Tuning

Task goal

The imaging quality of an event camera is determined jointly by several groups of parameters. This page covers on-the-spot tuning decisions for these parameters — what symptom maps to what parameter. Conceptual content (what bias is, the meaning of accumulation time) is at Bias & Tuning and Fundamentals; this page focuses on "symptom → action".

Three main tuning dimensions

Dimension	What it controls	Where to tune
bias / sensitivity	The pixel event-firing threshold (fire or not)	Bias & Tuning
accumulation time	The window for accumulating events into a frame (smeared vs empty)	The clear period in Guide 05
ROI (Region of Interest)	Output only some pixels, reducing data volume and compute	see below

On-the-spot tuning decision tree

Address the symptom directly:

Symptom	Cause	Action
Picture is "dirty", full of noise when static	Sensitivity too high / heavy background activity noise	Lower the sensitivity (raise the threshold), or plug in denoising
Missing fast / faint targets	Sensitivity too low	Raise the sensitivity slightly (lower the threshold)
Picture smeared, heavy motion trail	Accumulation time too long	Shorten the accumulation window (smaller clear period in Guide 05)
Picture nearly all black, no events	Accumulation time too short or sensitivity too low	Lengthen the accumulation window, or raise the sensitivity
Data volume too large, processing can't keep up	Full-frame output + high sensitivity	Enable ROI, or lower the sensitivity, or add denoising

Balancing noise and sensitivity

Raising the sensitivity reveals weaker targets but also increases noise. A common combination is: keep sensitivity fairly high + plug in a denoise algorithm, so you don't miss targets but suppress noise. For the bias concept and trade-offs see Bias & Tuning.

Programmatic entry points for bias / ROI

The currently public HV Toolkit DeviceConfig has no general-purpose bias/ROI setting methods, and this page will not invent interfaces that do not exist. The recommended approach is to adjust bias and display parameters in the host software MultiVision Studio, confirm the effect, then bake the values into your project through the corresponding product SDK's interface.

The bias items, default values, and ROI capabilities of S100 and X5 are determined by their respective sensor/carrier-board SDKs and cannot be copied across platforms; if your application needs programmatic tuning, refer to the corresponding platform SDK and product specification.