Introduction
Seedream 5.0 Lite Image to Image takes one to ten reference images plus an instruction and returns a single frame at the preset you name. That combination is what makes it useful for series work: the same character, restyled, always delivered in the same frame shape.
Seedream 5.0 Lite is the preset-sized tier of ByteDance's Seedream line: eight 2K output presets from 2048×2048 square to a 3136×1344 strip, 3,000-character prompts, JPEG or PNG, and editing that accepts up to ten references. Uncensored.
Choose a Seedream 5.0 Lite mode by starting material
| Starting material | Mode |
|---|---|
| A written art direction or design brief | Seedream 5.0 Lite Text to Image |
| An approved source image to revise | Seedream 5.0 Lite Image Editing |
Key features & highlights
Layout-aware design exploration
Seedream 5.0 Lite is useful for testing a visual hierarchy across posters, feeds, banners and illustrative layouts. Start with the intended placement and message, then judge the result for readability before adding decorative complexity.
Illustration and graphic treatments
The model can support both graphic and illustrative directions alongside photographic concepts. Name the desired treatment and the elements that must remain legible so review does not collapse into a vague preference.
Create from a brief or revise a source on purpose
Text-to-image begins with an original composition. Image editing begins with supplied material; use it when the task is a targeted visual change rather than another attempt to describe the source from memory.
Request parameters
| Parameter | Required | Description |
|---|---|---|
prompt | Yes | 3 to 3,000 characters. Name what each reference contributes and what must survive — pose, line weight, grade — and remember the result is restaged into the preset you chose. at least 3 characters · at most 3000 characters |
image_urls | Yes | One to ten publicly fetchable jpeg, png or webp URLs for images you hold the rights to. at least 1 items · at most 10 items |
size | No | The delivered frame, chosen from eight 2K presets. |
output_format | No | jpeg by default, or png. png is lossless and the safer handoff into further editing; it changes the delivered file, not the image that was generated. |
Pricing
$0.035 / image
Use cases
Refine a supplied visual
Use this mode when one to ten existing images should carry the subject, costume or style into the result. List them in image_urls in the order your prompt names them, say what each contributes and what must stay, and pick the preset that matches the source's shape.
Ten-reference restyling
Send a character sheet, two costume references and a background plate to the editing endpoint, name what each contributes, and get one frame back at the size you asked for. Useful for keeping a recurring character consistent across a series.
Costume character illustration sets
Keep a character from an approved frame and redraw the costume — leotard, tights, ears and tails — from a second reference, delivered at 20482048 or 17282304 so the whole set shares one frame shape.
Print comps and layout mocks
Send an approved layout frame and change one element — the background, the costume, the palette — while the composition stays. Set size to the comp's preset, 23041728 or 17282304, so the frame is not reshaped.
Prompt guide & pro tips
Pick the size the placement actually needs
Choose from the eight presets before writing, because the composition is built for that frame. Square for feed assets, 23041728 for a print comp, 31361344 for a hero strip.
Poll the task and keep the brief
Create the task, poll for a terminal state and fetch the image from platform storage. There is no seed field, so keep the references and the instruction with any frame you want to extend into a series.
Adapt the included example
Start with the published example using output_format: png, size: 2048*2048. Replace the subject and creative direction while keeping the input contract for this mode intact.
Keep the character, face and pose from the first image. Redraw her in the black satin leotard, sheer tights and ear headband from the second, keeping the same cel shading and line weight, on a deep crimson and gold casino background.
A two-reference costume swap that names the line weight and shading to preserve, delivered square as png. You supply both the character sheet and the costume reference.
Check the result before iterating
One to ten references, each needing a role assigned in the prompt; unnamed sources get blended. size still applies here, so a reference of a different shape is reframed to the preset you choose — set it deliberately when the source's proportions matter. Long edit chains drift, so go back to the original frame rather than re-editing an edit.
API quick start
# Generate the idempotency key once; reuse it when retrying after a lost response.
export SPICY_IDEMPOTENCY_KEY="${SPICY_IDEMPOTENCY_KEY:-$(uuidgen)}"
curl -X POST https://api.spicyapi.ai/api/v1/jobs/createTask \
-H "Authorization: Bearer $SPICY_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: $SPICY_IDEMPOTENCY_KEY" \
-d '{
"model": "bytedance/seedream-5.0-lite/edit",
"input": {
"image_urls": [
"https://example.com/reference.png"
],
"output_format": "jpeg",
"prompt": "A ceramic teapot on a linen cloth, soft window light, muted palette",
"size": "2048*2048"
}
}'
# Copy data.taskId from the response into TASK_ID, then poll the status.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.spicyapi.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
-H "Authorization: Bearer $SPICY_API_KEY"
See the API reference for authentication, the complete schema, polling, and error handling.
