VideoCapture image size

imrasp commented 6 years ago

[x] increase size of input
[x] increase quality of input

Be aware that at 1280x720 you can expect 15FPS or less this is a limitation of USB 2.0 bandwidth and not the camera itself. Since USB 3.0 webcams are new to the market you can expect to pay a premium REF : http://webcache.googleusercontent.com/search?q=cache:c6z0_izJCcAJ:www.emgu.com/forum/viewtopic.php%3Ft%3D3747+&cd=6&hl=en&ct=clnk&gl=th

imrasp commented 6 years ago

https://arstechnica.com/civis/viewtopic.php?f=6&t=978436

RGB32: The same quality as RGB24 with the bonus of 33% more useless data.
RGB24: Although twice the horizontal color resolution as YUY2, it tends to get produced by upsampling and thus with most capture devices you won't get any additional quality by using it, just 50% more unnecessary data. However, it is by far the most compatible format when it comes to codecs and encoders.
RGB16: Gives you the same data rate as YUY2 (1/3rd less than RGB24), but with significant banding. Good for video displays, bad for video storage.
YUY2: In most cases, the best. It's usually the closest you can get to data right out of the video decoder and it offers significant speed and/or size improvements over RGB24 (it's one-third smaller). This is the preferred format of Avisynth and Huffyuv. UYVY is a byte ordering change from YUY2 and either will do if both capture device and codec accept them.
YVU12 (YV12/I420): 25% less data than YUY2 in exchange for slightly more color bleeding vertically; this is the color space of MPEG-1. There is no difference in luminance. Not very well supported, unfortunately, and I wouldn't use it even if you were targetting MPEG-1.
YVU9: 25% less data than YVU12 and 43% less data than YUY2, but significant color bleeding -- this is the color space of Indeo. Not very well supported and I wouldn't recommend it.

Huffyuv is a lossless codec by Ben Rudiak-Gould that gives around 2:1 compression for decently clean source. Although it is slow at decompression, it has very fast compression and is competitive with PICVideo MJPEG for high-quality capture. It gets significantly better compression with YUY2 or UYVY input than with RGB. This is my method of choice, but it requires a LOT of disk space if you have a noisy feed -- something like 10GB per half hour for me, at 640x480x29.97fps with CD-quality audio.

imrasp commented 6 years ago

Global shutter cam IQinVision

v4l2-ctl -d /dev/video1 --list-formats-ext

ioctl: VIDIOC_ENUM_FMT
    Index       : 0
    Type        : Video Capture
    Pixel Format: 'YUYV'
    Name        : YUYV 4:2:2
        Size: Discrete 1280x720
            Interval: Discrete 0.017s (60.000 fps)
        Size: Discrete 1280x960
            Interval: Discrete 0.019s (54.000 fps)
        Size: Discrete 640x480
            Interval: Discrete 0.019s (54.000 fps)

    Index       : 1
    Type        : Video Capture
    Pixel Format: 'GREY'
    Name        : 8-bit Greyscale
        Size: Discrete 1280x720
            Interval: Discrete 0.017s (60.000 fps)
        Size: Discrete 1280x960
            Interval: Discrete 0.019s (54.000 fps)
        Size: Discrete 640x480
            Interval: Discrete 0.019s (54.000 fps)

imrasp commented 6 years ago

mobius - ros calib result

('D = ', [-0.26195662857200297, 0.0441263311300231, 0.0015978692595019826, -0.001618442666510325, 0.0])
('K = ', [309.2038080910573, 0.0, 317.47738187192186, 0.0, 348.4980876888013, 238.2653550509399, 0.0, 0.0, 1.0])
('R = ', [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0])
('P = ', [203.9306182861328, 0.0, 318.13921615756954, 0.0, 0.0, 294.8394775390625, 239.066151332554, 0.0, 0.0, 0.0, 1.0, 0.0])
None
# oST version 5.0 parameters

[image]

width
640

height
480

[narrow_stereo]

camera matrix
309.203808 0.000000 317.477382
0.000000 348.498088 238.265355
0.000000 0.000000 1.000000

distortion
-0.261957 0.044126 0.001598 -0.001618 0.000000

rectification
1.000000 0.000000 0.000000
0.000000 1.000000 0.000000
0.000000 0.000000 1.000000

projection
203.930618 0.000000 318.139216 0.000000
0.000000 294.839478 239.066151 0.000000
0.000000 0.000000 1.000000 0.000000

imrasp commented 6 years ago

IQinVision camera ros calib

('D = ', [-0.3708541546088663, 0.16722282677278488, -0.0012121279795248379, 0.0006352124631734436, 0.0])
('K = ', [589.0841496480523, 0.0, 283.7350200893131, 0.0, 588.5584428987099, 255.7768915817189, 0.0, 0.0, 1.0])
('R = ', [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0])
('P = ', [518.1420288085938, 0.0, 274.9534675317591, 0.0, 0.0, 549.4267578125, 257.5508008793695, 0.0, 0.0, 0.0, 1.0, 0.0])
None
# oST version 5.0 parameters

[image]

width
640

height
480

[narrow_stereo]

camera matrix
589.084150 0.000000 283.735020
0.000000 588.558443 255.776892
0.000000 0.000000 1.000000

distortion
-0.370854 0.167223 -0.001212 0.000635 0.000000

rectification
1.000000 0.000000 0.000000
0.000000 1.000000 0.000000
0.000000 0.000000 1.000000

projection
518.142029 0.000000 274.953468 0.000000
0.000000 549.426758 257.550801 0.000000
0.000000 0.000000 1.000000 0.000000

