When to Photograph a Product and When to Render It: A Straight Answer

A question that was easy five years ago has become genuinely difficult. Photograph the product or build it in 3D? Rendering has reached the point where, for hard surface products under controlled lighting, most viewers cannot reliably tell the difference and once quality stops being the deciding factor, the decision becomes about economics, timing and what the product is made of.

Both get chosen for the wrong reasons routinely. CGI gets picked because it sounds advanced, on products where it will look wrong. Photography gets picked out of habit, on projects where rendering would have been cheaper and more flexible.

What material tells you

The single most reliable predictor of whether CGI will hold up is what the product is made of.

Rendering is strongest on manufactured, geometrically regular, hard surfaced things: electronics, packaging, bottles, cosmetics, appliances, watches, automotive parts, furniture. These have defined geometry and predictable material behaviour, which is precisely what a renderer models well.

Rendering struggles with irregularity and softness. Fabric drape, embroidery, food, fresh produce, human skin, hair, anything handmade where the appeal is that no two are identical. These can be rendered, at a cost that usually exceeds simply photographing them and with a persistent risk of the uncanny result where everything is technically correct and nothing looks real.

Textiles are the clearest case in this market. When we shot Shemmagh Al Rayyan's product range, the entire proposition was material credibility, the weave, the fabric weight, the stitching. Those qualities are perceived through microscopic irregularity and the specific way light passes through and across real cloth. A render would have needed to fabricate exactly the imperfections that make the product convincing, at greater expense than photographing the object that already had them.

Food is not a close call

Photograph food. Rendered food reads as artificial almost immediately, because appetite appeal comes from irregularity, uneven browning, condensation, imperfect melt, steam, the particular gloss of a real sauce. Rendering these convincingly is among the hardest problems in CGI and there's no reason to attempt it when the subject is sitting in a kitchen. The gap between good and bad food imagery is a lighting and styling problem, not a technology one.

Where CGI wins decisively

Four situations where rendering isn't just viable but clearly correct:

• The product doesn't exist yet. Preproduction marketing, crowdfunding, trade presentations, investor material. You cannot photograph a prototype that hasn't been manufactured.
• Many variants of one form. Twelve colourways, eight sizes, four finishes. Photography multiplies: every variant is another setup. CGI front loads the cost into building the model once, after which additional variants are close to free. This is where the economics flip hardest, and where photography quotes start looking expensive for reasons clients don't expect.
• Impossible views. Cutaways, exploded assemblies, internal mechanisms, cross sections. A camera cannot see inside a sealed product.
• Perfect repeatability. When an asset must be rerendered next year at a different angle with a changed label, and match exactly. Recreating a photograph's precise lighting months later is difficult; rerendering a scene is trivial.

Where photography wins decisively

• The product's appeal is tactile or irregular. Covered above: textiles, food, craft, anything organic.
• You need it soon. A studio shoot can deliver in days. Modelling, texturing, lighting and rendering a product properly takes longer, and the first render is where most of the time goes.
• One product, few images. If you need six shots of one item, photography is almost always cheaper. CGI's advantage depends on amortising the model across many outputs.
• Authenticity is the claim. Handmade, locally produced, heritage, categories where the credibility of the object is the point. Rendered imagery undercuts a handmade claim even when nobody can articulate why.
• Human interaction. Product in use, in hands, in a real environment. Rendered people remain the weakest element in most CGI.

The combination most brands should be using

Framing this as a binary is the actual mistake. The strongest product libraries we build use both, allocated by task.

Photograph the hero imagery, the product in real light, in context, with human presence, where authenticity and warmth matter. Render the systematic material: variant grids, ecommerce views on clean backgrounds, technical diagrams, packaging mockups, anything needing repetition or perfect consistency. Then match them deliberately so the photographed and rendered assets share a lighting logic and colour treatment and read as one library rather than two.

That last step is what's usually skipped, and it's the one that makes the approach work. A feed where rendered and photographed images obviously come from different sources looks inconsistent, which defeats the point of having a large library at all.

The decision in one question

Ask how many distinct images you'll need from this product over the next two years, and whether its appeal depends on texture.

High image count, hard surface, build it in 3D and stop paying for repeat shoots. Low image count or an appeal that lives in material and irregularity, photograph it. Both, in most product ranges of any size, which is why the useful conversation is about allocation rather than choosing a side.