Just sharing some test video I've done recently, related to motion and pixel blur. This is all related to Computer Vision, specifically the vibration side of Vision hardware...
The majority of my testing is done on a simulated gimbal shaker. Posted in the my thread IMU Data Viewer, Yes with MatLab, describes a little bit of my efforts. Recently I had some time to go fly and perform some real world testing.
In this video I am targeting a test pattern used for focus imaging equipment. The Siemens Star is one of the common targets used and provides an ability to analyze pixel and motion blur. As well the fencing provides some ability to estimate speed of travel, in this case about 2.5m a second.
The video was converted to 24fps for the three variations...24, 30 and 60fps. The first sequences of videos is zoomed to 4X, giving a detailed perspective of the blur.
The Siemens Star is about 6.25" Diameter and the typical distance away is about 15'
Stock GP4B settings were: 2.7K PT/GPC, Auto WB, Sharp-Med, Spot-Off, 400
Testing on this overcast day was ideal, less light means more exposure time...more blur. You can pause the video to see the target at the frame level. Below is also a frame grab image so you can zoom in further to see how blur affects the reveals on the target. The center of the target reveals the blur effect by losing the target's details, I estimate about a 2"D...so not real bad consider motion and all. As well the pixel motion blur being more prominent laterally on the target, about 2 pixels.
In general, the Solo gimbal performs better laterally when moving left to right when flying fast, related to lack of prop wash on the gimbal. In the video you can also see the affects of rolling shutter. The fence's pickets are slanted slightly to the right, as centered frame.
Again this is just stuff I piddle with in my spare time. I know it's a strange endeavor, but I find the subject of computer vision and vibration damping interesting. Constructive comments are always welcome.
The majority of my testing is done on a simulated gimbal shaker. Posted in the my thread IMU Data Viewer, Yes with MatLab, describes a little bit of my efforts. Recently I had some time to go fly and perform some real world testing.
In this video I am targeting a test pattern used for focus imaging equipment. The Siemens Star is one of the common targets used and provides an ability to analyze pixel and motion blur. As well the fencing provides some ability to estimate speed of travel, in this case about 2.5m a second.
The video was converted to 24fps for the three variations...24, 30 and 60fps. The first sequences of videos is zoomed to 4X, giving a detailed perspective of the blur.
The Siemens Star is about 6.25" Diameter and the typical distance away is about 15'
Stock GP4B settings were: 2.7K PT/GPC, Auto WB, Sharp-Med, Spot-Off, 400
Testing on this overcast day was ideal, less light means more exposure time...more blur. You can pause the video to see the target at the frame level. Below is also a frame grab image so you can zoom in further to see how blur affects the reveals on the target. The center of the target reveals the blur effect by losing the target's details, I estimate about a 2"D...so not real bad consider motion and all. As well the pixel motion blur being more prominent laterally on the target, about 2 pixels.
In general, the Solo gimbal performs better laterally when moving left to right when flying fast, related to lack of prop wash on the gimbal. In the video you can also see the affects of rolling shutter. The fence's pickets are slanted slightly to the right, as centered frame.
Again this is just stuff I piddle with in my spare time. I know it's a strange endeavor, but I find the subject of computer vision and vibration damping interesting. Constructive comments are always welcome.
