http://www.clarkvision.com/articles/evaluation-canon-1div
R. N. Clark Email contact (is encoded to prevent spam):
has the following form: username@qwest.net where
username is rnclark
This page shows an analysis of noise, dynamic range, and full well capacity of a Canon 1D Mark IV camera.
Procedures for performing this analysis are described in: Procedures for Evaluating Digital Camera Noise, Dynamic Range, and Full Well Capacities; Canon 1D Mark II Analysis
The lowest possible noise from a system detecting light is the noise due to Poisson statistics from the random rate of the arrival of photons. This is called photon statistics, or photon noise. Noise from the electronics will add to the photon noise. Noise in Canon 1D Mark IV images is limited by photon statistics at high signal levels and by electronic noise from reading the sensor (called readout noise) and noise from the downstream electronics at very low signal levels. In the case of high signal levels, a system that is photon statistics limited enables us to directly measure how many photons the sensor captures, and by increasing the exposure, we can determine how many photons are required to saturate the sensor. That is called the full well capacity, or simply, maximum signal capacity. With data on the lowest noise to the highest signal, we can then determine the dynamic range of the sensor.
The data and analysis results below show how the canon 1D Mark IV sensor performs. Table 1 shows the results and these results are shown on the graphs at Digital Sensor Performance summary for comparison with other cameras. The data show that the 1D4 is performing significantly above other sensors with similar sized pixels and even above some older cameras with larger pixels at the time of this writing. Canon in their white paper on the 1D4 ( PDF reference here) states they have made several improvements to the focal plane assembly, including higher transmission of the color Bayer filters over each pixel, better fill factor, and gapless micro lenses. These combine to deliver a high signal (collect more photons) in a given exposure, and the results below confirm Canon's claims.
The read noise, reaching a low 0f only 1.7 electrons at ISO 12,800 is the lowest I have yet measured and the lowest I have seen on any room-temperature sensor.
But even more impressive than the high signal and low read noise, is the far better control of fixed pattern noise. Figures 1 and 2 show histogram and a highly stretched image of the read noise at ISO 12,800. It is very difficult to see any fixed pattern noise. The control of fixed pattern noise and the very low read noise enables the best high ISO/low light performance I have seen. It will be interesting to see some astrophotos with this camera.
Table 1
-------------------------------------------------
Maximum Measured
ISO Gain Read Noise signal Dynamic range
e/DN (electrons) (electrons) stops
100 4.20 22.2 55600 11.3
200 2.10 11.5 27800 11.2
400 1.05 6.6 13900 11.0
800 0.52 4.0 6900 10.8
1600 0.26 2.9 3400 10.2
3200 0.13 2.3 1700 9.5
6400 0.066 1.9 860 8.8
12800 0.033 1.7 430 8.0
25600 0.016 1.7 e 215 7 e
51200 0.008 1.7 e 108 6 e
102400 0.004 1.7 e 54 5 e
e = estimated by extrapolation
Pixel pitch: 5.7 microns.
16.1 megapixels.
S/N on 18% gray card, ISO 100 = 100.
Sensor Full Well Capacity at lowest ISO: 55,600 electrons.
Sensor dynamic range = 55600/1.7 = 32,700 = 15.0 stops.
ISO at unity gain (scaled to 12 bit) = 1680 (14-bit unity gain = ISO 420).
Low Light sensitivity Factor: 988. (=12-bit unity gain / read noise)
Full Sensor Apparent Image Quality, FS-AIQ = 80.5.
Focal Length Limited Apparent Image Quality, ISO 1600, Constant output Size, FLL-AIQ1600 = 60
Values in the above table are described at Digital Sensor Performance summary.
