A mini vector editor — artboards, shapes, text, and your images. Export to PNG, SVG or PDF.
Loading…
Your projects
Each project is a separate canvas. Open one to start generating.
Loading projects…
Your characters
Define a character with reference sheets and a description. Drop the same character into any project — no retraining needed.
Loading characters…
Your products
Build a product with reference images (and a description). Mark your best with ★ — the AI uses only those. Drop the same product into any project to keep it consistent.
Loading products…
Your brand profiles
Capture a brand's colors, fonts, tone and style — by hand or from a website. Drop the same brand into any project so every generation stays on-brand.
Loading brand profiles…
Shared with you
Canvases other people invited you to. Your access (view / edit) is shown on each.
Loading…
Run a preset
Preset
Shots
0 selected
Brand profile
Website URL
GPT-4o vision reads the homepage, extracts colors / fonts / tone / style, and drops a brand card onto your canvas. Auto-anchors it so every future generation in this project picks up the brand look. ~15-30s, ~$0.03-0.05 per analysis.
Name (for @mentions)
Tagline (optional)
Color palette
Fonts
Tone / brand voice
Visual style / description
Guidance prompt (optional — auto-built from the fields above if left blank)
🔍
Fetching the site…
New object
Name
① Add reference images
Paste a product URL to auto-import, or upload your own. Mark your best with ★ ref — those become the AI reference.
AI scrapes the product page and pulls every product image it can find. Fills name + description if empty.
② Description
Auto-describe writes it from your images, or type your own.
③ Generate more views optional
Turn your marked reference into extra studio angles. Pick a product type, tick the shots, then Generate.
Recommended shots
✓ Neutral background — white or light grey
✓ Soft, even light — no harsh shadows or glare
✓ Sharp focus, product fills the frame, high resolution
Mark your best uploads with ★ ref (front + label work well together). No mark = first image. GPT Image 2 · high quality · ~167 cr each. Context shots (in-room, on-model, scale) must be uploaded.
New character
Name
Description (optional)
Reference sheets
Clone from video — optional shortcut
Drop a short video of the person — we pull a face frame for the likeness AND a voice clip to clone, in one step. Best: 10–15s, clear front-facing face, clean single-speaker audio.
Real person?
To animate a real person on Seedance 2.0, verify them once in the
BytePlus console
— they do a ~1-min liveness check & consent — then paste the resulting
asset://… ID here. Leave blank for AI characters (those use Seedance 1.5 Pro / Kling).
Voice
Give this character a voice, then ▶ Preview before saving.
The clone copies your clip exactly — quality in, quality out. For the best result:
10–15 seconds, one speaker only — longer isn't better.
Clean audio — no music, background noise, or echo.
Clear, natural speech at a steady pace; no long pauses.
Match the mood you want — the clone copies the clip's energy and accent.
Good mic, even volume, no clipping. WAV or MP3.
Loading…
Generative Fill
Reference images (optional — guide the edit, e.g. "match this style/object")
What should change in the painted area?
Paint over the area you want to change (drag to brush; use the eraser to fix mistakes). Only the painted area is regenerated — everything else stays. Add reference images to steer the look. Result lands as a new linked node. Powered by GPT Image 2.
Version history
Automatic snapshots of this canvas. Restoring replaces the current canvas — and first backs up the current state, so a restore is itself reversible.