| Planetary Photo Techniques (page 6, Updated 3/24/08) This page continues with the Skynyx Camera & Lucam Recorder |
I'd like to add a note that may be useful if you wish to compare the FOV available between two
cameras. This is motivated by a discussion on the Lumenera forum, where an individual wanted
to compare characteristics of the Skynyx 2-0 to the 2-1 camera. Specifically FOV he'd get, frame
rates, and dynamic range. This write-up looks at the FOV comparison.
We will assume that you have one scope that you would like to put the selected camera on,
hence, there will be a common focal length, say 1000mm used throughout this comparison.
Since we are using the same focal length for both cameras, we really don't even need to know
this, if we are only interested in a ratio of the different camera FOVs. As a first order
comparison of the two cameras, just look at the sensor chip size of the two cameras under
consideration, the physically larger one will most likely have the larger FOV, regardless of how
many pixels either of them has. How did I arrive at this?
The image scale, or FOV of a single pixel, in arcsec/pixel is approximately (assumes square
pixels):
Image scale, or FOVpixel (arcsec) = 200 X Pixel size (microns) / focal length (mm)
This is found by converting the actual FOV of one pixel, given in radians by the formula:
FOVpixel (radian) = 2 X Arctan( pixel size (mm) / 2 X focal length (mm) )
into FOVpixel (arcsec) = 206,265 X FOVpixel (radian)
and using the small angle approximation for the tan, or Arctan of an angle.
The total horizontal FOV is just the number of horizontal pixels times the image scale of the
pixel. The total vertical FOV is the number of vertical pixels times the image scale of the pixel.
The image scale given above is in arcsec/pixel, if you want the FOV in arcmin, just divide the
total FOV calculated above by 60. If you want it in degrees, just divide the total FOV calculated
above by 3600.
As an example, say we have two different cameras that we are considering. The sensor in the
first camera has 500X500 pixels, each 10u. The sensor in the second camera has 1200X1000
pixels, each 5u. Lets also assume, for just a moment, that all pixels on both sensors are actively
used for imaging (this is never the case on real CCDs). This means that the first sensor will be
about 5000u X 5000u, or a 5mm square (it has a 7mm diagonal). The second sensor will be 6000u
X 5000u, or 6mm X 5mm (it has a 7.8mm diagonal). In the case were we assume that all pixels are
actively imaging, we could estimate that the FOV of the second sensor will be slightly larger
than the first, but only by ~11%.
If we had performed the tedious arithmetic, we would find that the actual FOV of the first
camera's sensor is arrived at as:
200 X 10 / 1000 = 2"/pix, 500 X 2 = 1000" square, 1414" diagonal
The actual FOV of the second camera is given by:
200 X 5 / 1000 = 1"/pix, 1200 X 1 = 1200" horizontal, 1000 X 1 = 1000" vertical, 1562" diagonal
The calculated ratio of the two FOVs is 1.10, in other words, the second camera has a 10%
bigger FOV as compared to the first, when used on the same scope, at the same focal length.
This is very close to the much easier chip size estimate, and no arithmetic was required for that
first order estimate.
I assumed that all pixels are actively imaging in the above argument. This is never the case. A
few rows, and/or columns are used for black level calibration. The manufacturer may also use
some of the pixel columns/rows for other purposes. Let's look at the case were both camera's
sensors are assumed to have some rows used for purposes other than active imaging. If both
cameras considered above gave up 20 of its rows, so that the first camera's sensor active
imaging region is 500X480, and the second camera's active imaging region is 1200X980, then we
would find that the chip size estimate would err on the side of the camera with the larger pixel,
ie the first camera. We would say that the first camera had a FOV 11% smaller when in reality
may be a little more than this. This can be seen as:
1st camera: 500 X 2 = 1000" horizontal, 480 X 2 = 960" vertical, 1386" diagonal
2nd camera: 1200 X 1 = 1200" horizontal, 980 X 1 = 980" vertical, 1549" diagonal
Now the second camera will have a calculated FOV that is a full 12% larger. As the number of
rows used for purposes other than active imaging increases, the camera with the larger pixel
will experience a greater decrease in FOV relative to the FOV of the camera with the smaller
pixels. Even still, for this case the estimate based on chip size was still good.
In order to drag this out even further, and finally get to the actual comparison of the two
Lumenera cameras, let's look at the Sony ICX424 (Skynyx 2-0) and ICX205 (Skynyx 2-1)
datasheets. The 424 sheet shows that this type 1/3 chip is 5.79mm X 4.89mm, with 692 X 504
pixels, each 7.4um. The Skynyx 2-0 uses a 640 X 480 region of the pixels for active imaging. The
205 datasheet shows that this type 1/2 chip is 7.6mm X 6.2mm, with 1434 X 1050 pixels, each
4.65um. The Skynyx 2-1 uses a 1392 X 1040 region for active imaging.
