Buy a dedicated 3D scanner when you will capture physical parts repeatedly and need faster setup, live coverage feedback, or more consistent results on small and medium objects. Use photogrammetry when the budget is tight, the subject is textured and easy to photograph, the job is occasional, or you want to prove that scanning helps before buying hardware.
Neither option automatically creates a dimensionally exact, print-ready replacement part. Both usually produce a mesh that must be cleaned, aligned, scaled, checked, and sometimes rebuilt as CAD. The right purchase depends less on the best demo scan and more on the objects you will capture, the accuracy you need, and how much cleanup time you can accept.
Quick verdict
| Your real use case | Better default | Why |
|---|---|---|
| Repeat capture of brackets, housings, grips, trim pieces, or medium mechanical parts | Dedicated 3D scanner | Live tracking and coverage feedback can reduce capture time and missed areas. |
| Occasional statues, terrain, props, rocks, furniture, or textured objects | Photogrammetry | A capable phone or camera may be enough when texture and lighting give the software strong features to match. |
| Small plain part with bores, slots, snap fits, and critical mating faces | Scanner plus manual measurement | Capture can provide the overall form, but calipers and CAD should control functional dimensions. |
| Shiny, black, transparent, or featureless object | Neither is effortless | Surface preparation, markers, controlled lighting, or another measurement method may be required. |
| One uncertain project and no established scan-to-CAD workflow | Photogrammetry or outsourced capture | Validate the value of the mesh before committing to dedicated hardware. |
What a dedicated 3D scanner buys you
A handheld scanner is designed around guided capture. Depending on the model, it may project structured light or use another optical depth method while its software shows whether the object is tracking, which areas are covered, and where another pass is needed. That feedback is the practical advantage. It can turn a slow photo-planning exercise into a repeatable bench workflow.
A scanner is most valuable when the same operator captures similar parts every week. Faster reacquisition after tracking loss, controlled working distance, turntable modes, marker support, and a predictable software pipeline can save more than the hardware costs over time. The purchase makes less sense if the scanner spends months in a drawer or if every job still requires hours of manual remodeling.
Software matters as much as the sensor. Before buying, inspect the current export formats, computer requirements, licensing terms, alignment tools, hole filling, decimation, scale controls, and update history for the exact model. The Creality Scan overview is one example of the software side buyers should evaluate rather than treating a scanner as a self-contained appliance.
What photogrammetry buys you
Photogrammetry reconstructs geometry by matching visible features across many overlapping photographs. It can start with hardware you already own, making it the lower-risk way to test whether a captured mesh helps your design process. It also scales well to larger textured subjects that are awkward to sweep with a small-part scanner.
The method rewards disciplined photography. The object must stay still while the camera moves through overlapping viewpoints. Exposure, focus, white balance, background separation, and coverage under overhangs all affect reconstruction. More photographs are not automatically better if they are blurred, reflective, repetitive, or taken from nearly identical positions.
Phone apps can make the process approachable, but do not assume every phone-based scan is photogrammetry. Some devices and apps may also use depth sensors or combine methods. What matters is the exported geometry, scale reliability, cleanup burden, and whether the workflow retains enough detail for the next step. The Scaniverse workflow guide shows the kind of capture-to-export path to inspect before purchasing separate hardware.
Accuracy is not the same as a printable CAD model
A dense mesh can look convincing while still being wrong where fit matters. Rounded noise can shift a hole center. A filled shadow can invent a surface. Automatic smoothing can soften an edge that was meant to locate against another part. Global scale can be close while two critical features remain too far apart.
For decorative duplication, visual similarity may be enough. For a replacement housing, latch, bearing seat, gasket groove, or bolt pattern, the mesh is usually reference evidence rather than final authority. Record known dimensions with calipers, gauges, or the mating assembly and use them to scale and verify the capture.
If the object will become a functional replacement, the reverse-engineering guide explains why a scan often needs to be rebuilt around design intent. The companion guide to critical replacement-part dimensions helps separate surfaces that merely look right from the features that control fit.
Object size and surface finish can decide the winner
Small mechanical parts
Small parts are often harder than scanner marketing suggests. Thin walls, deep holes, narrow slots, sharp edges, and hidden mating faces may fall below useful capture resolution or remain outside line of sight. A scanner that performs well on a shoe, bust, or medium housing may not be the right tool for a tiny clip. Manual measurement and parametric CAD can be faster for simple geometry.
Medium objects
Medium matte objects with varied geometry are the strongest dedicated-scanner lane. The operator can move around the part, inspect live coverage, and add targeted passes. Photogrammetry can still work well, but the scanner may win on throughput when many objects share this size and surface behavior.
Large textured objects
Large sculptures, terrain, architectural details, and props can favor photogrammetry because the camera can move farther away and capture broad overlapping coverage. Processing time and image organization increase, but the method does not require a small sensor to maintain one narrow working distance across the whole subject.
Difficult surfaces
Glossy, transparent, very dark, reflective, repetitive, or nearly featureless surfaces can defeat both workflows. Removable scanning spray, temporary matte coatings, coded targets, a textured background, cross-polarized lighting, or careful exposure may help when compatible with the object. Never coat a valuable, porous, delicate, or safety-critical item without confirming that the preparation can be removed safely and will not damage it.
Compare the full workflow, not the capture button
The useful output is not the first point cloud or preview mesh. It is a verified model that can support the intended print. Time the complete chain:
- Prepare the object, background, lighting, markers, and turntable if used.
