Image-to-image editing
Same scene. A new direction.
An emerald jacket becomes ivory. Then the warm theatre light becomes cool blue. Compare each edit with its actual source, using the slider or your keyboard.



Create. Refine. Make it yours.
Create lifelike scenes, refine the details and build ready-to-compose assets with Seedream 5.5 on MaxVideoAI.
Guide successive edits to clothing, light and the details that matter.
Put text into signs, posters and the scene itself.
Explore natural skin, tactile fabrics and lifelike lighting.
Generate native transparent PNGs, ready to compose.
Image-to-image editing
An emerald jacket becomes ivory. Then the warm theatre light becomes cool blue. Compare each edit with its actual source, using the slider or your keyboard.


Text in an image
A theatre entrance, a passing character and a short message inside the scene. Give the text a surface, a position and a reason to be there.
The prompt requests “ONE NIGHT ONLY”. This selected image shows the result; check every letter before publishing.
Cinematic image generation
A runner catches her breath. Examine the skin, flyaway hair and textured fabric in this close portrait, then open the original at full size.
Seedream 5.5 · Generation · 2K · 3:4 · JPEG

This selected result illustrates one portrait brief. Inspect facial details and anatomy before using any generated image.
Native transparent PNG
Generate an isolated subject with a real alpha channel. Change the preview background to inspect the edges, then open the unchanged original PNG.

The dancer is the same image on every background. Transparency was verified in this original and its WebP display copy; no background-removal effect is applied here.
Prompt guide
Describe the action, framing and light. Put exact text in quotes. Use the example below, or shape your idea with Milo.
The idea behind this image
A woman in a burgundy coat emerges from a Paris metro at blue hour. She turns towards someone outside the frame. Rain catches cyan and amber light. An illuminated street poster reads “MAKE YOUR MOVE”.
Shortened for clarity. Use or copy the complete original prompt below the image.Say which image defines the person, clothing or location. Use up to 10 reference images to guide one output.
Use the previous result as the next reference. Name what must stay, then ask for one change. For an opaque scene, keep Background set to Opaque.
Select PNG and Transparent in Create. Describe the isolated subject and leave space around its edges. JPEG cannot preserve transparency.
Formats, settings and limits to check before creating.
| Setting | In MaxVideoAI |
|---|---|
| Text-to-Image | Supported |
| Image-to-Image | Supported |
| Resolution options | 1K / 1.5K / 2K |
| Aspect ratios | auto / 1:1 / 4:3 / 3:4 / 16:9 / 9:16 / 3:2 / 2:3 / 21:9 |
| Output format | JPEG / PNG |
| Reference images | Up to 10 |
| Images per request | 1 |
| Background | Auto · Opaque · Transparent (PNG) |
Settings available in Create. The table describes MaxVideoAI’s implementation of this model. Source: MaxVideoAI
FAQ
Seedream 5.5 is an AI image generation and editing model developed by ByteDance. On MaxVideoAI, use text or up to 10 reference images to create portraits, cinematic scenes, campaign images and transparent PNG assets. Each request returns one still image.
ByteDance presents 5.5 as an advance in successive edits, realism and text rendering, with native transparent PNG output. In MaxVideoAI, 5.5 also offers 1K, 1.5K and 2K instead of Pro’s 2K-only output. Both accept up to 10 references and return one image per request. The examples above show selected results, not a controlled head-to-head benchmark.
Use the selected image as a reference. State what to preserve and ask for a focused change, then inspect the new result before using it for the next edit.
Yes. Choose PNG output and the transparent-background setting. JPEG is suitable for an opaque scene.
MaxVideoAI offers 1K, 1.5K and 2K, with JPEG or PNG output. Image editing accepts up to 10 references and returns one image; use Lite when you need image sets.
Save the still and choose a video model for a separate generation. A readable silhouette, clear hands and space around the subject make the opening frame easier to plan.
MaxVideoAI uses pay-as-you-go credits, with no required subscription. Choose your resolution and references in Create, then review the current price before generating. Each new generation or edit has its own quote.