AI images · ByteDance

Seedream 5.0 Flash — Fast image generation and reference edits

Find the moment. Try the next angle.

An ollie on a sunny street, a dancer crossing a station floor, friends reacting in the background: give your idea a visible action. Flash creates one JPEG or PNG in 1K, 1.5K or 2K from a prompt or reference images.

KEEP GOING — a street-skate campaign
Text → Image
2K16:9

KEEP GOING — generated with Seedream 5.0 Flash

A 2K frame with people in action and text painted into the setting. Open the examples for the original image and prompt; video generation is a separate step.

Open image workspaceView pricingPrompt examples

Explore an active scene

Describe a gesture that can be read in a single frame. Add a believable supporting pose, readable face and a setting with depth.

Try a different direction

Change the framing, light or expression in a new request. Flash returns one image at a time, so you can select the composition to develop.

Build a campaign around people

Place short exact copy on a real surface within the scene. Use the character, location and lettering together to give the image a purpose.

Guide the next edit

Bring up to 10 references for the person, clothing or location. Choose 1K, 1.5K or 2K and download the single result as JPEG or PNG.

Seedream 5.0 Flash — current image pricing

The exact current price is shown before you confirm generation. Review the current quote before each request.

View full pricing

1K image

$0.03

Single generated still

1.5K image

$0.03

Single generated still

2K image

$0.03

Single generated still

Generation example: 1 image · 2K · Text to image — $0.03

Check the current quote for your exact settings before generating.

People, motion and words inside the scene

Two original 2K frames generated with Seedream 5.0 Flash. Open an image to inspect the original and copy the exact English prompt. These are independent text-to-image examples.

Check the faces, hands, lettering and framing before saving the original image. Use the still as a reference for a separate image-to-video generation.

Create with Seedream 5.0 Flash

Explore image-to-video with Seedance 2.5

Seedream 5.5

Refine a scene with reference-guided edits, or create a transparent PNG asset. One image per request.

Discover Seedream 5.5

Seedream 5.0 Lite

Explore related character, product or storyboard images in a batch before choosing the frames to keep.

Discover Seedream 5.0 Lite

Seedream 5.0 Pro

Create one detailed 2K portrait, campaign still or infographic, with JPEG or PNG output.

Discover Seedream 5.0 Pro

Prompt examples

Try an original English brief for a human scene. Adapt the characters, action and setting, then check hands and lettering in the result.

Prompt examples

How to prompt Seedream 5.0 Flash

Start with an action you can see in one instant. Add the body position, expression, camera angle and light, then describe the surroundings and any text.

A plausible pose is more useful than an extreme stunt. Check hands, feet and props before using the image as a video reference.

Source: Official model guide

Work with reference images

Write one visible action

Choose the gesture, body position and expression before adding camera and lighting details. Keep hands, feet and props easy to distinguish.

Compare one change

Run a new request for a different angle or atmosphere. Keep your chosen brief nearby so you can judge whether the change helps the scene.

Use focused references

Identify which source defines the person, wardrobe or location. Image editing accepts up to 10 references and returns one image.

Select the opening frame

Save the image with a clear silhouette and room for movement. Take that still to a separate video model when you are ready to plan motion.

Prompt writing tips

  • Give the scene one action and a clear focal person.
  • For movement, describe the supporting foot, balance and readable hand positions.
  • Put exact words in quotation marks and place them on a physical surface.
  • Change one part of the brief per request so you can compare the direction.

Seedream 5.0 Flash — prompt examples

Original English prompts for human scenes and campaign stills. Copy a brief, then adapt the setting, characters and exact wording to your project.