- Capture all visible surfaces without moving the object unexpectedly.
- Align passes or photographs and remove the bench, supports, and stray geometry.
- Merge data, close only justified holes, reduce noise, and preserve important edges.
- Set scale from a trustworthy measurement.
- Rebuild functional geometry in CAD when the mesh cannot define it reliably.
- Print a test piece and check it against the mating assembly.
A scanner may shorten steps two and three while leaving steps four through seven nearly unchanged. Photogrammetry may cost little in hardware but demand more capture discipline and compute time. Compare total hands-on minutes per accepted model rather than scanner specifications alone.
Do not ignore computer, software, and file costs
A dedicated scanner can require a particular operating system, graphics capability, port, cable length, or mobile device. Photogrammetry can require substantial storage, upload bandwidth, cloud processing, or a capable local computer. Some apps limit export resolution or file formats behind a subscription. Check the exact current terms before building a workflow around them.
Also confirm what your CAD and mesh tools can ingest. STL and OBJ are common mesh formats; STEP is normally expected for editable solid or surface CAD, not as the automatic result of an optical scan. Converting a triangulated mesh to STEP does not magically recover clean cylinders, design history, or tolerances. If the model is going to another party, the guide on sending STL, STEP, or both explains why the file type should match the work being requested.
When buying a scanner is worth it
A dedicated scanner is usually worth serious consideration when most of these are true:
- You have a recurring queue of physical parts rather than one curiosity project.
- The objects sit inside the scanner's demonstrated size and surface range.
- Live tracking and coverage feedback will reduce operator time.
- You can verify critical dimensions independently.
- Your computer supports the current software comfortably.
- The export and cleanup pipeline fits your CAD tools.
- You have budget for targets, surface preparation, fixtures, and training as well as the scanner.
Do not buy primarily for the advertised point spacing or accuracy number. Ask how that figure was measured, at what object size and distance, under what surface conditions, and whether the result describes repeatability, local detail, or whole-object dimensional error. A controlled sample of your own part is more useful than a specification detached from your workflow.
When photogrammetry is the smarter first move
Photogrammetry is the better starting point when capture is occasional, the object has visible texture, you already own a suitable camera, or the main goal is to create an artistic or spatial reference. It is also a sensible proof-of-concept. If a phone-based mesh does not improve the downstream modeling work, dedicated hardware may not solve the underlying problem.
It can also be the better format for large subjects where walking a camera around the object is easier than maintaining a scanner's tracking distance. The trade is process control: consistent images, adequate overlap, nonchanging light, careful background treatment, and patience during reconstruction.
If all you need is a quote or feasibility check, a full scan may be unnecessary. Good photos, a few trustworthy dimensions, and the intact mating assembly can sometimes communicate more useful evidence. The guide to requesting a quote from images and dimensions shows where simple evidence works and where a real model is still needed.
How to test the decision before buying
- Select the hardest representative object from your real six-month queue, not a textured demo object chosen to flatter scanning.
- Try a phone or camera photogrammetry workflow under controlled light and record capture, processing, and cleanup time.
- Measure five critical features independently and compare the cleaned mesh at those locations.
- Have the same object scanned by a dealer, service, makerspace, or borrowed machine using the exact model you may buy.
- Import both results into the CAD or mesh software you will actually use.
- Build or repair one print-ready model, make a test print, and check fit on the real assembly.
- Multiply the time saved by the number of realistic jobs per year and compare that value with the complete hardware and software cost.
This test exposes the real bottleneck. If capture is slow but cleanup is easy, a scanner may pay off. If both meshes still require the same long CAD rebuild, improving modeling skills or outsourcing occasional capture may be the better investment.
Bottom line
Buy a 3D scanner for repeatable capture throughput, not because you expect one-click printable CAD. Use photogrammetry when you need a low-cost starting point, have textured or large subjects, and can trade more capture discipline and processing time for less hardware expense.
For functional 3D printing, use either mesh as measured reference data. Verify scale, holes, mating faces, wall positions, clearances, and load paths independently. The cleanest buyer decision is the one that produces an accepted part faster across your real queue, not the one that creates the most impressive first preview.
Frequently asked questions
Is a 3D scanner more accurate than phone photogrammetry?
Often, but not automatically. A suitable scanner can provide more controlled geometry and repeatable tracking on objects inside its intended range. Surface finish, calibration, distance, alignment, preparation, and software still matter. Verify critical dimensions instead of assuming the denser-looking mesh is correct.
Can photogrammetry create an STL for 3D printing?
Yes, many workflows can export a mesh that can become an STL. It may still need scaling, cleanup, hole repair, decimation, wall-thickness work, and CAD remodeling before it is printable or dimensionally useful.
Can a 3D scanner copy a broken replacement part?
It can capture surviving geometry, but it cannot directly know the missing shape, original design intent, tolerance, or load case. Opposite-side geometry, mating parts, photographs, measurements, and test fitting may still be required.
Do you need LiDAR for photogrammetry?
No. Photogrammetry can reconstruct geometry from overlapping photographs alone. Some phone apps may combine photographs with depth-sensor data, but the value should be judged by the exported mesh and its verified accuracy for the intended job.
What should you scan before buying a 3D scanner?
Use a representative difficult object from your real queue: the expected size, surface, recess depth, and required fit. A dealer demo on an easy textured bust does not prove performance on your black housing, thin clip, or reflective trim piece.