Docs/Creative Tools/Video & Animation

Video & Animation

Turn still images into AI-powered video clips and create motion content.

LetzAI's video generation tools let you generate AI-powered video clips from text prompts, still images, keyframes, and multimodal references (images, videos, and audio). Add realistic camera movement, character animation, synchronized dialogue and sound effects, or extend an existing clip into a longer continuous scene.

Getting started with video

You can create videos from the Animate tab, inside LetzAI Chat, or directly on Canvas:

  1. Select a starting point: Upload or select a reference image, a start and end frame, multimodal references (images, video clips, or audio), or start from a text prompt alone.
  2. Choose a video model: Different models specialize in multimodal references, synchronized audio, keyframe transitions, or multi-shot storytelling:
    • Seedance 2.0 & Seedance 2.5 (ByteDance): Flagship multimodal video generation with native synchronized audio (dialogue, ambient sound, and sound effects). Supports text-to-video, image-to-video, first-and-last frame keyframes, and Omni references (up to 12 files on Seedance 2.0 at up to 4K resolution and 15s; up to 50 files on Seedance 2.5 at up to 4K resolution and 30s). Enterprise organizations also unlock Seedance 2.0 Enterprise, Seedance 2.5 Enterprise, and Seedance 2.5 Draft Mode (fast 480p drafts that can be upscaled to 1080p).
    • Kling V4 (Kuaishou): Cinematic 720p or 1080p video (3–30s) with always-on native audio, text-to-video, image-to-video, first-and-last frame control, and automatic shot-splitting when you write your prompt as a shot list.
    • MiniMax H3 & H3 Max (Hailuo): 2K video generation (H3, 5–15s) and fast draft generation (H3 Max, 480p–768p) supporting text-to-video, start/end frames, and multimodal image, video, and audio references with strong subject consistency.
    • Gemini Omni 1.2 Flash (Google): Fast, cost-effective multimodal video generation and prompt-driven video editing (3–15s at 360p, 720p, 1080p, or 4K; up to 10s from a first and a last frame) with native audio and image/video reference support.
    • Flux 3 Video & Flux 3 Max (Black Forest Labs): 720p and 1080p video clips (5–20s) with always-on synchronized audio, supporting text-to-video, image-to-video, and first-and-last frame keyframe animation.
  3. Write a motion prompt: Describe the movement, camera work, and sound you want to see and hear (for example: "Slow tracking shot following her along the rain-slick street as she turns toward the camera, distant traffic and soft rain").
  4. Set duration and resolution: Choose clip length (from 3 to 30 seconds depending on the model), resolution (480p, 720p, 1080p, 2K, or 4K), and aspect ratio.
  5. Generate: Click Animate (or send your prompt in Chat or Canvas) and wait for your video to render.

Visit the Models catalog for a live comparison of every active video model, supported input modes, and credit costs.

Extending Videos, Keyframes, and Multi-Shot

Extending & continuing videos: You can extend an existing video clip or continue directly from its final frame in Chat, on Canvas, or in the Animate tab. With multimodal models like Seedance 2.0, Seedance 2.5, MiniMax H3, or Gemini Omni Flash, pass the existing clip as a video reference (or use its ending frame as the start frame) and write a single continuous take prompt describing what happens next from the final frame. Keep the same subject, wardrobe, lighting, and camera momentum without adding scene cuts so the continuation flows seamlessly.

First and last frame keyframes: Models including Seedance 2.0, Kling V4, MiniMax H3, and Flux 3 Video allow you to supply both a starting image and an ending image. Describe the motion that bridges the two frames, and the model animates a smooth transition from the first keyframe to the last.

Several shots in one clip (Kling V4): Write your prompt as a timed shot list (for example, "Shot 1 (4s): wide shot … Shot 2 (5s): close-up …") and Kling V4 splits the clip into those shots, up to 30 seconds in total. Use Seedance's multi-cut syntax when you want intentional camera cuts on Seedance.

Writing effective motion prompts

Unlike image prompts that describe a static scene, video prompts should focus on what moves and how:

  • Camera movement: "slow zoom in", "pan from left to right", "tracking shot following the subject", "orbiting camera", "handheld drift"
  • Subject motion: "person turns and smiles", "car drives down the road", "bird takes flight", "waves crash on the shore"
  • Environmental motion: "leaves falling", "clouds moving across the sky", "flickering candlelight", "rain falling"
  • Sound & dialogue (on audio-enabled models): Put spoken lines in quotes and name specific ambient sounds or sound effects (for example, "footsteps on wet pavement, low thunder").
  • For continuations vs. new scenes: When extending a clip, describe a single continuous shot continuing from the final frame. Only use multi-cut instructions when you want hard cuts to new angles.

Best practices

Start with a strong reference image. The quality of your starting frame or reference assets directly affects the output. Well-composed, high-resolution images with clear subjects animate best.

Iterate cheaply with Draft Mode. When working with Seedance 2.5 or MiniMax H3 Max, generate fast 480p draft clips first to test your framing, motion, and audio, then upscale your favorite Seedance 2.5 draft take to 1080p.

Match the model to the workflow. Use Seedance 2.0 or 2.5 for rich multimodal references, synchronized dialogue, and 4K/long-form shots; Kling V4 for long takes and multi-shot sequences up to 30s; MiniMax H3 for 2K reference consistency; Flux 3 for up to 20s keyframe takes; and Gemini Omni Flash for fast, budget-friendly multimodal clips.

Credit costs

Video generation is more compute-intensive than image generation and costs more credits per run. Credit costs depend on the model, duration, resolution, and whether native audio is enabled. Check the credit cost indicator before generating, or ask the assistant in Chat or Canvas to compare options.

Ready to bring your scenes to life? Head to the Animate tab, Chat, or Canvas and create your first video.

Can't find what you're looking for?

Our team is here to help you with any questions.

Contact Support