How do they compare? By chip size, the 424 has a 7.6mm diagonal, the 205 has a 9.8mm
diagonal, so the first order FOV estimate would give the 2-1 ~28% more FOV. No arithmetic
needed, just look at the datasheet (but be sure that the datasheet you use has the correct
numbers). How about the actual FOV available, per the active imaging area. Both of these chips
give up 40 or 50 rows, and 10 or 20 columns for purposes other than active imaging, but the
smaller 424 gives up more, so it's actual imaging FOV will suffer more than the 205. The chip
size estimate will be off slightly, penalizing the 205 in the comparison against the 424. In fact,
the 424 uses only 88% of its pixelized area for active imaging in the 2-0, while the 205 uses 96%
of its pixelized area for active imaging in the 2-1 (the icx274, used in the 2-2 uses 95% of its
pixelized area for active imaging). For our hypothetical 1000mm scope, we have for the 424:
200 X 7.4 / 1000 = 1.48"/pix, 640 X 1.48 = 947" horizontal, 480 X 1.48 = 710" vertical, 1184" diagonal
For the larger 205 we have:
200 X 4.65 / 1000 = 0.93"/pix, 1392 X 0.93 = 1295" horizontal, 1040 X 0.93 = 967" vertical, 1616"
diagonal
From which we see that the true imaging FOV, using our 1000mm scope, for the 205 is just over
1/3 larger than that with the 424. The ratio of FOV for the two sensors, the basis of the
comparison, will remain constant whether we use a 1000mm, or a 10000mm focal length. The 205
will have 1/3 bigger FOV. By the way, the Skynyx 2-2, using a larger type 1/1.8 sensor provides a
50% larger FOV than the 2-0.
We know that the Skynyx 2-1 is more expensive than the 2-0, does the extra FOV add any real
capability? For planetary work, the FOV of the 2-0 adequately handles planetary bodies, even
Jupiter at 10-15m focal length. In order to nyquist sample blue light, we want a focal ratio of
about f/40 for the 2-0, f/25 for the 2-1. This means that in real use, both will be set for equal
image scale, in the case of a 10" scope, it will be about 0.15"/pix. We will have about 270 pixels
across the body of Jupiter, regardless of whether we use the large 2-0 pixels, or smaller 2-1
pixels. Even if we are using a 14" scope, we would find that there would be 400 pixels across
the planetary body. Either sensor, and in particular the smaller 2-0, still accommodates the
largest planetary body when used on most amateur scopes. The extra FOV of the 2-1, and even
more so the larger 2-2, may allow for a moon or two to be included, but otherwise isn't worth
much. Lunar and solar photographers, of which I am not, may find the extra FOV to be a real
help, but when one considers the Qe, full well, read noise, the resulting dynamic range,
exposure times, and frame rates, the better option is still the 2-0.
cameras. This is motivated by a discussion on the Lumenera forum, where an individual wanted
to compare characteristics of the Skynyx 2-0 to the 2-1 camera. Specifically FOV he'd get, frame
rates, and dynamic range. This write-up looks at the FOV comparison.
We will assume that you have one scope that you would like to put the selected camera on,
hence, there will be a common focal length, say 1000mm used throughout this comparison.
Since we are using the same focal length for both cameras, we really don't even need to know
this, if we are only interested in a ratio of the different camera FOVs. As a first order
comparison of the two cameras, just look at the sensor chip size of the two cameras under
consideration, the physically larger one will most likely have the larger FOV, regardless of how
many pixels either of them has. How did I arrive at this?
The image scale, or FOV of a single pixel, in arcsec/pixel is approximately (assumes square
pixels):
Image scale, or FOVpixel (arcsec) = 200 X Pixel size (microns) / focal length (mm)
This is found by converting the actual FOV of one pixel, given in radians by the formula:
FOVpixel (radian) = 2 X Arctan( pixel size (mm) / 2 X focal length (mm) )
into FOVpixel (arcsec) = 206,265 X FOVpixel (radian)
and using the small angle approximation for the tan, or Arctan of an angle.
The total horizontal FOV is just the number of horizontal pixels times the image scale of the
pixel. The total vertical FOV is the number of vertical pixels times the image scale of the pixel.
The image scale given above is in arcsec/pixel, if you want the FOV in arcmin, just divide the
total FOV calculated above by 60. If you want it in degrees, just divide the total FOV calculated
above by 3600.
As an example, say we have two different cameras that we are considering. The sensor in the
first camera has 500X500 pixels, each 10u. The sensor in the second camera has 1200X1000
pixels, each 5u. Lets also assume, for just a moment, that all pixels on both sensors are actively
used for imaging (this is never the case on real CCDs). This means that the first sensor will be
about 5000u X 5000u, or a 5mm square (it has a 7mm diagonal). The second sensor will be 6000u
X 5000u, or 6mm X 5mm (it has a 7.8mm diagonal). In the case were we assume that all pixels are
actively imaging, we could estimate that the FOV of the second sensor will be slightly larger
than the first, but only by ~11%.