Table 2: ISO 100 Sensor Data and Analysis
Offset= 2047
Model gain = 4.2 e/DN
Model read noise = 22.2 electrons
Observed signal - offset
--------------------------
min max mean 2-img std noise S/N signal ISO relative S/N S/N
file (DN) (DN) (DN) (DN) (DN) (electrons) exposure model obs/model
_73C5280 13236.00 13237.00 13236.30 0.70 sensor saturated
_73C5281 13236.00 13237.00 13236.42 0.70 sensor saturated
_73C5282 9838.00 10783.00 10255.89 70.23 49.66 206.51 42645.9 100 0.650000 206.4 1.00
_73C5283 9813.00 10676.00 10256.78 70.23 49.66 206.53 42653.2 100 0.650000 206.4 1.00
_73C5284 6164.00 6786.00 6479.05 57.55 40.69 159.21 25348.6 100 0.384616 163.5 0.97
_73C5285 6158.00 6824.00 6531.78 57.55 40.69 160.51 25762.9 100 0.384616 164.2 0.97
_73C5286 3776.00 4192.00 3979.53 45.36 32.08 124.06 15390.7 100 0.250000 127.4 0.97
_73C5287 3794.00 4256.00 4031.59 45.36 32.08 125.68 15795.9 100 0.250000 128.3 0.97
_73C5288 2327.00 2630.00 2467.75 35.54 25.13 98.20 9643.2 100 0.161290 99.5 0.99
_73C5289 2281.00 2603.00 2434.49 35.54 25.13 96.88 9384.9 100 0.161290 98.8 0.99
_73C5290 1424.00 1648.00 1538.85 28.08 19.85 77.51 6008.0 100 0.096154 77.5 1.00
_73C5291 1432.00 1652.00 1536.87 28.08 19.85 77.41 5992.6 100 0.096154 77.4 1.00
_73C5292 869.00 1028.00 951.19 22.34 15.80 60.22 3626.3 100 0.062500 59.6 1.01
_73C5293 872.00 1032.00 953.63 22.34 15.80 60.37 3644.9 100 0.062500 59.7 1.01
_73C5294 537.00 662.00 598.09 18.13 12.82 46.65 2176.5 100 0.040651 45.8 1.02
_73C5295 531.00 661.00 594.71 18.13 12.82 46.39 2152.0 100 0.040651 45.7 1.02
_73C5296 327.00 425.00 375.24 15.02 10.62 35.34 1248.7 100 0.024154 34.6 1.02
_73C5297 324.00 418.00 374.13 15.02 10.62 35.23 1241.4 100 0.024154 34.6 1.02
_73C5298 195.00 275.00 233.72 12.71 8.98 26.01 676.8 100 0.015625 25.6 1.02
_73C5299 190.00 275.00 233.10 12.71 8.98 25.95 673.2 100 0.015625 25.5 1.02
_73C5300 112.00 183.00 149.07 11.01 7.78 19.15 366.9 100 0.010122 18.7 1.02
_73C5301 117.00 191.00 149.15 11.01 7.78 19.17 367.3 100 0.010122 18.7 1.02
_73C5302 68.00 124.00 94.63 9.86 6.98 13.57 184.0 100 0.006024 13.3 1.02
_73C5303 65.00 123.00 93.30 9.86 6.98 13.38 178.9 100 0.006024 13.2 1.03
_73C5304 32.00 84.00 57.75 8.99 6.36 9.08 82.4 100 0.003906 8.9 1.02
_73C5305 33.00 86.00 58.96 8.99 6.36 9.27 85.9 100 0.003906 9.1 1.00
_73C5306 15.00 62.00 37.23 8.48 6.00 6.21 38.5 100 0.002533 6.1 1.01
_73C5307 14.00 64.00 38.70 8.48 6.00 6.45 41.6 100 0.002533 6.3 0.98
_73C5308 2.00 48.00 24.12 8.11 5.74 4.20 17.7 100 0.001506 4.2 1.01
_73C5309 2.00 49.00 25.41 8.11 5.74 4.43 19.6 100 0.001506 4.4 0.96
_73C5310 -13.00 38.00 13.91 7.86 5.56 2.50 6.3 100 0.000976 2.5 1.01
_73C5311 -7.00 37.00 15.88 7.86 5.56 2.86 8.2 100 0.000976 2.8 0.89
_73C5312 -13.00 35.00 10.27 7.69 5.44 1.89 3.6 100 0.000633 1.9 1.01
_73C5313 -12.00 34.00 9.93 7.69 5.44 1.83 3.3 100 0.000633 1.8 1.05
_73C5314 -17.00 28.00 5.67 7.64 5.40 1.05 1.1 100 0.000376 1.0 1.00
_73C5315 -18.00 28.00 6.00 7.64 5.40 1.11 1.2 100 0.000376 1.1 0.95
_73C5316 -20.00 27.00 4.58 7.60 5.37 0.85 0.7 100 0.000244 0.8 1.00
_73C5317 -18.00 25.00 3.36 7.60 5.37 0.63 0.4 100 0.000244 0.6 1.36
_73C5318 -19.00 26.00 2.60 7.55 5.34 0.49 0.2 100 0.000158 0.5 1.00
_73C5319 -21.00 24.00 2.66 7.55 5.34 0.50 0.2 100 0.000158 0.5 0.98
_73C5320 -21.00 24.00 1.36 7.55 5.34 0.26 0.1 100 0.000094 0.3 1.00
_73C5321 -21.00 23.00 1.62 7.55 5.34 0.30 0.1 100 0.000094 0.3 0.84
_73C5322 -25.00 24.00 1.27 7.52 5.32 0.24 0.1 100 0.000061 0.2 1.00
_73C5323 -22.00 22.00 1.95 7.52 5.32 0.37 0.1 100 0.000061 0.4 0.65
Table 3: Read Noise Data and Analysis
Apparent
Read Noise Gain
file min max mean 2-img std noise (electrons) (e/DN) ISO
_73C5408 2025.00 2068.00 2047.24 7.48 5.29 22.215 4.200 100
_73C5409 2025.00 2068.00 2047.19 7.48 5.29 22.215 4.200 100
_73C5412 2022.00 2074.00 2046.93 7.72 5.46 11.466 2.100 200
_73C5413 2022.00 2068.00 2047.01 7.72 5.46 11.466 2.100 200
