Relighting Product Shots: Three Lights, One Instruction
Relighting holds the product still and moves the light — a distinct operation from a regenerate, and the cheapest way to turn one shot into a campaign.
One hero shot, one campaign brief that calls for it in morning light, studio white, and warm dusk — the reflex is to reshoot or regenerate three times and hope the product looks like the same product across all three. It doesn't need to be three generations. Relighting is a narrower, more specific operation than "make a new version of this image," and treating it as its own tool rather than a re-prompt is what keeps the product identical while everything about how it's lit actually changes.
Relighting is not the same operation as regenerating
A full regeneration re-derives the entire frame from the prompt — geometry, material, background, product details, all of it up for reinterpretation, with lighting as just one property among many that might shift. Relighting is deliberately narrower: hold the subject, its shape and its material exactly where they are, and change only the light source's position, color and intensity. That distinction matters because it's also a risk boundary. A regenerate that's supposed to only change the lighting can just as easily change the label text, the product's proportions or the packaging color along with it — three things a client will notice immediately and a lighting change alone never should have touched.
What a purpose-built relighting tool actually exposes
Magnific's Relight tool is a clear illustration of what a dedicated relighting mechanism looks like when it's built as its own interface rather than bolted onto a general edit prompt. The interaction model is a hard cap of three lights, with position, angle and color set independently for each one — a UI built around the physical unit a lighting setup is actually made of, rather than a single "change the lighting" text box. The underlying model has to understand the photo's own geometry to make that work at all, since a light placed by a user still has to cast a shadow that agrees with the product's actual shape rather than one guessed from a caption. Versely doesn't carry Magnific — it isn't in our own model catalog — so treat that as a real, useful reference point for what the category looks like at its most explicit, not as a workflow available inside Versely today.
What Versely does carry in this space works a level more abstract: Kling Image O1 handles whole-frame lighting changes from a text instruction rather than a per-light control panel — shifting a scene from daylight toward dusk is one of the standard examples used to demonstrate it — at three credits a call, which makes it one of the cheapest single-instruction options in the catalog for a broad lighting shift across an entire frame.
The instruction-driven path for a precise, single-element change
Between "three lights I place by hand" and "one instruction that rewrites the whole frame's lighting" sits the more common product-photography need: change the light, touch nothing else, and do it without cropping the request down to a per-light UI. Seedream 5.0 Pro Edit is built almost exactly for that brief — its own listing describes grounded, region-precise editing that changes one element while keeping the rest of the frame intact, at seven credits a call, which is the right description for a relight specifically because holding everything except the light source constant is the entire point of the operation.
It isn't the only option at that price point. Flux 2 Max Edit sits at the same seven credits for ultra-detailed photorealistic output, Qwen Image 2 Edit Pro runs eight credits with an explicit design goal of preserving the original image's integrity while applying an instructed change, and Nano Banana Pro Edit costs more at fifteen credits but is worth the difference on a hero shot where the relit result is going to anchor an entire campaign rather than sit in a batch. None of the four is a universal right answer — Versely's best image editing model comparison is worth a look before committing a hero shot to any one of them, since the model that holds a label's typography crispest under a relight isn't necessarily the one that handles a reflective surface best, and a product shoot usually needs both at different points in the same campaign. Versely's full model catalog has the complete spread if the brief calls for something outside those four entirely.
The economics of one shot covering a campaign
The interesting part isn't any single relight — it's what a handful of them add up to. One well-lit hero product photo, run through three or four relighting passes at seven to fifteen credits each, produces a morning, studio and dusk version of the same shot for a fraction of what a full reshoot or three independent regenerations would cost, with the product itself never actually changing between versions. That's a genuinely different economic shape than the flat-per-call math covered on Versely's cost breakdown for a hundred product images — this isn't about generating more images, it's about extracting more usable variants from one image that's already right.
It also changes what's worth spending on upstream. A single hero shot that's going to be relit four or five different ways justifies a better original than a shot that's only ever going to exist once — clean geometry, an unambiguous single light source in the source photo, and a background simple enough that a relight pass isn't fighting existing shadows it has to overwrite. A messy source shot compounds across every relit variant instead of costing you once.
The shadow-direction tell
There's a fast, reliable way to catch a relight that didn't actually hold the geometry: check where every shadow points. A single convincing light source produces shadows that all fall away from it consistently — a product's own shadow, any shadow cast by its label or texture, and any secondary shadow from a prop in frame should all agree on where the light is, because they're all responding to the same source. A relight that's drifted usually shows the tell before anything else does: a shadow angled one way under the product and a highlight suggesting the light is somewhere else entirely, or a shadow with no matching highlight anywhere in frame.
The second tell lives on anything reflective. A glossy product surface, a metal cap, a glass bottle — all of them should show a coherent catch-light in roughly the position the new light source implies. A relit bottle with warm dusk tones across its label but a cold, unchanged highlight in the glass is the same failure as the shadow mismatch: two parts of the image disagreeing about where the light actually is. Both checks take seconds and catch the failure mode that a quick glance at "does it look warmer now" reliably misses.
A Versely walkthrough
For a single product shot that needs to hold its identity across a relight:
"Relight this product shot to warm golden-hour light coming from camera-left. Keep the product, background and packaging text exactly as they are — only the lighting should change."
That's a generate_image_from_image call with the source photo attached and Seedream 5.0 Pro Edit as the model — the region-precise, single-element-change behavior described above is exactly the constraint this instruction is asking for. Run the same instruction again with "cool studio white light from directly above" or "dusk light from camera-right" swapped in, and a single seven-credit hero shot becomes a small lit set rather than a single frame. Before anything ships, run the shadow check above on each variant — a preview render is cheap enough to catch a mismatched shadow before it goes out under a client's name. Versely's AI photo editor is the general entry point if the brief turns out to need more than a relight once you're in there.