Start with the verdict. What goes wrong in an appearance design process is rarely the drawing — it is that nobody agreed where one stage ends. The client believes week three buys a 3D model ready to cut steel; we are still arguing about proportions in a sketch. After ten years and more than 320 products, I have compressed the process into six delivery gates: requirement definition → concept sketches (2–3 form directions) → 2D renderings and CMF → 3D surface appearance model → appearance prototype validation → appearance freeze and tooling drawing.
The first three gates decide what the product looks like. The last three decide whether it can be built and still look like that.
The dividing line is not how attractive the rendering is. It is whether each gate produces something the next stage can pick up directly. Every gate must end with a document or physical object someone can sign. Verbal approval feels comfortable in the meeting and comes back at you at gate four.
Gates 1 and 2: fix the visual weight before you draw a line
The requirement gate produces one page — the appearance positioning sheet. It answers four questions: which products will this sit next to on shelf, how large are the hands of the target user, is the main color cool or warm, and should the volume feel more restrained or more expansive than the competition. That page records visual judgment only — no structure, no function.
A recent desktop appliance brief opened with "simple and premium". Those words cannot be drawn. Three rounds of questioning got to what the client meant: it should sit on a wooden desk without demanding attention. That translates into low saturation, small volume and a matte finish — not "premium like Apple". Get that wrong and every sketch after it is wasted paper.
At the concept gate we present 2–3 form directions, 3–5 sketches each, and they must go onto a single comparison sheet, with the main character line, the parting tendency and the visual center of gravity marked on each. The pass criterion is blunt: can the client say where the three directions differ within 30 seconds? If not, the sketches are unclear — it is not that the client does not understand design.
The most common failure is drawing details too early. Button fillets, decorative grooves, logo placement — all drawn, while the overall proportion is still undecided. Then the client picks a direction, we move to 3D, the length-to-width ratio is wrong, and the entire design language starts over. Sketches settle exactly three things: volume proportion, the path of the main character line, and how one surface turns into the next.
Gate 3: color only counts under real light
At the 2D gate we deliver multi-view renderings plus a CMF list detailed enough to purchase from: not "white matte plastic" but "PC+ABS, matte texture to VDI 3400, color matched to the chip supplied by the client". We always render three lighting setups — soft studio, hard raking side light, and a real in-use environment. Soft light alone misleads: plenty of surfaces only expose a broken highlight under hard light.
How CMF actually gets confirmed: the color on a rendering is not confirmation. The supplier makes a physical chip to the color standard, and you look at it once in daylight and once under warm indoor light. The same color code reads noticeably lighter or darker on matte versus gloss, and a screen cannot show you that. Our rule: without a physical chip in hand, CMF is not confirmed.
One burn we remember: a handheld device with large areas of light grey matte looked clean and crisp in the render. The prototype made fingerprints and skin oil painfully obvious, because the texture was too fine — a little hand sweat and you get a dark patch. We fixed it by going one step coarser and changing the most-touched areas to a fine matte-to-gloss transition. That class of problem never shows up in a render.
| Surface treatment | Visual character | Fingerprint resistance | Where to use it |
|---|---|---|---|
| Fine matte texture | Clean, restrained | Poor | Decorative faces, rarely touched |
| Coarse matte texture | Solid, tactile | Good | Handheld, frequently touched faces |
| Gloss | Sharp, premium | Very poor | Small accents and edge strips |
| Two-shot / two-material transition | Clear layering | Moderate | Needs tooling coordination; assess cost |
Gates 4 and 5: the surface and the hand feel set here
The 3D appearance model is the slowest part of the chain. Our criterion: all surfaces G2 continuous, and every surface-to-surface intersection must take a fillet whose radius transitions naturally — no place where the highlight breaks. The check is crude but effective: switch the model to zebra stripes or highlight reflection and rotate it slowly through a full turn, watching whether the highlight line kinks — a kink means the surfaces were never joined properly.
Handheld products must fix grip parameters here. Palm width typically falls in the 75–90mm range, and the diameter of the grip section decides whether the thumb web opens wide enough. Once the diameter passes the limit of a natural power grip, users switch from gripping to clawing, and fatigue rises immediately. Button position is settled at this gate too: the button center goes where the finger naturally lands, not where the structure team finds it convenient.
