Aspect ratio, size and quality answer different questions
Aspect ratio describes an image’s shape: square, portrait or landscape. Size describes its output tier, while quality is a separate setting where the selected model offers it. Decide the placement before generating. A product listing, a vertical social image and a wide website banner need different compositions even when they show the same object. Asking for a “high quality image” alone does not tell the model where the product should sit.
According to Google’s Nano Banana developer overview, native aspect-ratio choices support composing images for different formats. That is general background rather than a SeaDance specification. Use the supported options below for this site. The image controls are separate from video generation: SeaDance videos currently use only 16:9, so a portrait still does not establish that the video workflow can return a portrait clip.
Supported choices in the SeaDance image workflow
The table lists the ratios accepted by the current SeaDance image workflow, grouped by model family. Nano Banana Pro and GPT Image 2 use the site’s common ratio list, including 21:9; the other families have their own lists. An auto choice asks the model to choose composition rather than fixing a ratio yourself. Use a fixed ratio when a particular placement needs a predictable shape.
| Image model family | Accepted aspect-ratio options | Size and quality choices |
|---|---|---|
| Nano Banana 2.1 | 1:1, 1:4, 1:8, 2:3, 3:2, 3:4, 4:1, 4:3, 4:5, 5:4, 8:1, 9:16, 16:9, 21:9, auto | 1K, 2K, 4K; fixed quality |
| Nano Banana Pro | 1:1, 16:9, 9:16, 4:3, 3:4, 21:9, 3:2, 2:3 | 1K, 2K, 4K; fixed quality |
| GPT Image 2 | 1:1, 16:9, 9:16, 4:3, 3:4, 21:9, 3:2, 2:3 | 1K; fixed low quality |
| GPT Image 2.5 Flare and Sunburst | auto, 1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9 | 1K, 2K, 4K; low, medium, high |
Choose a composition for the placement
For an e-commerce main image, 1:1 is suitable when the destination calls for a square product view. Keep the object recognizable with room around its silhouette. Use a clear background and avoid decorative objects that imply accessories are included when they are not. The destination’s own listing rules still need to be checked; a model’s ability to generate a square image does not establish marketplace compliance.
For a vertical social image, 9:16 is suitable when the design needs a tall canvas. Ask for the product in a clear part of the frame and leave intentional space where your design needs it. For a horizontal website visual, 16:9 suits a conventional landscape layout. A 21:9 option suits a wider banner when a subject on one side and open space on the other serve the layout. These are composition recommendations rather than guaranteed platform dimensions.
A more extreme strip can be suitable for a decorative treatment where the model supports it, but it can also make a detailed product harder to recognize. Decide whether the image needs to show an entire object or only a material detail. Do not choose the widest ratio simply because it is available. The visual purpose should explain the shape, and the prompt should identify which parts of the subject must remain visible.
Text-to-image credit prices by family
These prices are for one text-to-image output without reference uploads. Selecting a different supported ratio does not add a ratio surcharge to this calculation. Size and, for GPT Image 2.5, quality determine the listed charge. Flare and Sunburst have identical current credit prices here. Equal prices should not be interpreted as proof of identical visual output on your material.
| Model | Quality | 1K | 2K | 4K |
|---|---|---|---|---|
| Nano Banana 2.1 | Fixed | 1 credits / $0.20 | 2 credits / $0.40 | 3 credits / $0.60 |
| Nano Banana Pro | Fixed | 2 credits / $0.40 | 2 credits / $0.40 | 3 credits / $0.60 |
| GPT Image 2 | Fixed low | 1 credits / $0.20 | Not offered | Not offered |
| GPT Image 2.5; either variant | low | 1 credits / $0.20 | 1 credits / $0.20 | 1 credits / $0.20 |
| GPT Image 2.5; either variant | medium | 1 credits / $0.20 | 1 credits / $0.20 | 2 credits / $0.40 |
| GPT Image 2.5; either variant | high | 2 credits / $0.40 | 4 credits / $0.80 | 6 credits / $1.20 |
All generation dollar estimates allocate prepaid credits at Basic monthly: $29.90 divided by 149 credits, approximately $0.200671 per credit. They are rounded to cents after calculating the whole request. They are not separate checkout prices. Other plans have slightly different purchase rates; compare the pricing page before purchasing. New accounts start with 0 credits.
Reference-based images have a separate quote
If you already have a product photo, Image to Image is suitable for creating a new image from supplied material. Its input count can affect the quote, so do not reuse a text-only estimate blindly. Nano Banana 2.1 counts uploaded reference images in its calculation; GPT Image 2 and GPT Image 2.5 include the first reference and charge for additional inputs. Integer rounding can make different underlying costs show the same credit amount.
As an example of the mode difference, GPT Image 2.5 Flare at 1K high quality uses 3 credits / $0.60 with one reference. Compare that with its text-only row before submitting. Use the form’s quote for the selected count and settings. The reference controls are useful for grounding the subject, but they do not guarantee exact packaging text or a perfect copy of every visible detail.
Write the layout into the prompt
For a square product view, describe the object, the background and how much surrounding space you want. For a portrait social image, explain whether the subject belongs near the center or lower in the composition. For a wide banner, specify the side that holds the product and the side that should remain visually quiet. This makes the ratio part of a layout decision rather than an isolated technical selection.
Start with Text to Image when you are exploring a fictional scene or have no source photo. Use the product photo workflow when the aim is a recognizable product presentation. Inspect text, materials, proportions and unwanted props after delivery. A higher size or quality setting does not remove that review step. Failed generations return their deducted credits; purchase-level questions follow the refund policy.
Frequently asked questions
Does changing the ratio change the credit price?
There is no separate aspect-ratio surcharge in the current calculation. Check model, size, quality and reference count, which can change the request quote.
Which shape suits a vertical social image?
9:16 is suitable when a tall composition fits the intended placement. Choose it explicitly and describe where the subject and any open design space should appear.
Can I use 21:9 for a wide banner?
Yes, it is accepted in the current SeaDance image workflow for the listed families. Compose for the wide layout and inspect the resulting subject framing before using it.
Do all models offer a quality selector?
No. GPT Image 2.5 offers low, medium and high. The other families use their fixed quality behavior; GPT Image 2 is pinned to its 1K low-quality route.
Will a portrait still produce a portrait video?
No. The video workflow currently outputs only 16:9. Image aspect-ratio choices describe still-image generation and do not add new video output formats.
Sources
Related articles
Choose the subsequent product video · Compare image model workflows · Understand image and video credits
