Image-to-video works best when the opening image already contains the composition you want to preserve. The image establishes the subject, setting, proportions, lighting, and visual hierarchy. The prompt then has a simpler job: explain how that image should come to life.
This workflow is useful for product photography, fashion stills, character portraits, architecture, key art, and storyboards. It is not about adding movement everywhere. It is about giving one well-designed frame a controlled next moment.
1. Build a strong first frame
Start with an image that is clear before it moves. The primary subject should be easy to identify, and the framing should already match the delivery format as closely as possible.
For a product shot, make sure the product silhouette is unobstructed. For a character, keep the face, wardrobe, and body position readable. For a location or interior, protect the lines and major depth cues that make the scene feel stable.
Before uploading, check four things:
- The main subject has enough contrast against the background.
- The composition leaves room for the planned camera move.
- Important product or character details are visible at a useful size.
- The source image does not depend on small readable text to communicate the idea.
A beautiful image with too many competing subjects can become a difficult animation brief. A simpler image gives the model fewer decisions to reinterpret.
2. Choose one kind of movement
Image-to-video prompts are strongest when one motion idea leads the shot. Choose either camera motion, subject motion, environmental motion, or a small material change as the main event.
Good starting choices include:
- A slow push-in toward a product or portrait.
- A lateral camera track that reveals depth in an interior.
- A small head turn, step, or hand movement from a character.
- Wind moving fabric, foliage, mist, or reflections.
- A single water ripple, light sweep, or rotating object.
Avoid asking for an orbit, a zoom, a scene change, a wardrobe change, and multiple visual effects in the same short clip. A focused first pass makes it easier to know what to improve.
3. Write the motion prompt after the image is selected
Use the image to carry visual identity. Use the text prompt to carry timing and intent. A simple prompt structure is:
- State what should remain stable.
- Describe the primary movement.
- Specify the camera path and framing.
- Add one lighting or material change.
- Define the final composition.
For example:
Keep the product silhouette and packaging geometry stable. The camera makes a slow three-quarter push-in while a soft warm reflection travels across the glass. Fine mist moves behind the product, then the camera settles into a centered hero frame.
The image already provides the bottle, background, and initial composition. The prompt only needs to make the motion, timing, and final frame explicit.
4. Use references for the decision that matters most
An image reference can anchor a subject or product. Add a video reference when the intended timing, camera path, or performance matters more than a still image can show. Add audio direction when the sound needs to support the rhythm of the shot.
Give each reference a role. For example:
- Use an image to preserve a product package.
- Use a source clip to communicate a walking pace or camera track.
- Use a text prompt to define the new environment and lighting.
The reference video generator is useful when a source clip should guide motion. The multimodal video generator is useful when a project needs a combination of image, video, audio, and prompt direction.
5. Match the output format to the destination
Choose the aspect ratio before you write the final prompt. The framing affects camera distance, subject placement, and how much environmental motion can fit into the shot.
- Use 9:16 for mobile-first social placements and vertical product stories.
- Use 1:1 when the product needs a balanced square composition.
- Use 16:9 for website headers, presentations, wider ads, and cinematic scenes.
For a vertical shot, keep the subject's essential movement in the center column. For a wide shot, use the side space to reveal the environment or create a deliberate camera path.
6. Review one variable at a time
The fastest way to improve an image-to-video result is to review it against a short checklist:
- Does the subject remain stable?
- Is the planned movement readable?
- Does the camera move in the intended direction?
- Do the light and materials remain believable?
- Does the final frame serve the intended use?
When a result misses the target, change the prompt or reference for only one of these variables. Do not rewrite every part of the brief after a single run. The strongest workflow produces a sequence of small, informed revisions rather than unrelated variations.
7. Move to final quality after the shot is approved
Use an efficient draft to establish the composition and motion, then increase the output setting once the direction is clear. Seedance 2.0 provides the broadest resolution range for final-quality short clips, including 1080p and 4K. Seedance 2.5 is the better fit when the action needs a longer single generation at 480p or 720p.
The Seedance 2.5 vs Seedance 2.0 comparison explains that tradeoff in more detail.
Start with a reusable shot plan
Open the image-to-video generator when you have a first frame ready. For structured examples, use the prompt library and adapt a product, fashion, architecture, or cinematic shot plan to your own image.