The appearance prototype is the acceptance gate. This build has no structural parts — it validates appearance only: proportion, surface smoothness, how the CMF reads, and how it feels in the hand. When a prototype comes back we always do four things: photograph it side by side with the rendering, touch it wearing gloves, set it on a desk and look at the side profile, and hand it to someone outside the project to catch their first reaction. On one product everyone agreed it felt "off somehow" and could not say why; measuring showed the bottom fillets were larger than the top ones, so the object read top-heavy. Invisible in the render, obvious in the hand.
Gate 6: an appearance freeze is a handover package, not a sentence
Freezing the appearance is not something you announce. It is a package containing: the frozen 3D appearance model; a CMF confirmation report with physical chip number and texture grade; prototype photographs and the adjustment log; zebra-stripe continuity screenshots; and an annotated drawing of grip and button positions. Miss one item and the structure team fills the gap with their own interpretation — what comes out is no longer the form you agreed on.
The freeze criterion: the structural engineer can complete the internal stack-up without modifying any exterior surface. If exterior surfaces have to move, the freeze was called too early or the stack-up assessment came too late. Our practice is to pull structure in at gate four, so problem zones surface before the freeze, not after.
One step that gets skipped easily: the appearance sealing sample at tooling handover. The mold shop must be told which faces are cosmetic faces, what defect level is acceptable on them, and where ejector pin marks and gate marks are absolutely not allowed. Leave that unsaid and the shop applies its own standard — gates on visible surfaces, then days of polishing and rework.
The six gates at a glance
| Stage | Deliverable | Pass criterion |
|---|---|---|
| Requirement definition | One-page appearance positioning sheet | Can state who it sits beside, who uses it, where visual weight goes |
| Concept sketches | Comparison sheet of 2–3 directions | Differences explainable in 30 seconds |
| Renderings and CMF | Multi-view renders + CMF list + physical chips | Color confirmed on a physical chip, not on screen |
| 3D appearance model | G2-continuous surface model | No kink in zebra check; grip and button positions annotated |
| Appearance prototype | Validation prototype + comparison record | No visible gap versus the render; hand feel approved |
| Appearance freeze | Handover package + sealing sample standard | Internal stack-up completed without touching exterior surfaces |
FAQ: the questions clients actually ask
How long does the appearance design process take?
For the consumer products we usually handle, six to ten weeks from requirement definition to appearance freeze. Broken down: sketches 1–2 weeks, renderings and CMF 2–3 weeks, the 3D surface model slowest at typically 2–3 weeks, prototype and validation another 1–2 weeks. Every week the client's internal decision drags pushes the schedule back a week, so each gate gets a confirmation deadline at kickoff.
The rendering was approved, so why does the prototype look different?
A rendering is ideal light; a prototype is shaped by material, texture, process and real light. Color depth, reflection and the path of a highlight across a curve will almost certainly differ between screen and object — which is why prototype validation exists. Shrink the gap two ways: confirm CMF on physical chips rather than screens, and run the zebra continuity check before delivering the model. If the model is sound and the chip is on target, the remaining gap is acceptable.
Can the form still be changed after the appearance freeze?
Yes, but cost escalates fast with each stage. Changing the form after freeze means redoing structure and redrawing the tooling; if the mold already exists, changing one exterior surface can cost on the order of tens of thousands of RMB. So push form decisions into the prototype stage. We advise clients to move the internal review up to gate four, with structure, marketing and management looking at the 3D model together.
Can structure use the appearance 3D model directly?
The appearance model defines exterior surfaces and overall form. Internal structure, wall thickness, snap fits and assembly relationships are not in it. Structure builds the internal stack-up under those surfaces, treating them as a fixed reference. At handover we annotate which faces are locked and which may be micro-adjusted. Some surfaces need draft for ejection — permitted micro-adjustment, but small enough to be visually invisible.
Can we skip the prototype and go straight to tooling?
Technically yes, but not recommended. The prototype is the only chance to hold the object and see the real surface and color before cutting steel. Skipping it loads all form risk onto the mold, and once a mold needs rework the time and cost are in a different league. If the schedule is tight, at minimum validate proportion and hand feel on a prototype; CMF can wait for trial-shot parts. Proportion and grip problems cannot be fixed by modifying a mold.
If you remember one thing: every appearance gate needs something that can be signed. After three rounds of "let's go with that", someone still says "this feels a bit off" the week before tooling. Put a physical object on the table and the judgment converges.
Written by Hu Yashe (Asher), Hertz Industrial Design — 10 years, 320+ products in production.
Originally published at https://www.hezidesign.com/jingyan/1230.html
