Use 9:16 for a tall, vertically viewed composition, 16:9 for a wide scene, and 1:1 for a square placement. Choose the shape that keeps the essential action visible in the destination player. If you need several versions, check whether the whole moving subject fits each crop before reusing one clip.
For creators using Seedance 2.5, this is a framing and editing guide, not a promise that every model exposes the same output settings. Select the ratio in the active generator when available; writing “vertical” in a prompt alone does not set the file dimensions. The examples below are untested planning prompts, not results generated for this article.
Choose a ratio by what must remain visible
| Ratio | Useful starting point | What to protect |
|---|---|---|
| 9:16, portrait | A vertical full-screen clip, one presenter, a product held above a table | Hands, handles and sideways movement near the narrow edges |
| 16:9, landscape | A wide website player, presentation, environment or two-person scene | The relationship between separated subjects and the surrounding setting |
| 1:1, square | A square feed placement or compact product tile | A complete central subject with enough space for its movement |
These are composition choices, not universal upload requirements. Check the actual placement: a feed preview and a full-screen player may show different crops. A square file is not automatically the best choice for every feed, and a mobile audience does not always require vertical video.
If you need help finding the controls, use the Seedance2Video beginner's guide. Here, the next decision is whether the intended shot survives a different shape.
Aspect ratio is not resolution
Aspect ratio is width divided by height. Resolution describes the pixel dimensions. A 1920 × 1080 image and a 1280 × 720 image both have a 16:9 ratio, even though the first contains more pixels.
Those dimensions are arithmetic examples, not export-size promises for Seedance2Video. A higher-resolution source gives an editor more pixels to work with, but it cannot restore an object removed by a crop.
How much do you lose when cropping 16:9 to 9:16?
A full-height portrait crop from a wide frame keeps much less of the original picture than many people expect.
Start with an illustrative 1920 × 1080 frame. Keeping the full 1080-pixel height:
- 9:16 crop width: 1080 × 9 ÷ 16 = 607.5 pixels, approximately 608 after rounding.
- Share of original width retained: 607.5 ÷ 1920 = about 31.6%.
- 1:1 crop width: 1080 pixels, retaining 56.25% of the original width.
The rounded portrait dimensions are approximate; an editor may adjust dimensions to meet its export constraints. Because the height stays fixed in this example, the retained area has the same percentage as the retained width.
For a centered portrait crop, roughly the middle third of the original wide frame remains. A mug at the center may fit, while the hand entering from the far left disappears. Enlarging that crop to a larger vertical export does not recreate the discarded action.
The reverse has a similar cost: a full-width 16:9 crop from a 9:16 portrait frame retains only about 31.6% of its original height. A face may survive while the hands and product below it are lost.
Test the movement, not just the opening frame
Before cropping, identify the complete area occupied by the essential action over time. That includes the product, fingers, props and any camera movement that changes their position.
For example, imagine a mug initially centered in a wide shot. A hand lifts it toward the upper right. The first frame fits a portrait crop, but the handle crosses the crop edge halfway through. The correct decision depends on the entire lift, not the thumbnail.
Use this check in your editor:
- Put the original clip in a sequence with the destination ratio. Keep its proportions locked.
- Position the crop around the essential subject.
- Watch from start to finish with the crop boundaries visible.
- Pause at the widest hand position, the highest lift and any camera change.
- Check whether captions or interface overlays will cover the remaining action in the destination preview.
An animated crop can follow one moving subject when there is enough source image available. It cannot fit two widely separated people into a narrow frame at the same moment. It may also introduce a distracting pan, so review it as part of the edit.
Crop, fit or create another composition?
| Situation | Practical choice | Tradeoff |
|---|---|---|
| One compact subject stays within the new boundaries | Crop the existing clip | You discard surroundings and some source pixels |
| Edge details or separated subjects are essential | Fit the complete image inside the new canvas | You need to design the unused space around it |
| The scene becomes tiny, awkward or incomplete | Prepare a separate composition | You need another creative pass and, if generating again, another attempt |
Do not stretch a wide video into a tall frame. Stretching changes proportions: a round mug can become oval and faces can become unnaturally narrow.
Fitting the whole picture is useful when completeness matters more than filling every pixel. For example, you can place a wide demonstration inside a vertical layout with explanatory text above or below it. Preview the subject at the actual viewing size; preserving everything is not useful if the important detail becomes unreadable.
Plan one product idea in three compositions
The concept is the same in each version: a hand lifts a ceramic mug so the viewer can see its handle and surface. Keep the mug's identity and lighting consistent, but change the staging to fit the frame.
If you use a reference image, prepare its intended crop first. Leave room for the whole handle and the hand's path. A tightly cropped source does not document the missing background or hidden surfaces; asking the model to reveal them requires it to invent information.
These prompts are untested examples. They describe creative intent and do not guarantee framing or product fidelity. Set the matching ratio separately in the generator if offered.
9:16: keep the lift within a narrow column
A single ceramic mug sits near the center of a tall composition on a simple tabletop. Keep the complete handle visible with breathing room above the mug. A hand enters from the lower right, lifts the mug a short distance upward and pauses. Keep the fingers and mug inside the frame throughout. Locked camera, soft window light, uncluttered background. No generated text.
Review the highest point of the lift and the elbow or fingers near the side edge. Do not assume that centering the mug at the start keeps it centered in motion.
16:9: give the action horizontal space
A ceramic mug sits slightly left of center on a wide tabletop. Leave a quiet background area to the right for a headline added later. A hand enters from the left and lifts the mug gently. Keep the complete mug, handle and fingers visible throughout. Locked camera, soft window light, minimal background detail. No generated text.
This arrangement uses the width for both movement and edited copy. It is a separate layout: cropping it vertically may remove the hand's entrance or the space reserved for text.
1:1: keep the action compact
A ceramic mug is centered in a balanced square composition. Leave space around the complete handle and above the mug. A hand lifts it gently and holds it near its starting position. Keep the full mug and fingers visible. Locked camera, soft window light, plain tabletop and background. No generated text.
The limited movement makes the composition easier to evaluate. If the hand still leaves the frame, adjust the source framing or simplify the action before adding more prompt instructions.
Finish each version for its destination
Add captions and exact wording in your editor. There is no universal safe margin for every player: controls, captions and other overlays vary by placement. Check the destination preview before considering the edit finished.
For each version, confirm:
- The complete product and required action remain visible throughout.
- The crop has not removed an essential relationship, gesture or detail.
- The picture retains its original proportions.
- Text is readable at the intended viewing size and avoids visible overlays.
- The final frame still works as an ending, rather than cutting off the movement.
Before the next generation, plan the shot in the Seedance 2.5 workspace with the destination shape in mind. Check the exact selected model and available controls; Seedance 2.5 and Seedance 2.5 Lite are separate selections. No new generation is needed simply to evaluate a crop in an editor.
For a campaign that needs consistent references across versions, continue with the reference-to-video workflow. Before using the finished clip, apply the broader product-video QA checklist.
