When you animate a single photo, the video model has to guess where the scene should end. A start image and an end image give it two visual targets. Your prompt can then focus on what happens between them.
Here’s a simple example: a closed gift box opens to reveal a bottle.
1. Prepare two frames that belong to the same shot
For the start frame, use an image of the closed box on a table. For the end frame, show the same box open, with the bottle visible inside.
Keep the camera angle, crop, background, and lighting as similar as possible. If the box moves across the table or changes color between frames, the model has to reconcile those differences while also animating the lid. Start with one change you can describe clearly: the lid opens.
2. Describe the motion between the frames
Upload both images to Hey Dream AI Free Image to Video. Then try this prompt:
The box stays centered on the table. Its lid slowly swings upward on the rear hinge, revealing the bottle inside. The camera remains fixed. Keep the background and lighting consistent. End with the lid open and the bottle still.
Notice that the prompt specifies what moves, what stays still, and how the shot ends. Those details give the model more direction than a phrase like “cinematic product reveal.”
3. Change one thing after each preview
Watch the clip and identify the first thing that breaks the illusion:
- The box slides or changes shape: Check whether it occupies the same position and has the same proportions in both images.
- The camera drifts: Remove any camera movement from the prompt and ask for a fixed shot.
- The bottle appears before the lid opens: Make the sequence explicit: “The lid opens first, then the bottle becomes visible.”
Change one variable per attempt so you can tell whether the frame pair or the prompt was causing the problem. Once the reveal is readable, you can try a slow camera push or a lighting change in a separate version.
Top comments (0)