_73C5416 2009.00 2087.00 2046.85 8.91 6.30 6.615 1.050 400
_73C5417 2011.00 2103.00 2046.88 8.91 6.30 6.615 1.050 400
_73C5422 1985.00 2127.00 2047.00 10.90 7.71 4.045 0.525 800
_73C5423 1978.00 2106.00 2047.00 10.90 7.71 4.045 0.525 800
_73C5428 1912.00 2197.00 2047.15 15.81 11.18 2.906 0.260 1600
_73C5429 1942.00 2208.00 2047.06 15.81 11.18 2.906 0.260 1600
_73C5434 1795.00 2330.00 2045.84 25.20 17.82 2.334 0.131 3200
_73C5435 1796.00 2320.00 2046.29 25.20 17.82 2.334 0.131 3200
_73C5440 1619.00 2553.00 2046.14 40.92 28.94 1.910 0.066 6400
_73C5441 1520.00 2546.00 2046.15 40.92 28.94 1.910 0.066 6400
_73C5446 1529.00 2583.00 2043.83 74.45 52.64 1.737 0.033 12800
_73C5447 1487.00 2750.00 2043.78 74.45 52.64 1.737 0.033 12800
Data acquired by Peter A. Hawrylyshyn, M.D. February, 2010
Analysis by R. N. Clark February 13. 2010
Figure 1. Histogram showing the noise distribution if the read noise in a Canon 1D mark IV
camera. Histogram limits are 0 to 1024 out of 0 to 16383 original camera
levels (DN). The image of this stretched histogram, indicated by the 3 tick
marks below the histogram, whose limits are
432 to 644 original camera DN, are shown in Figure 2. The data range is 0 to 33.8
electrons from the left to right limits of the graph. The scale below the
three marks shows the range in electrons.
The histogram vertical scale is logarithmic in A, and linear in B (Photoshop).
The electron scale only aplies the histogram A; the horizontal scale in B is not
linear.
Figure 2. Stretched read noise from the Canon 1D Mark IV at ISO 12,800.
Note that pattern noise is barely perceptible. The histogram of this image
is in Figure X. Minimum in this image = 432 and max = 644 in original camera
DN, out of 0 to 16383 (14-bits). That is a range of 7 electrons from black to
white in the image. This image is direct from a single raw file, with no
demosaicking (from dcraw). Note the black and white pixels are due to clipping
the levels, and are not hot or dead pixels.
Conclusions
The Canon 1D Mark IV sensor sets new performance standards, including higher sensitivity per pixel (a combination of filter transmittance times effective fill factor times quantum efficiency), lower read noise, and lower fixed pattern noise. Hopefully the new technology that went into this sensor will makes its way into other cameras as well. That will enable a new era in performance, especially for high ISO / low light photography.
Dynamic range is still limited to a little over 11-stops, apparently by downstream electronics that must process the data extremely fast (at 10 frames per second; that is over 160 megapixels/second). I would like to see a camera option that used a slow 16-bit A/D converter and low noise amplifiers to deliver data with the full capability of the sensor, which is 15 stops. Even if it took 10 seconds to read out the sensor (longer readout times are used in scientific applications), there are situations where high dynamic range imaging would benefit.
References
1) CCD Gain. http://spiff.rit.edu/classes/phys559/lectures/gain/gain.html
4) http://www.photomet.com/library_enc_fwcapacity.shtml
Notes:
DN is "Data Number." That is the number in the file for each pixel. I'm quoting the luminance level (although red, green and blue are almost the same in the cases I cited).
16-bit signed integer: -32768 to +32767
16-bit unsigned integer: 0 to 65535
Photoshop uses signed integers, but the 16-bit tiff is unsigned integer (correctly read by ImagesPlus).
The sensor analysis was done with custom, in-house written software. Raw data were extracted from the camera raw files using DCRAW. Custom software read that data and all processing was done in 32-bit floating point.
Home Page: ClarkVision.com
Back to: Digital Imaging Information index on this site: http://www.clarkvision.com/articles
First published February 13, 2010.
Last updated February 15, 2010.