('Wrote calibration data to', '/tmp/calibrationdata.tar.gz')

imrasp commented 6 years ago

gun's logitech cam - ros calib

('D = ', [-0.0063537315596489865, 0.09196839081625377, 0.001295396987758146, 0.0013147129093750876, 0.0])
('K = ', [813.6804878474821, 0.0, 331.0379749750803, 0.0, 808.9466037567273, 253.97783194048048, 0.0, 0.0, 1.0])
('R = ', [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0])
('P = ', [816.9920043945312, 0.0, 331.48460130085004, 0.0, 0.0, 812.4269409179688, 254.3105466265897, 0.0, 0.0, 0.0, 1.0, 0.0])
None
# oST version 5.0 parameters

[image]

width
640

height
480

[narrow_stereo]

camera matrix
813.680488 0.000000 331.037975
0.000000 808.946604 253.977832
0.000000 0.000000 1.000000

distortion
-0.006354 0.091968 0.001295 0.001315 0.000000

rectification
1.000000 0.000000 0.000000
0.000000 1.000000 0.000000
0.000000 0.000000 1.000000

projection
816.992004 0.000000 331.484601 0.000000
0.000000 812.426941 254.310547 0.000000
0.000000 0.000000 1.000000 0.000000

imrasp commented 6 years ago

roscore rosparam set cv_camera/device_id 1 rosrun cv_camera cv_camera_node

rosrun camera_calibration cameracalibrator.py --size 9x6 --square 0.023 image:=/cv_camera/imageraw camera:=/cv camera/

imrasp commented 6 years ago

mobuis - ros calibration

('D = ', [-0.2765147597541606, 0.052479268381039945, -0.0013199539860483027, 0.0004802657728054864, 0.0])
('K = ', [408.76256460526787, 0.0, 407.2447748597686, 0.0, 463.6651343352479, 252.18756274716878, 0.0, 0.0, 1.0])
('R = ', [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0])
('P = ', [269.6871337890625, 0.0, 415.58530354675895, 0.0, 0.0, 424.81304931640625, 253.95481774115615, 0.0, 0.0, 0.0, 1.0, 0.0])
None
# oST version 5.0 parameters

[image]

width
848

height
480

[narrow_stereo]

camera matrix
408.762565 0.000000 407.244775
0.000000 463.665134 252.187563
0.000000 0.000000 1.000000

distortion
-0.276515 0.052479 -0.001320 0.000480 0.000000

rectification
1.000000 0.000000 0.000000
0.000000 1.000000 0.000000
0.000000 0.000000 1.000000

projection
269.687134 0.000000 415.585304 0.000000
0.000000 424.813049 253.954818 0.000000
0.000000 0.000000 1.000000 0.000000

('Wrote calibration data to', '/tmp/calibrationdata.tar.gz')

2nd

('D = ', [-0.270115209535647, 0.04775925645447495, -0.0013526057156239871, -0.0015524817571016716, 0.0])
('K = ', [321.40359116573444, 0.0, 312.6202058997774, 0.0, 359.5424147556039, 250.7085802378792, 0.0, 0.0, 1.0])
('R = ', [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0])
('P = ', [211.55117797851562, 0.0, 315.4040554332423, 0.0, 0.0, 306.044677734375, 254.0685189615906, 0.0, 0.0, 0.0, 1.0, 0.0])
None
# oST version 5.0 parameters

[image]

width
640

height
480

[narrow_stereo]

camera matrix
321.403591 0.000000 312.620206
0.000000 359.542415 250.708580
0.000000 0.000000 1.000000

distortion
-0.270115 0.047759 -0.001353 -0.001552 0.000000

rectification
1.000000 0.000000 0.000000
0.000000 1.000000 0.000000
0.000000 0.000000 1.000000

projection
211.551178 0.000000 315.404055 0.000000
0.000000 306.044678 254.068519 0.000000
0.000000 0.000000 1.000000 0.000000

('Wrote calibration data to', '/tmp/calibrationdata.tar.gz')

imrasp commented 6 years ago

('D = ', [-0.26902156103134517, 0.047568364162286235, -0.0027215413820488896, -0.001491015127786872, 0.0])
('K = ', [329.7775161035554, 0.0, 316.77688357221393, 0.0, 370.7263015978578, 251.2873606498051, 0.0, 0.0, 1.0])
('R = ', [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0])
('P = ', [212.90399169921875, 0.0, 315.3086176447541, 0.0, 0.0, 320.19140625, 253.29800943564624, 0.0, 0.0, 0.0, 1.0, 0.0])
None
# oST version 5.0 parameters

[image]

width
640

height
480

[narrow_stereo]

camera matrix
329.777516 0.000000 316.776884
0.000000 370.726302 251.287361
0.000000 0.000000 1.000000

distortion
-0.269022 0.047568 -0.002722 -0.001491 0.000000

rectification
1.000000 0.000000 0.000000
0.000000 1.000000 0.000000
0.000000 0.000000 1.000000

projection
212.903992 0.000000 315.308618 0.000000
0.000000 320.191406 253.298009 0.000000
0.000000 0.000000 1.000000 0.000000

imrasp commented 6 years ago

33207238_1715499361859477_2677766999290937344_n 33397529_1715499505192796_1991845848647991296_n 33305595_1715499605192786_4266328290861514752_n 33222885_1715499688526111_2496414784061177856_n 33222898_1715499788526101_5935413592097554432_n 33232494_1715499861859427_3164010889800056832_n

imrasp / LearnVI_Drone

VideoCapture image size #18