This is the official implementation of Physical-based Event Camera Simulator（ PECS）.This is a trial version used to validate the algorithm. Since PBRT has actually been deprecated by us due to its slow speed, limited functionality, and the frequent, headache-inducing complex compilation issues it causes, this version only has UE version. If you need more information, please refer to the paper with the same title.
- The sensor/src part is partly from ICNS because our algorithm is unrelated to the electronic circuitry except for the photoelectric effect.Thus, we use ICNS to achieve electronic circuitry simulation. In addtition, some functions need metavision to run because our camera is EVK4. If you don't want to use metavision, you can comment them out.
-please note that you need to generate photocurrent based on your own camera's quantum efficency.Sensor/utils is an example of quantum efficency implementation.
we edit Airsim to generate event data real-time in UE. The usage instructions are as follows:
-
Install AirSim
-
use eventcamera_sim to replace the folder with the same name in Airsim
-
generate library for eventcamera_sim
python sensor/utils/setup_display.py build
and move the output library to the eventcamera_sim folder. Then you need to generate a dsi lib and move it to the eventcamera_sim folder, which you can refer to ICNS. Finaly, you can use PECS in Airsim.
PECS's algorithm contribution is firstly proposed the effect of lens simulation and quantum efficency. This means you can combine arbitory sensor with rendering engine you like, and using above algorithms to get a better simulaton result in a specific camera.
Details in the paper.(tested in windows 11)
- Part of our code refers to ICNS
- We use blender and UE(airsim) as tools
We may implement the algorithmic concepts of PECS into a larger project in the future, and there will be further updates at that time.