ByteDance Editorial Image Editor
ByteDance's editorial-grade editor changes scenes, styling, and even the words on the poster — while the art direction of the original frame stays intact.
Your generated image will appear here.
Enter a prompt and click 'Generate' — or start from an example:
Examples
Drag the handle to compare each source and result, or open the viewer for a closer look.
scene rewrite — signature strength
Prompt: Rewrite this rainy night street as a snowy winter evening: falling snowflakes, a thin layer of snow on the pavement and cart awning, the same neon signs glowing through the snowfall, breath-visible cold air feel, wet reflections replaced by soft snow-muffled light. Keep the street layout, the food cart, the sign positions, and the composition exactly the same.
Source: Steven Pisano — CC-BY 2.0 via Flickr.
style transfer
Prompt: Repaint this studio portrait as a classical oil painting: visible canvas weave, confident brushstrokes in the background, rich warm skin tones, soft Rembrandt-style lighting from the left, a deep umber background. Keep his face, gray hair, stubble, navy sweater, pose, and expression clearly recognizable as the same man.
style transfer
Prompt: Restyle this flat illustration as a two-color risograph print: overprint texture, slight ink misregistration, grainy dithering, a terracotta and slate-blue ink palette on off-white paper. Keep the composition — person with laptop inside the giant mug, bookshelf and lamp, steam clouds, negative space on the right — exactly the same.
scene rewrite
Prompt: Rewrite this cozy bookshop café as a mysterious midnight scene: warm lamps dimmed to pools of amber light, moonlight through the tall windows, dust motes in the beams, the cat awake with glowing eyes, deep blue shadows between the shelves. Keep the isometric cutaway structure, both floors, the staircase, the furniture positions, and every room element readable in the same places.
creative edits
Prompt: Rewrite the underwater scene as a cosmic nebula version: the swimmer drifting through deep space toward a glowing golden nebula, the silver fish becoming tiny stars and constellations, deep teal fading to black at the edges. Keep the square composition, the swimmer's pose and direction, and the typography — "MIRA VALE" at the top and "saltwater hymns" at the bottom — exactly the same and legible.
creative edits
Prompt: Restyle this racehorse photo as a bold vintage screen-print racing poster: flat ink colors, visible halftone grain, a limited palette of racing green, cream, and burnt orange, the number-5 horse and fence rendered as graphic shapes with a slightly distressed texture. Keep the horse's jumping pose, the fence position, and the overall composition clearly recognizable.
Source: Paolo Camera — CC-BY 2.0 via Flickr.
Seedream V4 Edit applies ByteDance's Seedream V4 generation engine to existing images, and it carries that family's poster sensibility with it: the model understands layout, styling, and the role type plays in a frame. Ask it to swap a headline, restyle a set, or redress a model, and the result reads like a retouched original rather than a patched composite.
On RenderFlow AI, the canvas accepts up to four reference images, outputs at any size from 1024 to 4096 pixels per side (2048 default), and offers a Max Images control from 1 to 14 for pulling several variations from one instruction — all at 5 credits per image.
Start with the image that carries the art direction, then enable Inputs 2–4 for supporting material — alternate product angles, a palette reference, a logo lockup. The model matches what it can see, so concrete references beat longer descriptions.
State what changes, what must stay, and quote any copy verbatim: "change the headline to 'AUTUMN EDIT', keep the model, the set, and the grade." Layout-aware edits respond to position words like top-left, second line, or behind the subject.
Width and height run independently from 1024 to 4096 pixels, with 2048×2048 as the starting point. Drop toward 1024 while layouts are being approved, then render the final at the dimensions the hero slot or placement actually requires.
Raise the Max Images slider above 1 and say the count in the prompt — "create three crop variations of this layout" — to collect options from a single instruction instead of re-running the same brief by hand.
When the job is a simple object removal or background cleanup with no text or layout at stake, SeedEdit V3 does it for 4 credits. Save Seedream V4 Edit for frames where the editorial look and the type matter.
Width and height are set independently anywhere from 1024 to 4096 pixels, with a 2048×2048 default. That range covers everything from quick approval drafts to high-resolution hero placements without a separate upscaler.
Yes — text-aware editing is a family strength. Quote the exact replacement copy, describe the type style ("same condensed sans, all caps"), and keep strings short; the model resets the words in a matching treatment rather than pasting them on.
It sets how many edited results one run may return, from 1 to 14. State the number you want in the prompt as well — for example, "generate 4 layout variations" — so the instruction and the slider agree, then pick the strongest of the set.
The canvas accepts up to four inputs: the frame being retouched plus three optional references. Use them for alternate angles, palette or style targets, and brand assets the edit must incorporate.
SeedEdit V3 (4 credits) is the everyday repair tool — removals, background swaps, quick cleanups on single photos. Seedream V4 Edit (5 credits) is the editorial tool: text-aware changes, multi-reference art direction, and output sizes up to 4096 pixels for frames that ship to placements.
Long strings, unusual letterforms, and very small type are the hard cases for any image editor. Shorten the copy, name a common type style, and give the text more room in the frame; pinning the Seed once you are close lets you retry without losing the layout you liked.
Edit existing images with Google Nano Banana Edit.
Edit related image sets with ByteDance Seedream V4 Sequential Edit.
Edit images with context-aware FLUX prompt controls.
Create images with strong text rendering and layout control.
Generate images with Tencent Hunyuan Image 2.1.
Generate high-quality ByteDance Seedream V4 images for campaigns, posters, and social visuals.