 1080p. drive a character image with a motion reference video
  • MiniMax H3 Max (MiniMax) — 2–15s, 480p / 768p / 1080p. prompt, one still, or first + last frame; native audio, up to 1080p
  • MiniMax H3 Reference (MiniMax) — 2–15s, 480p / 768p / 1080p. up to 9 images, 3 clips and 3 audio files cited in the prompt; native audio
  • Motion transfer (Genjutsu) — 4–30s, 480p / 720p / 1080p. copy the motion from a clip onto pictures you upload
  • Object swap (Genjutsu) — 4–30s, 480p / 720p / 1080p. swap an object in a clip using pictures you upload
  • Restyle (Genjutsu) — 4–30s, 480p / 720p / 1080p. redraw a clip in a style you pick
  • Motion transfer Light (Genjutsu) — 4–15s, 480p / 720p / 1080p. Genjutsu Light — copy the motion from a clip of up to 15 seconds onto pictures you upload
  • Object swap Light (Genjutsu) — 4–15s, 480p / 720p / 1080p. Genjutsu Light — swap an object in a clip of up to 15 seconds using pictures you upload

Create a video

Which video model should I choose?

Compare reference inputs, modes, durations, resolutions, and audio controls before generating.

Video models support different kinds of input. Some start from text, some require a source image or reference images, and others accept frames, audio, or a motion video. Select the model whose input type matches your goal; the form only shows controls supported by that model. The current capability guide is:

  • Seedance 2.5 (BytePlus) — Optional multimodal references (images, video, or audio); optional first/last frames; Aspect ratio: 16:9 / 4:3 / 1:1 / 3:4 / 9:16 / 21:9 / adaptive; Frame-guided aspect ratio follows the first/last frame; Duration: 4–30s; Resolution: 480p / 720p / 1080p; Up to 2 videos per request; Output sound: on by default; Prompt enhancement available
  • Seedance 2.0 (BytePlus) — Optional multimodal references (images, video, or audio); optional first/last frames; Aspect ratio: 16:9 / 4:3 / 1:1 / 3:4 / 9:16 / 21:9 / adaptive; Duration: 4–15s; Resolution: 720p / 1080p / 4k; Up to 2 videos per request; Output sound: on by default; Prompt enhancement available
  • Gemini Omni Flash (Google) — Optional images in Generate; required video in Edit video; Aspect ratio: 16:9 / 9:16; Duration: approximately 10s, set by the model; Resolution: 720p; Up to 1 video per request; Audio generated automatically; no audio toggle; Prompt enhancement available
  • HappyHorse 1.1 T2V (DashScope) — No reference media; Aspect ratio: 16:9 / 4:3 / 1:1 / 3:4 / 9:16; Duration: 2–15s; Resolution: 720p / 1080p; Up to 2 videos per request; Prompt enhancement available
  • HappyHorse 1.1 I2V (DashScope) — Required source image; Aspect ratio follows the source image; Duration: 2–15s; Resolution: 720p / 1080p; 1 video per request; Prompt enhancement available
  • HappyHorse 1.1 R2V (DashScope) — Required 1–9 reference images; Aspect ratio: 16:9 / 4:3 / 1:1 / 3:4 / 9:16; Duration: 2–15s; Resolution: 720p / 1080p; 1 video per request; Prompt enhancement available
  • Wan 3.0 T2V (DashScope) — No reference media; Aspect ratio: 16:9 / 4:3 / 1:1 / 3:4 / 9:16; Duration: 2–30s; Resolution: 720p / 1080p; Up to 2 videos per request; Native audio is always on; optional custom audio track; Prompt Extend available
  • Wan 3.0 I2V (DashScope) — Required source image; Aspect ratio follows the source image; Duration: 2–30s; Resolution: 720p / 1080p; 1 video per request; Native audio is always on; optional custom audio track; Prompt Extend available
  • Wan 3.0 R2V (DashScope) — Required 1–5 reference images; Aspect ratio: 16:9 / 4:3 / 1:1 / 3:4 / 9:16; Duration: 2–30s; Resolution: 720p / 1080p; 1 video per request; Prompt Extend available
  • Kling 3.0 (Kling) — Optional first/last frames; Aspect ratio: 16:9 / 9:16 / 1:1; Duration: 3–15s; Resolution: 720p / 1080p / 4k; Up to 1 video per request; Output sound: off by default; Prompt enhancement available
  • Kling 3.0 Turbo (Kling) — No reference media; Aspect ratio: 16:9 / 9:16 / 1:1; Duration: 3–15s; Resolution: 720p / 1080p; Up to 1 video per request; Audio generated automatically; no audio toggle; Prompt enhancement available
  • Kling 3.0 Omni (Kling) — Optional first/last frames; optional scene/style reference images (up to 7); Aspect ratio: 16:9 / 9:16 / 1:1; Duration: 3–15s; Resolution: 720p / 1080p / 4k; Up to 1 video per request; Output sound: off by default; Prompt enhancement available
  • Kling Motion Control (Kling) — Required motion video and appearance image; Aspect ratio follows the motion inputs; Duration: follows the motion video; Resolution: 720p / 1080p; Up to 1 video per request; Keep original sound: on by default; Prompt enhancement available
  • MiniMax H3 Max (MiniMax) — Optional first/last frames; Aspect ratio: 16:9 / 9:16 / 1:1 / 4:3 / 3:4 / 21:9; Duration: 2–15s; Resolution: 480p / 768p / 1080p; Up to 1 video per request; Prompt enhancement available
  • MiniMax H3 Reference (MiniMax) — Optional first/last frames; Aspect ratio: adaptive / 16:9 / 9:16 / 1:1 / 4:3 / 3:4 / 21:9; Duration: 2–15s; Resolution: 480p / 768p / 1080p; Up to 1 video per request; Prompt enhancement available
  • Motion transfer (Genjutsu) — Required clip and 1–8 pictures; Duration: follows the clip (4–30s; longer clips are refused); Resolution: 480p / 720p / 1080p; Up to 1 video per request; Prompt enhancement available
  • Object swap (Genjutsu) — Required clip and 1–8 pictures; Duration: follows the clip (4–30s; longer clips are refused); Resolution: 480p / 720p / 1080p; Up to 1 video per request; Prompt enhancement available
  • Restyle (Genjutsu) — Required clip; pick a style; up to 5 extra pictures; Duration: follows the clip (4–30s; longer clips are refused); Resolution: 480p / 720p / 1080p; Up to 1 video per request; Prompt enhancement available
  • Motion transfer Light (Genjutsu) — Required clip and 1–8 pictures; Duration: follows the clip (4–15s; longer clips are refused); Resolution: 480p / 720p / 1080p; Up to 1 video per request; Prompt enhancement available
  • Object swap Light (Genjutsu) — Required clip and 1–8 pictures; Duration: follows the clip (4–15s; longer clips are refused); Resolution: 480p / 720p / 1080p; Up to 1 video per request; Prompt enhancement available

Compare video options

Tips for creating great AI videos

Write clearer prompts and use references for more controlled results.

  • Be specific in your prompt — describe the scene, subject, lighting, camera angle, and mood, in short clear sentences.
  • Generate a few variations and compare.
  • Creating reference visuals in AI Images first, then using them to guide video generation, gives you far more control over the look.

Create a video