XO3D Studio · Standards
Product truth.
Most product CGI is judged on whether it looks real. That is the wrong test. The question that decides whether a launch visual works is narrower and harder: is the product on screen the product you are shipping. This is what that standard consists of, where it is held in production, and what it deliberately does not claim.
Product truth, carried across the whole launch.
The studio behind this guide
The standard this describes
Product truth is XO3D’s own production standard for checking a render against the approved source, material and colour decisions, not a product-data system: the geometry comes from your CAD where it exists, the materials are developed against product photographs, specifications and physical samples where supplied, the colour is reviewed against the value you specify, and where you supply a physical sample the render is compared against it before look development is signed off. Come to us when a buyer knows the product well enough to notice what is wrong with it, and being nearly right would cost you credibility. Two limits worth stating plainly: this is not an engineering sign-off and cannot certify a tolerance, and if what you need is a fast approximation rather than an accurate one, a volume batch renderer is the cheaper route.
How XO3D compares to the other five supplier modelsThe definition
What it
actually
means.
Product truth is XO3D’s standard for checking a render against the approved source, material and colour decisions for a product. It is a production standard, not a product-data or content-management system. In short, the product on screen is the product you are shipping. The geometry comes from your CAD where it exists, rather than an approximation of it. Materials are developed or adapted against the available product references: photographs, finish specifications and physical samples where supplied. The colour is the value you specified, not the one that happens to render well. The lighting is designed for this product rather than inherited from the last one. And where you can supply a physical sample, the render is compared against it before the look is signed off.
It is a production standard with named checkpoints, not a style or a claim about quality. Every part of it can be pointed at in the process and shown to have happened, which is the only kind of standard worth publishing.
It exists because the alternative is expensive in a way that is hard to see. A render that is nearly right does not fail loudly. It just quietly costs the product some credibility with every person who looks at it.
Five things that have to be true
The standard,
in parts.
Each one is a technical discipline with a commercial consequence. The discipline is the work. The consequence is the reason the work is worth paying for.
The geometry is yours
Where CAD and engineering drawings exist, the model is built from them rather than rebuilt by eye. Proportion, wall thickness, radius and the relationship between parts come from the approved source revision rather than from an artist deciding what looks about right. CAD can be out of date, incomplete or simplified, so we say which details are interpreted and agree them with you.
The details the camera sees are reviewed against the approved revision, so a buyer who knows the product is not shown the wrong one.
The materials are matched to the reference
CAD appearance data exists so engineers can read a model. It does not survive a camera. Each finish is developed from the available product references, including photographs, specifications and physical samples where supplied: how the anodising shifts at a grazing angle, which way a brushed grain runs, how a soft-touch coating scatters light instead of reflecting it. A library material can be a starting point where it fits; the test is the reference, not whether the shader was built from scratch.
The surface behaves the way the viewer expects that surface to behave, so the product reads as an object rather than as a picture of one.
The colour is the value you specified
Brand colour has a specified value, whether that is a Pantone reference or a CMYK build. That value is the reference for look development and the grade, rather than whatever grades most attractively. A screen, a video encode or a print process can still shift how a colour appears, so each destination output is checked rather than assumed to match.
The launch film, the campaign still, the retailer listing and the printed piece are recognisably the same product, which is what brand consistency actually means in practice.
The lighting is designed for this product
A rig built for this product, this sector and your brand language. Not a studio preset carried over from the last job. Where the sector benchmark is the finest product photography in the world, the lighting is measured against that rather than against other renders.
The frame earns its place instead of merely filling a slot on a launch page.
Where you can supply a sample, we check against it
A model can look correct on screen and still be wrong. Where you can give us a physical sample or a finish swatch, the render is compared against it directly rather than judged from memory, and that happens before look development is signed off.
The comparison is made against something real, and the record says what was compared.
Why it sells
Accuracy is
not the enemy
of desire.
There is a persistent assumption that technical accuracy and emotional appeal pull in opposite directions, and that making a product desirable means taking liberties with it. We do not think they do.
The people who look hardest at launch imagery often know the product, or products like it. When a surface behaves the way it should, attention stays on the product. When a highlight on aluminium reads like plastic, or a proportion is slightly off, a knowledgeable viewer may notice, even without saying what is wrong. The practical risk is showing a detail the buyer knows, wrongly.
Precision is what buys the room to be beautiful, because a frame nobody is quietly doubting is a frame that can carry an argument.
The longer version of this argument is in material obsession and our philosophy.
