Verified against Flux · 2026-08-02
Drop an existing product photo into a new lifestyle scene with Flux.1 Kontext
A Flux.1 Kontext edit prompt that takes a product photo you already have and places it convincingly into a brand-new lifestyle environment, preserving the exact product geometry and label instead of letting the model regenerate — and often subtly alter — the product itself.
The prompt
Ready to copy — highlighted parts are example details you can swap.
SOURCE PRODUCT IMAGE bottle_hero_v3.png — a 500ml amber glass serum bottle with a white pump cap, shot on plain white. This exact product — its shape, proportions, label text, and color — is the fixed input; it is being relocated into a new scene, not regenerated from a description. WHAT MUST NOT CHANGE ON THE PRODUCT the front label reading "LUMINEUX — Vitamin C Serum, 500ml" in a thin serif font, and the brushed-gold pump cap. Do not redraw, re-letter, or "clean up" the label — every character of any visible text must match the source exactly, since a model asked to place a product into a new scene will sometimes quietly re-render label text as slightly different or blurred instead of preserving it pixel-for-pixel. NEW ENVIRONMENT a sunlit bathroom shelf beside a folded white towel and a small potted eucalyptus sprig, morning light through a frosted window HOW THE PRODUCT SHOULD SIT IN THE SCENE standing upright, slightly angled toward camera, resting on the marble shelf edge with nothing touching it. The product's scale relative to the surrounding objects in the new scene must be physically plausible — roughly the same height as the folded towel stack beside it — not oversized or shrunk to fit the composition. LIGHT MATCHING Relight the product to match the new environment's actual light source and direction, including a correctly angled contact shadow and any color cast the ambient light would realistically throw onto the product's surface. A product lit like a studio shot dropped into an outdoor or ambient-lit scene will look pasted in, so the light has to change even though the product itself does not. WHAT SHOULD NOT APPEAR Because there is no negative-prompt field, state this positively: the new scene contains only the environment described above and the single product from the source image — no duplicate of the product, no second brand's product nearby, no added text or graphic overlay beyond what was already on the product's label. REFLECTIONS AND SURROUNDING SURFACES If the new environment includes a reflective or glossy surface near the product — a marble counter, a mirror, a wet surface — the product should actually appear in that reflection, faintly and correctly positioned, the way it would if it were physically sitting there; a glossy surface with no reflection of the object resting on it is one of the more common tells that a product was composited in rather than photographed in place. CHECK BEFORE FINALIZING Compare the product's edges against the new background at the pixel level in your own review: a hard, slightly too-clean cutout edge, or a thin halo of the old white background bleeding through around the product's silhouette, means the edit needs another pass rather than being accepted as-is. Also confirm the label text is still exactly legible and unchanged from bottle_hero_v3.png — a 500ml amber glass serum bottle with a white pump cap, shot on plain white — a label that has become slightly blurred or re-angled during compositing defeats the entire purpose of anchoring to a real source image. OUTPUT One composited image where the product's geometry and label are unchanged from bottle_hero_v3.png — a 500ml amber glass serum bottle with a white pump cap, shot on plain white, now lit and placed as if it were physically photographed inside the new scene, with the seam between "product" and "environment" invisible.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Kontext's core capability — editing a real input image rather than generating from text — is precisely what pixel-accurate label preservation needs. Asking base Flux.2 to "generate a serum bottle with this label text" invites the model to hallucinate slightly different label wording on every single generation, since text rendering inside a fresh generation is never guaranteed letter-perfect; anchoring to an actual source image and editing around it is the only reliable way to keep real label text intact across a batch of lifestyle scenes. The explicit relight instruction exists because Kontext preserves an input image's content faithfully by design, and that includes its original lighting unless told otherwise. A studio-lit product simply composited into a sunlit bathroom scene without a relighting instruction will look exactly like what it is — a cutout — because its shadow direction and color temperature won't match its new surroundings. Naming the specific new light source is what makes the edit read as one coherent photograph rather than a collage of two mismatched images. The scale_reference constraint matters because compositing tasks are where image models most commonly get relative proportions wrong: nothing else in the prompt anchors how large the product should read next to the towel or the plant, and without a stated comparison, Flux will pick a size that looks compositionally pleasing in isolation rather than one that is physically correct relative to the objects actually surrounding it. Positively stating "no duplicate of the product, no second brand's product nearby" matters more here than in a pure text-to-image brief, because compositing prompts increase the odds of an unwanted second instance appearing — the model is reasoning about "a product in a scene" as a general category it has seen many times, not about one specific, singular object it needs to keep to exactly one count.
What you get back
The same amber serum bottle from bottle_hero_v3.png, label text pixel-identical, now standing on a sunlit marble bathroom shelf beside a folded towel and a eucalyptus sprig, with a soft warm-toned contact shadow and a faint reflection matching the shelf's real light — no visible compositing seam.
Verified against
Flux Flux.1 Kontext [pro] · 2026-08-02
Changelog
- 2026-08-02 — Initial publish, verified against Flux.1 Kontext [pro] for pixel-accurate label preservation during scene compositing.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
