Comparison 02 · Audit sample

Robot stereo and depth camera evidence table

Range, resolution, transport and timing claims without collapsing unlike camera architectures.

11 component records6 comparison fieldsUpdated 2026-08-10
Scope

USB active/passive stereo cameras, rugged GMSL/GMSL2 cameras, PoE GigE stereo systems and industrial GigE time-of-flight cameras.

Do not flatten

Maximum range, ideal range, optimal range and accuracy-tested range are not interchangeable.

FieldRealSense
D455
Orbbec
Gemini 335
Stereolabs
ZED X
Basler
blaze-101
RealSense
D435i
Orbbec
Gemini 2
Stereolabs
ZED 2i 2.1mm
RealSense
D457
Orbbec
Gemini 2 XL
Stereolabs
ZED X Mini 2.2mm
Basler
blaze-102
Published rangeRange wording and conditions are preserved.0.6–6 mMManufacturer-described ideal range; not a hard detection limit.Source ↗0.10–20+ m; optimal 0.26–3 mCCurrent product-page wording; V1.6 datasheet states a 0.1–20 m+ working range and a 0.3–3 m ideal range. Both state a theoretical maximum up to 65 m.Source ↗2.2 mm: max 0.3–20 m / ideal 0.3–12 m; 4 mm: max 1–35 m / ideal 1–20 mMV1.2 lens-option table; maximum and ideal ranges remain separate.Source ↗Short 0.3–1.5; long 0.3–10 mMSeparate short- and long-range modes.Source ↗0.3–3 mMManufacturer-described ideal range, not a hard maximum.Source ↗0.15–10 mMProduct specification; ideal range is narrower.Source ↗0.3–20 mM2.1 mm lens; ideal range 0.3–12 m.Source ↗0.6–6 mMManufacturer-described ideal range, not a hard maximum.Source ↗0.4–20 mMManufacturer product specification.Source ↗Max 0.1–8 m; ideal 0.1–4 mMV1.2 datasheet, ZED X Mini 2.2 mm lens column.Source ↗0.3–1.5 m short mode; 0.3–10 m long modeMTwo separate operating modes.Source ↗
Resolution / frame rateMode combinations matter.1280 × 720 up to 30 fps; 848 × 480 up to 90 fpsMMaximum frame rate depends on resolution.Source ↗1280 × 800 @ 30 fpsMMaximum published depth mode.Source ↗1920 × 1200 @ 60 fps; 960 × 600 @ 120 fpsMThe two maxima are separate operating modes.Source ↗640 × 480 at 20 fps default / 30 fps FastModeMFastMode changes the operating mode; it is not the default.Source ↗Up to 1280 × 720; up to 90 fpsMResolution and frame-rate maxima are listed separately and are mode-dependent.Source ↗1280 × 800 at 30 fpsMMaximum listed mode.Source ↗2 × 2208 × 1242 at 15 fps; 2 × 1920 × 1080 at 30 fps; 2 × 1280 × 720 at 60 fpsMSeparate modes.Source ↗Up to 1280 × 720; up to 90 fpsMResolution and frame-rate maxima are listed separately.Source ↗1280 × 800 at 10 fps; 640 × 400 at 20 fpsMSeparate modes.Source ↗2 × 1920 × 1200 at 60 fpsMMaximum listed mode.Source ↗640 × 480; 20 fps default / 30 fps fastMFrame rate depends on acquisition mode.Source ↗
Field of view / lensLens option and axis definitions matter.87 × 58 °MHorizontal × vertical.Source ↗90 × 65 °MHorizontal × vertical.Source ↗2 mm wide-FOV or 4 mm long-range lensMField of view depends on ordered lens option.Source ↗67 × 51 °MHorizontal × vertical.Source ↗87 × 58 °MHorizontal × vertical.Source ↗91 × 66 °MHorizontal × vertical.Source ↗110 × 70 × 120 °MHorizontal × vertical × diagonal; 2.1 mm lens.Source ↗87 × 58 °MHorizontal × vertical.Source ↗91 × 65 °MHorizontal × vertical.Source ↗110 × 80 × 120 °MHorizontal × vertical × diagonal; 2.2 mm lens.Source ↗67 × 51 °MHorizontal × vertical.Source ↗
Data interfaceGMSL2 and USB require different host hardware.USB-C 3.1 Gen 1MNo extra condition in selected source.Source ↗USB 3.0 Type-CMNo extra condition in selected source.Source ↗GMSL2MRequires a compatible deserializer or capture platform.Source ↗Gigabit Ethernet / GigE VisionM1000 Mbit/s communication interface.Source ↗UnknownNo adequate published value in the selected source set.USB 3.0 Type-CMNo extra condition in selected source.Source ↗USB 3.0 Type-CMExternal screw-lock cable.Source ↗GMSL over FAKRA; cable length up to 15 mMRequires compatible serializer/deserializer host hardware.Source ↗USB 2 Type-C; PoE Gigabit EthernetMOBox and cable configuration affect mass and integration.Source ↗GMSL2MRequires a compatible GMSL2 capture path.Source ↗Gigabit EthernetMIndustrial Ethernet camera transport.Source ↗
