Frequently Asked Questions 

General Questions 

What is a DVS?

DVS (Dynamic Vision Sensor) is a type of sensor that transmits only pixel-level changes. Unlike traditional image sensors that capture full frames at a fixed rate, DVS sensors are much faster and more efficient. Additionally, they have a very high dynamic range and a very low latency. Frames Events GIF
Davide Scaramuzza, Robotics and Perception Group, University of Zurich

Below is a comparison between a more realistic frame and the corresponding event output with an explanation of what an event is.

Realistic Image from Frame Camera	Corresponding Output from Event Camera

Read more about events and temporal resolution in our White Paper.

How do I use my Event Camera?

iniVation cameras are usable with our different software solutions.

I think I need both events and frames in my applications. Which camera should I choose?

All iniVation cameras provide pixel-level event output. In addition, all iniVation cameras support software reconstructed frames of intensity images, using the event output processed by iniVation’s DV Software. DV Software contains several algorithms to reconstruct frames of intensity images from events. Each of these algorithms has different characteristics affecting image quality and processor requirements.

In addition, the DAVIS346 camera can directly output conventional frames of hardware-measured intensity images. These are the same as the frames from conventional image sensors. The conventional frame output from the DAVIS346 is limited to 55 dB dynamic range and 40 FPS frame rate. Please refer to the DAVIS346 specifications for further details.

Software reconstructed frames (from events) work best for specific use cases where image quality is not critical and/or ultra-high speed is required. Some examples of such use cases are as follows:

General debugging output for users
Occasional camera calibration, lens focusing, general setup
Basic recognition tasks where the user is collecting their own training data
Extremely high-speed visualization that is not easily possible using conventional frames

Comparing to conventional frames, software reconstructed frames have the following pros and cons :

Pros:

Inherently HDR (>90 dB)
Variable frame rates (up to approx. 1k FPS depending on available processing power)
Robust reconstruction depending on the use case
Can be improved via software updates

Cons:

Lower image quality than conventional frames
Image artefacts (reconstruction algorithm dependent)
Unable to reconstruct static scene information
Requires extra computational power

If image quality is of primary importance to a certain use case, e.g. difficult image recognition, it is best to check beforehand to see if software reconstructed frames will be of sufficient quality. If you are unsure, please get in touch with our support team, and we will provide you with the best camera recommendation for your use case.

Can DVS cameras see infrared (IR)?

DVS sensors can see all visible light plus near-infrared (NIR). The luminosity function should be that of a standard CMOS sensor, sensitivity peaking around 700-750 nm and going until about 1100 nm wavelength (however, the highest wavelengths we tested were around 980nm). The default lens shipped with the camera may contain an IR filter. In order to see IR light, you may need to use a different lens.

Hardware Questions 

My camera is not working, what should I do?

If you can’t seem to access your camera, please try the actions below:

Ensure the camera is properly connected with the provided cable or a decent equivalent.
Try switching USB ports, preferably USB 3.0.
Check your USB permissions.
Check if your camera firmware can be updated.
If you are using DV GUI:
- Make sure DV is up-to-date to its latest version.
- Make sure to properly select the camera in the list of devices before running any other modules.

If none of this works, then please contact us through our support e-mail so that we can investigate your issue further.

Does iniVation offer drivers for the cameras?

On Linux and macOS, no driver is necessary. The cameras work out of the box.

On Windows, the driver should get installed automatically once you plug in the device. If that doesn’t work, install the driver manually by following this guide.

Can I use an external trigger signal?

Yes, DAVIS346 and DVXplorer support connecting an external trigger signal. Usage is documented here. Please note the DAVIS346 AER and the DVXplorer Micro do not support external trigger signals.

If you require a sync cable, we recommend the following supplier.

Can I synchronise timestamps between two cameras?

Yes, that is one of the purposes of the sync connector available on the DAVIS346 and DVXplorer. Usage is documented here. Please note the DAVIS346 AER and the DVXplorer Micro have no support for multi-camera timestamp synchronization.

What is the spectral sensitivity / quantum efficiency (QE) of the iniVation sensors?

The spectral sensitivity / quantum efficiency (QE) of DAVIS346 mono is published in https://ieeexplore.ieee.org/document/8334288 (FSI DAVIS curve in Fig. 3). The spectral sensitivity / QE of DAVIS346 color is not available, but the QE of the same color filters and similar pixel designs (same photodiode size) have been characterized in this thesis https://www.research-collection.ethz.ch/handle/20.500.11850/156363 (Chapter 5 Fig. 31 B, Page 71). So the QE of DAVIS346 color should be very similar to the R, G, and B (without W) channels in that figure. The QE specs of DVXplorer or DVXplorer Lite are not available. Their QEs are not directly measurable because their pixels have no intensity output.

More information specific to InfraRed light is available here.

How can I change the biases of my camera to, for example, increase sensitivity or slow down event output?

Biasing documentation is available here.

Why does my DAVIS camera give out frames that are all white / have vertical lines / look corrupted?

DAVIS camera frames are known to have issues:

All white frames are generally caused by overexposure of the camera. Try to:
- Not use it while looking directly at a very bright light source.
- Change the frame exposure settings.
Vertical lines in the frame might be caused by using the rolling shutter mode. Make sure to use global shutter setting.
Refer to this question to check that you are using your camera properly.

If you still have issues, contact us through our support e-mail.

Software Questions 

How can I read the raw data from an AEDAT4 file?

An .aedat4 file contains data under the AEDAT4 format. The data is compressed, it is therefore not directly human-readable. However, there are multiple solutions to extract the raw data from an AEDAT4 file:

Using dv-processing in C++ or Python.
Convert the event data to .csv using the corresponding DV module