1 · Subject / product (B) (optional — leave empty to generate from scratch)
Drop / paste / click to upload the product photo
the AI edits this exact product / places it into the background
the AI edits this exact product / places it into the background
1a · More subjects (optional — combine several products in one image)
Drag, click, or paste (Ctrl+V) product images here — up to 4 more subjects
Add more product photos to place several products together. The AI arranges every subject (the one above + these) into one image — into the Background below if set, else on a clean surface. Leave empty for a single hero product. Up to 5 subjects.
1b · Background (A) (optional — compose the subject into this scene)
Drop / click a background scene
the subject above is placed into this scene, in context
the subject above is placed into this scene, in context
When set, the AI composes your subject into this background at a believable scale / perspective / position (context-aware — it re-renders the scene, holding its look, not a paste). Needs a subject photo above.
For hero / main images where the label, logo and shape must be the real product. Cuts your product out and places it (with a shadow) on the background above — or a clean studio backdrop if none. The product pixels stay real; only the background is AI. Needs a Subject photo.
2 · Prompt
✨ Prompt assist — describe what you want, get a better prompt
2b · Text on the image (optional — the AI paints this styled text into the generated image)
The exact words are strongly nudged (the AI paints them — not pixel-guaranteed). For guaranteed labels stamped on a base image, use Text variants (flat) below.
3 · Aspect ratio
Custom
×
Any ratio beyond the native ones (16:9, 9:16, custom …) is rendered at the closest supported size, then centre-cropped to your exact dimensions.
4 · Asset type — tick one or more (one render each)
Each drives the composition of the render (a per-type template woven into the prompt): hero = clean studio; lifestyle = in-context scene; comparison = variants side-by-side; size chart = dimension callouts. Ticking N gives N images.
5 · Brand colours (optional)
Each set colour is passed to the model by role (background / logo-text / accent). Leave blank to let the model choose.
6 · Text variants (optional — leave blank for a single image)
Stamped deterministically — every output is pixel-identical except the label. Best for flat labels/graphics; free (no extra model calls).
7 · Generate
Runs on the workspace bot and shows the result below — you stay here. Engine set by the chip up top (⚡ subscription / 💳 metered) — metered is faster (per-image $), subscription is free but slower. Text quality is similar and varies run-to-run — for EXACT styled text use a deterministic overlay, not the AI.
Result
Your generated image appears here.