Pixboom Spark Review – First Hands-On with Beta Version of the High-Speed Cinema Camera

by Nino LeitnerOctober 23^rd, 2025

Subscribe to Reviews - Cameras

We took the Pixboom Spark pre-production unit into the wild and then onto the bench: butterflies in India with Gunther Machu, followed by a studio deep-dive to check what this new Super 35 BSI global-shutter camera can really do. Here is what worked already, what still needs polish, and where this could slot into real productions. Let’s dive into our Pixboom Spark review!

The Spark is a new high-speed camera from Chinese startup Pixboom. We first met the team at IBC (video interview here) after their preview at Cine Gear, and they loaned us an early beta for a week. Gunther took that unit to Goa to film butterflies and damselflies, then we sat down in the CineD studio for a thorough post-mortem and to review a short cut of his footage.

Sensor, frame rates and resolution

Spark uses a Super 35, backside-illuminated, global-shutter CMOS with an open-gate resolution of 4608 × 3072. In open gate it tops out around 670 fps. As you reduce the vertical readout while keeping full width, frame rates rise: 2160p is roughly 937 fps, 1080p reaches about 1820 fps, and around 768p clears 2400 fps. The global shutter eliminates rolling-shutter skew and allows unusually clean handheld shots at very high speeds, since motion is slowed so much that minor shakes vanish.

Field test in India: butterflies in slow motion

Gunther’s brief was simple, capture wing mechanics you cannot see at 4K/120. Macro and long lenses at high magnifications demand light and precision. In the field, the Spark held a rock-steady sensor temperature readout of approximately 42 °C despite heat and humidity, and with a 99 Wh V-mount battery it ran two to three hours while left powered to avoid the roughly 35-second boot. Global shutter plus very high frame rates made handheld usable for wildlife behavior and for low-angle macro crawls along the forest floor.

Image quality and dynamic range

This beta firmware reads the sensor in 10-bit raw. Our quick check with a Xyla 21 showed about 10 to 11 stops visible on the waveform, with roughly 8 to 9 stops usefully distributed before noise and clipping push in. Highlights must be protected, otherwise the image can feel contrast-heavy, similar to early DSLR high-speed looks. Noise has a fine, film-like character and, at this stage, the camera does not apply sensor-level noise reduction, which we consider a positive for raw control in post. We really hope they don’t add it for the final version of the camera!

Gunther and me both noted a faint horizontal line across some shots. The sensor apparently reads the top and bottom halves separately, and subtle non-uniformity at the seam can reveal a line at the match. Pixboom told us they will mitigate this upstream in sensor QA and downstream in processing. Some other beta users have reported fixed pattern noise in some occasions, but we have not observed this.

We will re-test this and all other aspects in a production camera once it becomes available, but we are confident that those issues will be addressed properly before units ship.

We think that the addition of an IR-cut filter is advisable. The Spark’s OLPF stack on our unit did not strongly suppress IR, so strong sources and heavy lighting can introduce IR contamination, especially in the shadows.

ISO behavior and shutter considerations

Pixboom describes five “native” analog gain positions, not digital pushes: ISO 400, 640, 800, 1250, and 1600. In practice this let us lift sensitivity without obviously trading highlight headroom at those set points. High-speed exposure still follows the usual math, for example 2400 fps implies a 180-degree shutter near 1/4800 s, so plan lighting accordingly, especially for macro at T5.6–T8.

Recording, media and workflow

A key differentiator is continuous recording directly to a proprietary 2.5 TB SSD cartridge. In open gate at around 600 fps, that yields about seven minutes of raw capture, which is a lot of action in super slow motion. The cartridge includes a USB-C 3.2 Gen 2 port, so it mounts like an external drive for fast offloads in the 1.0–1.2 GB/s range in our experience. The current beta lacks in-camera trimming, end or mid-trigger, or pre-roll options, which means you can fill media quickly while waiting for behavior. Pixboom says circular buffer and pre-record are planned, and they really are a “must” in a high-speed camera.

Conversion is via Pixboom’s app to CinemaDNG, and batch conversion and setting in and out points worked flawlessly. From there we graded in Premiere Pro using the Octopus RAW Studio plug-in, or in Resolve. A future 12-bit readout mode is planned for the Pixboom Spark, likely at reduced maximum frame rates, which should benefit tonal smoothness and usable dynamic range.

Mounts, lenses and anamorphic options

The Spark’s body accepts interchangeable mounts. We tested E-mount with electronic iris control through adapters and also ran EF glass with a Metabones. There is no autofocus, which we missed most with long wildlife lenses where depth of field at macro distances is measured in millimeters. The 3:2 sensor makes anamorphic practical, and the camera already offers in-camera de-squeeze ratios.

Ergonomics, power and sound

The body is compact, essentially a dense cube that rigs easily. The active cooling works, but the fan is loud, so Spark is not a sync-sound camera, and it has no audio inputs. Think of it as a dedicated motion tool that you roll in for select shots. An on-body UI is functional. Remote control is evolving, with USB-C to Ethernet browser control available on development builds and Wi-Fi plus Bluetooth hardware inside for a coming mobile app.

What still needs work

This was an early beta, so some gaps are expected. Our main asks are end or mid-trigger with circular buffer and pre-record, in-camera clip trimming and delete, black shading and fixed-pattern calibration finalized, optional OLPF or IR-cut in the optical stack, and some exposure tools like false color or a waveform. Autofocus is not promised, yet even a limited fast AF mode for static macro setups would be a killer feature for specialty work.

Pricing and availability context

During the campaign, prices discussed started around 8,000 USD, with a stated target retail around 13,000 USD including the 2.5 TB media. However, there is still a “pre-order special” available on the Pixboom website for just below $10,000. Exact bundles and timings will depend on production. Either way, Spark lands well below legacy high-speed cinema options, and below or near some competing compact high-speed systems, while offering continuous recording to internal media and a modern global-shutter BSI sensor.

Pixboom recently also hosted a Montreal workshop where they shot and uploaded extreme high-speed tests and walked through workflow and planned firmware features, check it out here:

Who is the Pixboom Spark for?

Commercial tabletop, product beauty, sports science inserts, wildlife behavior, VFX elements, water and glass work, and any time you need to turn a blink into a symphony. Owner-operators can finally justify bringing a true high-speed tool to set at this price point. You’ll be able to treat it like a specialist camera that complements your A-cam, not a replacement for it.

CineD Lab Test note

We only run full CineD Lab tests on production firmware, and then those results also end up in our Camera Database. Once Pixboom ships a production unit, we will measure dynamic range, rolling behavior confirmation, latitude, and more for our database.

Pixboom Spark review – verdict

Pixboom Spark is already delivering the one thing that matters in high speed, reliable capture at serious frame rates with a clean global-shutter image and a workable workflow. This looks like it can be the affordable high-speed camera everyone has been waiting for, for such a long time, and we applaud the team behind this camera. The firmware still needs to catch up on triggers and calibration, yet the foundation looks strong – and we look forward to working extensively with the production version once it’s ready.

What would you shoot first if you could freeze time this way? Let us know in the comments.