Caddx vs Runcam FPV Camera Comparison: Image Quality, WDR, Low-Light, and Latency — 2026 Guide

The FPV camera is the one component you experience for every second of every flight, yet pilots spend more time researching motors. A camera with poor wide dynamic range (WDR) turns a sunset flight into a choice between seeing the sky or seeing the ground — never both. A camera with glass latency above 20ms makes proximity flying feel drunk. Here’s how Caddx and Runcam stack up across the metrics that actually matter in the air, based on real flight testing.

The Sensor War: Sony Starvis 2 vs Everything Else

Both Caddx and Runcam license Sony sensors for their flagship models, but sensor choice matters more than brand:

• Sony Starvis 2 (IMX662/IMX585): The gold standard in 2025-2026. Dual-gain pixel architecture captures detail in shadows and highlights simultaneously — something single-gain sensors literally cannot do. Any camera with a Starvis 2 sensor will outperform any camera without one in dynamic range, regardless of brand or price.
• Sony Starvis 1 (IMX327/IMX385): Still excellent. The IMX327 in the Runcam Phoenix 2 and Caddx Ratel 2 delivers 90% of Starvis 2’s WDR at 60% of the price. For pilots who fly in consistent lighting (midday, overcast, or indoor), Starvis 1 is indistinguishable from Starvis 2.
• Non-Sony sensors: Found in budget cameras like the Caddx Ant and Runcam Nano. Acceptable in perfect lighting, useless in mixed light — a cloud passing overhead washes out the entire image.

Model-by-Model Comparison

Caddx Ratel 2 (Starvis 1 IMX327)

The Ratel 2 has been my daily driver on three quads for two years. 1200TVL resolution, switchable 4:3/16:9 aspect ratio via a physical button on the back of the camera (not in the OSD menu — you have to remove the camera from the frame to switch, which is annoying). WDR is excellent for a Starvis 1 sensor; flying through tree canopies with dappled sunlight, I can see branches and sky detail simultaneously. Glass latency is 12ms — competitive with everything except dedicated racing cameras.

Weakness: The 2.1mm stock lens has noticeable barrel distortion at the edges. Swapping to a 2.5mm GoPro-style lens fixes it but narrows FOV from 165° to 135°, which changes how you judge proximity.

Runcam Phoenix 2 (Starvis 1 IMX327)

Same sensor, different tuning. Runcam’s image processing leans cooler — blues and greens are more saturated, which helps with grass/foliage contrast but makes skin tones look washed out (relevant if you film people at events). The OSD menu is navigated with a single joystick on the included control board; Caddx uses a 5-button pad. I find the joystick faster but less precise — overshooting the setting you want is common until muscle memory kicks in.

The Phoenix 2’s low-light performance edges out the Ratel 2 by about half a stop. At dusk, I can fly the Phoenix 2 5 minutes later than the Ratel 2 before the image becomes unusable. Both cameras have a “night mode” that drops frame rate to boost exposure — useful for parking garage flying, not for actual night flying.

Caddx Walnut (Starvis 2 IMX662)

Caddx’s budget Starvis 2 offering at roughly $35. The WDR jump from the Ratel 2 is immediately noticeable — transitions from shadow to sunlight don’t wash out or black-crush. You see into shadows while still seeing sky detail. The tradeoff: the Walnut’s lens is fixed-focus and not replaceable, and the housing is plastic (Ratel 2 uses aluminum). In a crash, the lens barrel is the first thing to break.

Runcam Phoenix 3 (Starvis 2 IMX585)

Runcam’s answer to the Starvis 2 market. The IMX585 is a larger sensor (1/1.2″ vs 1/1.8″ on the IMX662), which means better light gathering and lower noise. In practical terms: flying through a parking garage at noon with the exit in frame, the Phoenix 3 shows detail inside the garage AND the outside world — the Ratel 2 would show a white rectangle where the exit is. Glass latency is 14ms, slightly higher than Starvis 1 cameras but imperceptible outside of championship-level racing.

The Phoenix 3’s OSD menu adds a “User” preset slot that saves all your image adjustments — brightness, contrast, saturation, WDR strength, sharpness — and lets you toggle between presets mid-flight via a spare UART control channel. This is the killer feature for pilots who fly in varying conditions: one preset for sunny, one for overcast, one for dusk, switchable from your radio.

Caddx Ant Nano (Budget, no Starvis)

At 1.7g and $18, the Ant Nano is the go-to for ultralight toothpick builds where every gram counts. Image quality is mediocre — fine for flying, bad for watching. WDR doesn’t exist; fly toward the sun and the entire image goes white until you turn away. Acceptable for racing and proximity where you need minimum weight, unacceptable for anything you plan to record.

Runcam Split 4 (4K Recording)

Hybrid FPV cam + onboard DVR at 4K30. The FPV feed is clean but standard-def — the split architecture means the image sensor feeds both the analog transmitter AND the onboard recorder, and the analog path is limited by NTSC/PAL resolution. If you want recorded footage that looks good, the Split 4 delivers — far better than DVR from goggles. But the FPV feed itself is no better than a $30 Ratel 2.

Parameter Comparison Table

Common Mistakes When Choosing an FPV Camera

Mistake 1: Chasing TVL numbers. 1200TVL vs 1500TVL sounds like a big jump, but your analog VTX transmits at roughly 600-700TVL equivalent after modulation and transmission losses. The extra resolution matters for on-board DVR (like the Split 4) but not for the live FPV feed. Choose based on sensor and WDR, not TVL.

Mistake 2: Ignoring glass latency for freestyle. Freestyle quad pilots don’t think about latency the way racers do, but a 22ms camera (like an older Runcam Eagle) vs a 12ms camera adds 10ms of control lag. At 30 m/s, that’s 30 centimeters of position error — the difference between threading a gap and clipping a branch. If you fly proximity, camera latency matters more than motor Kv.

Mistake 3: Buying the same camera for every build. A 5.8g Phoenix 2 on a 19g toothpick is absurd — the camera alone is 30% of the dry weight. Match camera to build: Ant Nano or similar ultralight for toothpicks, Ratel 2 / Phoenix 2 for 3-5 inch freestyle, Phoenix 3 for long-range where mixed lighting is guaranteed.

Mistake 4: Not checking lens thread compatibility. Caddx and Runcam use M12 lens threads (standard), but the lens housing depth varies between models. A 2.5mm lens that works on the Ratel 2 may not reach focus on the Phoenix 3 because the sensor sits deeper in the housing. Test fit before ordering replacement lenses.

Mistake 5: Judging cameras by bench testing. WDR advantages of Starvis 2 are invisible on the bench under indoor lighting. You need real-world mixed lighting — flying in and out of shadows, toward and away from the sun — to see the difference. Bench tests make a $18 Ant Nano look comparable to a $45 Phoenix 3. They’re not.

⚠️ Regulatory Notice: The flight recommendations in this article should be followed in accordance with the latest 2026 drone regulations in your country or region. Always verify local laws regarding flight altitude, no-fly zones, remote ID requirements, and registration before flying. Regulations vary significantly between the FAA (US), EASA (EU), CAA (UK), CAAC (China), and other authorities.

Recommended Hardware

The Caddx Walnut hits the sweet spot for most pilots — Starvis 2 WDR at just $35, with glass latency that keeps up with aggressive flying. We stock the full Caddx and Runcam lineup at uavmodel.com, including replacement lenses and control boards.