No-CAD Gridfinity: 5 Real Trays Built From a Single Phone Photo

What "Gridfinity tray from photo without CAD" actually means

A Gridfinity tray from photo without CAD isn't a hack — it's the saner workflow. You skip Fusion 360, you skip the calipers, and you let computer vision handle the measuring. Below are five real projects built this way, with the gotchas and the tweaks that made each one print-right on the first try.

Project 1: 17-piece socket set, 1u-deep tray

One of the most common no-CAD Gridfinity wins is a socket organizer. The user laid 17 metric sockets out in a 4 by 5 grid on a sheet of printer paper, dropped a credit card in the corner, and shot top-down from about 50 cm. The AI generated a 4u by 4u tray with 17 individually sized cavities — every socket got a 1.5 mm clearance band and a finger scallop on the lip.

The fix that mattered: rotating the largest socket (24 mm) so its long axis ran along the 4u direction. The AI had it short-axis-aligned by default, which would have eaten an extra cell. A 5-second drag in the preview saved an hour of reslicing.

Project 2: Kitchen junk drawer

This one had everything: scissors, a measuring spoon set, two pens, twine, a tape measure, and the inexplicable AAA battery you always find in a kitchen drawer. The reference was a printed Gridfinity 1u baseplate.

The AI snapped this to a 6u by 4u layout. The interesting part was the scissors — the AI recognized them as elongated and gave them a 2u by 1u cavity with the handles oriented toward the front of the drawer (closer to the user). That heuristic, "elongated objects orient toward the user," is the kind of thing CAD won't do for you.

Print weight: about 220 g of PLA. Total time from photo to printed tray sitting in the drawer: 4 hours, of which 3 hours and 50 minutes was the printer running.

Project 3: Electronics bench small-parts tray

Resistor reels, capacitors, a half-spent roll of solder, two pairs of tweezers, an SMD pickup tool. Photographed on a green cutting mat (high contrast against PCB-yellow components helps segmentation).

The AI under-sized the resistor reels initially because the masks bled into each other where the reels touched. Solution: separate the reels by a centimeter on the photo and rerun. Took 30 seconds. The output gave each reel its own circular cavity with a finger relief at the side so you can spin and unspool without lifting it out.

For more on small-parts and bench layouts, see Gridfinity for Electronics.

Project 4: Camera lens caps and filters

Edge case: round, low-profile objects with shiny coatings. Lens caps reflect the room. The first photo had the AI mistaking the reflections for separate objects, generating ghost cavities.

The fix: take the photo under diffuse light (against a white sheet) and shoot at a slight angle to break up specular reflections, then let the AI's perspective correction handle the rest. The second pass produced a clean 3u by 2u tray with circular pockets.

This is the no-CAD workflow's weakness — reflective and translucent objects need a little staging. But "stage the photo better" is a faster fix than "model everything in Fusion 360."

Project 5: Workshop hand tools

Six screwdrivers, three pliers, a utility knife, a small hammer. The hardest of the five projects because the objects vary wildly in size — 100 mm screwdrivers next to a 250 mm hammer.

The AI's first pass put the hammer in a 2u by 6u cavity, which made the tray rectangular and awkward to fit in the drawer. Manual override in the preview: rotate the hammer 90 degrees, drop it into a 4u by 2u cavity in the corner, and the tray became a clean 4u by 5u that actually fit the drawer.

Lesson learned: the AI is good at packing but it doesn't know your drawer dimensions. Set those before generating, or override after.

What all five projects had in common

• A scale reference in the frame. Every successful project had a credit card, a 1u baseplate, or a printed reference square. Without it, the AI is guessing.
• A top-down shot. Angled photos work but introduce perspective error. Top-down is the safe default.
• One review pass before download. The AI gets about 90% of cells right; the 10% is obvious in the preview and takes seconds to fix.
• Diffuse lighting. Shadows confuse segmentation. Daylight or a bright overhead is enough.

When no-CAD beats CAD, and when it doesn't

No-CAD is faster for any tray with three or more objects. For a single-object tray you can grab from a library, the library wins on time. For collectibles or precision parts where 0.1 mm matters, traditional CAD plus calipers still wins. For everything else — shop drawers, bench trays, kitchen drawers, hobby boxes — the photo workflow is genuinely better.

For a foundational walkthrough on the photo-first approach, see Custom Gridfinity Bins Without CAD.

How GridPilot does this

GridPilot is built around the photo-first workflow described in these projects. Drop a phone photo in, get a Gridfinity-snap tray back, drag cells around in the preview if needed, download a 3MF.

Conclusion

The shortest path from "messy drawer" to "Gridfinity tray on the printer" runs through your phone camera, not Fusion 360. Five real projects, five photos, five trays. The barrier was never CAD skill — it was time, and the AI workflow takes that off the table.

Build your Gridfinity tray from a photo →