Open image workspace

City skate — KEEP GOING

Text → Image

A vibrant cinematic live-action sports-campaign still, 16:9 widescreen. A fictional female skateboarder around twenty-six, wearing a burnt-orange windbreaker, loose navy trousers and simple cream sneakers, is captured at the clean apex of a small controlled ollie over a low curb on a sunny Lisbon street. Her whole body and the complete skateboard are visible, both feet positioned convincingly above the deck, natural hands balancing the movement, focused face clearly visible. Use a low three-quarter camera angle and a believable body pose rather than an extreme stunt. Two adult friends react naturally on the sidewalk behind her, with a cafe terrace and a real city street receding into the distance. A large painted wall advertisement on the left reads exactly "KEEP GOING" in two crisp, oversized dark-blue lines on a pale warm-yellow background; the lettering is physically painted on the wall and follows its perspective. Late afternoon sunlight creates long diagonal shadows, tactile stucco and real skin texture. Crisp subject with a small amount of natural motion blur in the moving background, 28mm cinema lens, rich but controlled colors, energetic editorial composition with real depth. Keep the skateboarder away from the frame edges. A candid decisive moment from a premium commercial film, one cohesive image with no added title, collage or border.

Station dance — NEXT STOP ANYWHERE

Text → Image

An arresting live-action cinematic still, 16:9 widescreen. In a grand early-twentieth-century European railway concourse, a fictional adult dancer in their late twenties, dressed in a flowing cobalt-blue overshirt, ivory T-shirt, charcoal trousers and worn white sneakers, is caught mid-turn during an impromptu dance. The full body is visible with a convincing grounded pivot, one foot on the stone floor, natural arms extended at different heights, relaxed readable hands and a face lit by a shaft of late-afternoon sun. Several adult commuters continue walking in the deeper background, some glancing toward the dancer. Tall windows, iron arches and a large station clock create a rich architectural backdrop. A physically installed departure-board-style sign in the middle background reads exactly "NEXT STOP" on its first line and "ANYWHERE" on its second line, in clean cream block lettering on a deep-blue panel. Clothing has a light natural motion blur at its edges while the face and torso remain sharp. Warm sun against cool interior shadow, long floor reflections, atmospheric depth, 35mm cinema lens, restrained grain, beautiful believable skin and fabric. Frame the scene like the opening of a moving human story with space for a slow dolly-in, not a studio dance portrait. One coherent realistic location, no floating lettering or graphic overlay.

Tips and boundaries

Best practices, common fixes, and important limitations to help you get the strongest results with Seedream 5.0 Flash.

What works best

  • Human action and expressive campaign scenes
  • Framing, lighting and wardrobe variations
  • A clear opening still before a video test

Hard limits to keep in mind

  • One generated image per request
  • Use one request per variation; Flash does not return a sequential image set.
  • Review rendered text and fine details before publishing.

Specifications

Formats, settings and limits to check before creating.

Price / image

$0.03/image

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

Image settings

Details
  • Generate from a prompt or edit reference images.
  • Choose 1K, 1.5K or 2K.
  • Up to 10 reference images
  • One generated image per request
  • Download JPEG or PNG.

Responsible image creation

Built-in safeguards and best practices for responsible creation with Seedream 5.0 Flash.

  • Use source images you have permission to edit.
  • Respect consent and rights when depicting real people.
  • Provider safety rules apply to prompts and reference images.

FAQ

When should I choose Seedream 5.0 Flash?

Flash is the fast variant of the Seedream 5.0 family. Use it for visual exploration: test a character pose, framing or campaign direction through individual image requests.

Can Flash return several variations in one request?

Flash returns one generated image per request. Launch a new request for another variation, or choose Seedream 5.0 Lite for image sets.

Can I edit an existing character image?

Yes. Use up to 10 reference images, identify the face, outfit and framing to preserve, and describe the change. Review the resulting identity and details before keeping it.

Which outputs are available?

Choose 1K, 1.5K or 2K and download JPEG or PNG. For a native transparent-background asset, choose Seedream 5.5.

Can a Flash image become the opening of a video?

Yes. Save the chosen still and use it as a reference in a compatible video model. Plan motion separately; Flash itself produces a still image.

How do I check the current price?

Choose your resolution and references in the image workspace. MaxVideoAI shows the quote before you confirm each request.