Accuracy / precisionOnly comparable with the same distance, target and metric.<2% at 4 mMManufacturer condition shown on the product page; see datasheet footnote for full method.Source ↗≤1.5% RMSEM1280 × 800 at 2 m over a 90% × 90% ROI on a >80% reflectivity plane; this is spatial noise, not the separate absolute depth-accuracy row.Source ↗2.2 mm: <0.8% at 2 m / <4% at 12 m; 4 mm: <0.4% at 2 m / <7% at 20 mMV1.2 lens-option table; metric definition, target, ROI, acquisition mode and test environment are not disclosed beside the values.Source ↗±5 mmM0.5–5.5 m; 20-minute warm-up; flat white 90% target; no ambient light; 22°C; central 40 × 40 pixels; 25-image average.Source ↗<2%MAt 2 m; manufacturer product-page value.Source ↗≤2%M1280 × 800 at 2 m over 81% ROI on a >80% reflectivity plane. The datasheet describes RMS distance to a best-fit plane without a ground-truth reference.Source ↗<0.8% at 2 m; <4% at 12 mM2.1 mm lens claims.Source ↗<2%MAt 4 m; manufacturer product-page condition.Source ↗≤2%M1280 × 800 at 4 m over 81% ROI on a >80% reflectivity plane. The datasheet describes RMS distance to a best-fit plane without a ground-truth reference.Source ↗<1.6% at 2 m; <3% at 4 mMV1.2 datasheet, 2.2 mm lens. Metric definition, target, ROI, acquisition mode and test environment are not disclosed beside the values.Source ↗±5 mmMAt 0.5–5.5 m under the documentation’s specified operating conditions.Source ↗
Timing / syncFeature support is not proof of achieved multi-sensor timing.Bosch BMI055 IMU; multi-camera use describedMProduct page does not turn this into a universal host-level synchronization guarantee.Source ↗Depth/color sync, multi-device sync and hardware trigger supportedMHost, cable and SDK configuration still determine achieved timing.Source ↗External frame syncMIntegration details depend on host hardware and cable topology.Source ↗PTP IEEE 1588, hardware trigger, software trigger, free runMMulti-camera operation is supported, but achieved timing depends on network and configuration.Source ↗UnknownNo adequate published value in the selected source set.8-pin sync; ≤3 ms depth/RGB differenceMAuto exposure off; follow SDK topology instructions.Source ↗UnknownNo adequate published value in the selected source set.UnknownNo adequate published value in the selected source set.External sync and triggerMProduct-level support; achieved timing depends on topology and software.Source ↗Hardware synchronization and triggerMFeature support; achieved system timing still requires validation.Source ↗Free run, IEEE 1588 PTP, software and hardware triggerMSupported modes; system timing still requires validation.Source ↗

Values are transcribed from the linked sources and were checked on the date shown by each record. V = independently verified; M = manufacturer claimed; D = RCI derived; C = conflicting; U = unknown.

Read the range and error-claim audit

RCI 017 separates 25 range observations and 16 selected error or precision values across these 11 records, then checks 13 fields needed to reproduce the public claim. No record publishes a complete reproduction package. Open the depth-camera audit →

Interpretation rule

The condition belongs to the number.

RCI does not compute winners from this audit sample. The table’s purpose is to show what can be compared now, what requires a shared test, and which public fields are still missing.

Read the full method →