Be precise about this
What it does
not claim.
- It is not an engineering sign-off
- A model built from your CAD and confirmed by your engineers is a strong basis for a technically credible image. It is not a validation of the manufactured product and it cannot certify a tolerance.
- It is not a regulatory approval
- On medical, industrial and safety-related products, regulatory acceptance stays with you. We represent what your approved source material supports, and no more than that.
- It depends on what you can supply
- The sample check happens where a sample exists. The colour is held to a specified value where a value is specified. Where the source is thinner, we say what the work can and cannot claim before it starts.
- It is not the same as looking real
- A render can be convincingly photographic and still misrepresent the product. Photorealism is a technique. Product truth is a standard, and the two are not interchangeable.
Where a specific accuracy or representational requirement exists, define it at brief stage and name who approves it. That is a short conversation before production and a serious problem afterwards. For the glass, liquid and label checks on a fragrance launch, see our perfume marketing guide.
The five checkpoints
What each part
needs and proves.
Each part of the standard depends on evidence you supply. Where that evidence is missing, the record says so rather than marking the check as passed.
| Part | Evidence needed | Studio check | Client approval | Remaining limitation |
|---|---|---|---|---|
| Geometry | CAD or drawings, with the approved revision named | Source audit: what the model can be built from and what is missing | Your engineering or product reviewer confirms the model | Not an engineering sign-off; interpreted details are listed |
| Material and finish | Photographs, specifications, and samples where supplied | Look development against the references | Look development sign-off | Without a sample, the finish is an informed interpretation |
| Colour | A specified value or approved reference | Render and grade reviewed against the value | Colour approved in look development | Screens, encodes and print can change how a colour appears |
| Lighting | Brand direction and the intended shots | A rig designed for this product, reviewed in context | Look and shot approval | A creative decision, not a measurement |
| Sample comparison | A physical sample or finish swatch | Render compared with the sample before sign-off | Agreed with you before sign-off, noting who compared it and under what light | Covers that sample only, not production batches |
Evidence
What your
evidence
supports.
Different evidence supports different claims. None of it is needed on every brief, and none of it certifies the whole product.
- CAD and drawingsSupport the geometry: proportions, radii and how the parts fit. They do not show how a finish looks.
- Reference images and specificationsSupport an informed interpretation of materials, colour and finish detail.
- Physical samplesSupport a direct visual comparison of the finish under agreed viewing conditions.
- Measurements, where they existSupport a stated value, such as a colour reading, for the item that was measured.
Provisional details are marked as provisional. A new CAD revision, a changed finish or a new sample is a reason to re-check the affected images.
A real material decision
Interpretation,
named as such.
On the Nanuskate modular ramp, Jack Robertson, the lead 3D artist, built the plastic composite finish from images supplied by the client and photographs of comparable finishes. No physical sample was provided. The first pass looked too uniform, so he added a slight colour shift on each panel. The references also showed fine speckling and small pellet-like detail.
What the record supports: the finish was built from those references, and the panel variation was a deliberate change after the first pass. What stays interpretation: how closely it matches the physical panels, because there was no sample to compare it with. If a sample arrived later, the right step would be to compare it with the approved look rather than assume the earlier interpretation is exact.
Approval record
Write down
what was
approved.
A short record makes the standard reviewable. It names the project and product revision, the source file or reference the work was built from, the basis for each material and colour, the simplifications you accepted, who approved it and when, the outputs it covers and anything still unresolved. The blank sheet below has those fields. It is a template: an unsigned sheet is not an approval.
In production
Where each
one is held.
None of this is a promise made at the brief and hoped for at the end. Each part is held at a named point, and the work stops until the relevant approval exists.
The source audit at stage one establishes what the geometry can be built from and what is missing. Look development at stage two is where the model, the materials, the colour value and the lighting rig are built and signed off together, with a sample comparison when a sample is available. The finishing grade at stage four has to hold the approved colour rather than trade it for a look. Your engineering reviewer confirms the model before anything is built on top of it.
The full sequence, including who signs off what, is on how we work. To plan the reviews, see how to review CGI and the product CGI timeline. The technical detail of getting from engineering data to a production model is in CAD to CGI. Our 3D modelling services cover the build itself, and the exploded view guide sets out what an exploded view proves and does not prove.
Related
Where to
go next.
Common questions
Straight
answers.
What does product truth mean in product CGI?
