HVS Hybrid Vision

Our cameras are based on the AlpsenTek HVS (Hybrid Vision) chip. HVS fuses conventional imaging (APS) and events (EVS) inside the same pixel array, which is the core differentiator that sets this product apart from pure event cameras. For the event principle see Fundamentals, and for data formats see Data Formats.

1. What is HVS

HVS is not "two sensors glued together". It is every pixel in the same pixel array simultaneously carrying an image-sampling chain and an event-detection circuit. One chip therefore simultaneously outputs:

APS image stream: conventional full-image frames (RAW), with texture and color;
EVS event stream: asynchronous events (x, y, p, t), recording pixel-level brightness changes.

The two streams are synchronized at pixel level and microsecond level — there is no parallax, and no post-hoc alignment is needed.

2. Three operating modes

Mode	Output	When to use
Image mode (APS)	Full image frames (RAW)	Static scenes, when texture / color / detail is needed
Event mode (EVS)	Async event stream `(x, y, p, t)`	High-speed motion, low latency, low power
Fusion mode (HVS)	Image + event dual-stream sync	When both detail and fast response are needed (most real-world scenarios)

3. Pixel-level fusion vs sensor-level fusion

	Sensor-level fusion	Pixel-level fusion (HVS / ours)
Approach	Two independent sensors (APS + EVS) side by side	Image + event integrated inside the same pixel
Parallax	Yes (the two sensors see differently)	Zero parallax
Time sync	Needs post-hoc alignment	Native pixel-level, microsecond sync
Size / cost	Large / high	Single chip

Our chip integrates image and event inside the pixel with architectures such as iampCOMB / GESP / IN-PULSE DiADC combined with PixMUX time-division-multiplexed readout, using BSI stacking and inter-pixel isolation to achieve "sense-store-compute integrated" pixel-level fusion.

HVS pixel internal dual-path concept diagram

The photodiode splits into two paths — the image-sampling chain and the event-detection circuit — outputting APS frames and EVS events respectively.

4. Dual-stream synergy: see clearly + see fast

APS and EVS each have their strengths and complement each other once fused:

APS handles "seeing clearly": texture, color, static detail — "what the target is".
EVS handles "seeing fast": high-speed change, low latency, low bandwidth, >120dB high dynamic range — "how the target is moving".

Typical synergy: use APS to provide scene context (background, texture), and overlay high-speed motion trajectories on the same picture with EVS; or use EVS to trigger a low-power "there is movement" signal that wakes APS up to capture a frame.

Dual-stream synergy: fusing the APS image frame with the EVS event stream

The same high-speed motion scene: APS delivers the sharp image frame, EVS delivers the event-point trajectory, and the two streams fuse into one picture.

5. HVS vs pure EVS — how to choose

Pure EVS (the pure-event approach of Sony / Prophesee): pursues the ultimate event performance (higher peak frame rate, lower latency) but usually does not output images, suited to cost-insensitive, ultra-high-speed dedicated scenarios that do not need image context.
HVS (ours): one chip covers both image and event, with equivalent high-speed response + high dynamic range, low cost, small size, suited to most real-world scenarios that need both picture and event (recognition, surveillance, interaction, SLAM, etc.).

6. How to use it on our products

USB camera (e.g. CF-NRS1): APS + EVS dual stream output over USB3.0; MultiVision Studio can preview both streams on one screen.
MIPI module (e.g. EVS_003, based on APX003): the dual stream is output over MIPI CSI to the carrier board; dual_vc_vin_src / evs_live_player on RDK X5 provides real-time dual-stream display and recording.
Programmatic capture: APS goes through HV_Camera::startImageCapture, EVS through startEventCapture; starting both together is the minimal HVS dual-stream demo — see Programming Guide 07 — Capture APS Image. Full API at Toolkit API.