Create Your Video

The most reliable way to produce a video is to control the first frame. Generate an image in Realism, refine it until the composition is right, then animate it.

​

Step 1 — Generate the first frame

Open the image generator and select a model. Compose the shot with animation in mind:

• Camera distance — mid-shot and full-body frames animate more naturally than extreme close-ups
• Pose — a neutral or slightly dynamic pose gives the model room to add motion without distorting anatomy
• Background — simple backgrounds produce cleaner motion separation than busy scenes
• Lighting — soft lighting is more stable across frames than hard dramatic shadows

When you have a frame you’re happy with, save it to your gallery.

​

Step 2 — Open a video session

Go to the video generator and start a new Session. Name it something that matches the concept you’re working on. All generations in this session will be grouped together. See Sessions for how sessions work.

​

Step 3 — Select Image-to-Video

Choose Image-to-Video and select your first frame from the gallery. Resolution must match the source image. If the resolution doesn’t match, the generator will flag it before you proceed.

​

Step 4 — Write a motion prompt

Describe what should move and how. Focus on action and camera, not on describing the scene — the first frame already defines the scene. Good: slow pan right, hair moving in the wind, subtle breathing motion, steady camera Bad: beautiful woman, soft light, photorealistic — this describes the image, not the motion. See Video Prompting for a full breakdown.

​

Step 5 — Generate

Click Generate. Render time is longer than image generation. The clip appears in your session when complete. If the result isn’t right, adjust the motion prompt and regenerate — keep the same first frame. Changing the first frame mid-session makes it harder to compare results.

​

Step 6 — Enhance (optional)

Run the output through Enhanced Video to upscale and sharpen the clip. This uses additional tokens. See Enhanced Video.

​

Tips

• Spend more time on the first frame than on the motion prompt — a strong frame produces better video regardless of prompt quality
• Generate two or three motion prompt variations before changing the first frame
• Keep sessions focused on one concept — start a new session when you switch characters or scenarios
• Five seconds is the maximum duration for image-to-video