What is the lowest resolution of the camera?

Hi,
I got the H7 - What is the lowest resolution possible of the camera?

I’m thinking of buying the Global Shutter Camera - What is the lowest resolution possible of this camera?

Reuben

The lowest resolution is QQQQVGA (40x30), you can set a smaller window/crop (with set_windowing()) to get a lower resolution (the minimum is 8x8).

Hi,
I got my Global Shutter sensor. When I use QQVGA I get ~215fps (example high_fps.py).
When I try smaller resolutions (QQQVGA or QQQQVGA) the image is being cropped instead of being downsized so I’m losing part of the image and the fps changes a bit QQQVGA & QQQQVGA =~221FPS

  1. Is there a way to reduce resolution without losing the field of view?
  2. Why the FPS doesn’t go up on lower resolution?

Reuben

  1. No, the sensor only supports binning at 4x4 from it’s max res. So, after 160x120 any speed ups are from cropping.

  2. Put the camera in triggered mode to increase the fps more beyond 221 fps. Also, turn the frame buffer off using the disable frame buffer button.

At 200+ fps the processor only had milliseconds per image… So, there’s not a lot you’re able to do.

I’m using the “triggered_mode.py” And I disabled the frame buffer. I get the following results:

VGA ~= 51FPS
QVGA ~= 77FPS
QQVGA ~=113FPS
QQQVGA ~= 101FPS :open_mouth:
QQQQVGA ~=104FPS

on the “high_fps.py” (frame buffer is disabled) I get:
VGA ~=68FPS
QVGA ~=125FPS
QQVGA ~=215FPS
QQQVGA ~=220FPS
QQQQVGA = ~221FPS

No more then 221FPS?

Sorry, you need to combine what both scripts do the camera. high_fps.py gets high fps because it allows the exposure to be really short. Triggered mode just forces the camera to expose in sync with the OpenMV Cam.

I tested it and it works well!!

Any plans in the future to add resize_img command to the library or maybe just a reduce_img zise command without changing the aspect-ratio?

  • I wish i knew c so I could add it by myself
    e.g 160x120:
    reduce(2) 80X60
    reduce(4) 40X30

I assume adding a resize may cause problem in cases up resizing for bigger resolution so maybe you can only add reduce image

Thank you very much for your answers and Openmv is a great product!!

Reuben

Use the mean_pool or midpoint_pool methods. They are designed to downscale.

There’s also literally a scale() method… Please read the API. image — machine vision — MicroPython 1.15 documentation

Use the mean_pool or midpoint_pool methods however. Scale is nearest neighbor.

http://docs.openmv.io/library/omv.image.html?highlight=scale#image.image.mean_pool
http://docs.openmv.io/library/omv.image.html?highlight=scale#image.image.midpoint_pool

I just checked it and it is working!
Thank you