Product truth means the product on screen is the product you are shipping. The geometry comes from your CAD rather than an approximation of it, the materials are developed or adapted against the available references for that product, the colour is held to the value you specify, the lighting is designed for that specific product, and where you can supply a physical sample the render is compared against it before the look is signed off. It is a production standard with named checkpoints, not a style.
Is product truth the same as photorealism?
No, and conflating them is a common and expensive mistake. Photorealism is a technique for making an image look photographic. Product truth is a standard for making the image correct about the product. A render can be highly photorealistic and still show the wrong revision, the wrong finish behaviour or a colour that has drifted. Realism without accuracy produces a convincing picture of something you are not selling.
How is colour accuracy held across different outputs?
Brand colour is treated as a specified value rather than a description, whether that is a Pantone reference or a CMYK build. That value is set during look development and kept through the finishing grade, so a colour is not traded away to make a shot look better. A reference value, a rendered pixel and a printed or encoded result are not the same thing, and lighting, screens and print processes change how a colour appears. So each output is checked against the reference, and any limit is stated rather than hidden.
How do you keep product finishes from looking too uniform?
Study the reference available for each finish and check the render under the intended lighting. For the Nanuskate ramp, Jack Robertson, the lead 3D artist on that project, had client-supplied images and photographs of comparable plastic composites, but no physical sample. He built the material from scratch because no library finish was close enough. The first pass looked too uniform, so he added a slight colour shift to each panel.
Do you need a physical sample to work accurately?
No. Much of our work happens before a physical sample exists, built from CAD, engineering drawings and written material and finish specifications. Where a sample or a finish swatch is available, it makes the material reconstruction faster and the accuracy verifiable, so it is worth sending. Where one is not, accuracy rests on the source material and we are explicit about what that supports.
Why does accuracy matter commercially and not just technically?
Because the people who look most closely at launch imagery often know the product: buyers, retailers, engineers and existing customers. A surface that behaves the wrong way or a proportion that is slightly off can be noticed even when nobody says what is wrong. Disciplined review removes that risk for the details you know matter. Accuracy is not the opposite of desire; it gives the creative work room.
Is product truth the same as a single source of truth for product data?
No. A product data system holds product information, such as specifications and descriptions, in one place. Product truth is how a render is checked and approved against that kind of information: the approved source, materials and colour.
Can CGI still be wrong?
Yes. A render can only be as right as its source and its review. If the CAD is out of date or a reference is misread, the image will be wrong in the same way. That is why the revision is named and the camera-visible details are reviewed before final output.
What does a sample comparison establish?
That the render was compared with that sample, under the conditions it was viewed in, by the person who reviewed it. It does not validate the manufactured product or later production batches, and it does not cover parts of the product the sample does not show.
What happens when the CAD changes during a project?
Send the new revision and say what changed. The studio checks which views or shots it affects and confirms what needs updating. Approvals are recorded against a revision, so it is clear which images match which version of the product.
How is AI-assisted work reviewed?
Against the same evidence as any other image: the approved source, the material and colour references, and the people who approve the product. See CGI versus AI images for how the two differ in evidence and control.
Past production insight from
Jack Robertson
Former Mid-Level Artist, XO3D
Match the evidence you have, and name its limits
Jack built a composite material from scratch when there was no close library starting point. The reference was photographic, including images supplied for the product and comparable finishes. A uniform first pass needed slight panel-to-panel colour variation to read more convincingly.
The important distinction is between an informed material interpretation and a verified match to a physical sample. Record which reference supports the finish, scale of detail and variation. If a sample becomes available later, compare it with the approved look rather than assuming the earlier interpretation is exact.
A material library can be a starting point where it fits. The acceptance test is the product reference and the intended shot, not whether the shader was built from scratch. See how look development establishes that test.
From a production discussion while Jack worked at XO3D, edited for clarity. XO3D's named expert on this subject is Thomas Howcroft, Founder and Creative Director.
Start with the product
Tell us what
has to be right.
Send the product, the source files you have and the finishes that matter. We will tell you what the work can claim accurately and what it cannot.
Start a conversationYour next project
Make the approval reference explicit.
Agree which drawings, samples and product revision define the image. Record deliberate explanatory simplifications so a compelling frame does not quietly become an unsupported product claim.
A route to consider
Technical CGI
An explanation of a product or mechanism built from approved source information, with technical review before finishing.
Bring these three things.
- Which revision is authoritative?
- Which features need specialist review?
- What visual simplifications are acceptable?
Start with an outline. We review the product, intended outputs and timing before agreeing the production scope.