If we had performed the tedious arithmetic, we would find that the actual FOV of the first
camera's sensor is arrived at as:
200 X 10 / 1000 = 2"/pix, 500 X 2 = 1000" square, 1414" diagonal
The actual FOV of the second camera is given by:
200 X 5 / 1000 = 1"/pix, 1200 X 1 = 1200" horizontal, 1000 X 1 = 1000" vertical, 1562" diagonal
The calculated ratio of the two FOVs is 1.10, in other words, the second camera has a 10%
bigger FOV as compared to the first, when used on the same scope, at the same focal length.
This is very close to the much easier chip size estimate, and no arithmetic was required for that
first order estimate.
I assumed that all pixels are actively imaging in the above argument. This is never the case. A
few rows, and/or columns are used for black level calibration. The manufacturer may also use
some of the pixel columns/rows for other purposes. Let's look at the case were both camera's
sensors are assumed to have some rows used for purposes other than active imaging. If both
cameras considered above gave up 20 of its rows, so that the first camera's sensor active
imaging region is 500X480, and the second camera's active imaging region is 1200X980, then we
would find that the chip size estimate would err on the side of the camera with the larger pixel,
ie the first camera. We would say that the first camera had a FOV 11% smaller when in reality
may be a little more than this. This can be seen as:
1st camera: 500 X 2 = 1000" horizontal, 480 X 2 = 960" vertical, 1386" diagonal
2nd camera: 1200 X 1 = 1200" horizontal, 980 X 1 = 980" vertical, 1549" diagonal
Now the second camera will have a calculated FOV that is a full 12% larger. As the number of
rows used for purposes other than active imaging increases, the camera with the larger pixel
will experience a greater decrease in FOV relative to the FOV of the camera with the smaller
pixels. Even still, for this case the estimate based on chip size was still good.
In order to drag this out even further, and finally get to the actual comparison of the two
Lumenera cameras, let's look at the Sony ICX424 (Skynyx 2-0) and ICX205 (Skynyx 2-1)
datasheets. The 424 sheet shows that this type 1/3 chip is 5.79mm X 4.89mm, with 692 X 504
pixels, each 7.4um. The Skynyx 2-0 uses a 640 X 480 region of the pixels for active imaging. The
205 datasheet shows that this type 1/2 chip is 7.6mm X 6.2mm, with 1434 X 1050 pixels, each
4.65um. The Skynyx 2-1 uses a 1392 X 1040 region for active imaging.
How do they compare? By chip size, the 424 has a 7.6mm diagonal, the 205 has a 9.8mm
diagonal, so the first order FOV estimate would give the 2-1 ~28% more FOV. No arithmetic
needed, just look at the datasheet (but be sure that the datasheet you use has the correct
numbers). How about the actual FOV available, per the active imaging area. Both of these chips
give up 40 or 50 rows, and 10 or 20 columns for purposes other than active imaging, but the
smaller 424 gives up more, so it's actual imaging FOV will suffer more than the 205. The chip
size estimate will be off slightly, penalizing the 205 in the comparison against the 424. In fact,
the 424 uses only 88% of its pixelized area for active imaging in the 2-0, while the 205 uses 96%
of its pixelized area for active imaging in the 2-1 (the icx274, used in the 2-2 uses 95% of its
pixelized area for active imaging). For our hypothetical 1000mm scope, we have for the 424:
200 X 7.4 / 1000 = 1.48"/pix, 640 X 1.48 = 947" horizontal, 480 X 1.48 = 710" vertical, 1184" diagonal
For the larger 205 we have:
200 X 4.65 / 1000 = 0.93"/pix, 1392 X 0.93 = 1295" horizontal, 1040 X 0.93 = 967" vertical, 1616"
diagonal
From which we see that the true imaging FOV, using our 1000mm scope, for the 205 is just over
1/3 larger than that with the 424. The ratio of FOV for the two sensors, the basis of the
comparison, will remain constant whether we use a 1000mm, or a 10000mm focal length. The 205
will have 1/3 bigger FOV. By the way, the Skynyx 2-2, using a larger type 1/1.8 sensor provides a
50% larger FOV than the 2-0.
We know that the Skynyx 2-1 is more expensive than the 2-0, does the extra FOV add any real
capability? For planetary work, the FOV of the 2-0 adequately handles planetary bodies, even
Jupiter at 10-15m focal length. In order to nyquist sample blue light, we want a focal ratio of
about f/40 for the 2-0, f/25 for the 2-1. This means that in real use, both will be set for equal
image scale, in the case of a 10" scope, it will be about 0.15"/pix. We will have about 270 pixels
across the body of Jupiter, regardless of whether we use the large 2-0 pixels, or smaller 2-1
pixels. Even if we are using a 14" scope, we would find that there would be 400 pixels across
the planetary body. Either sensor, and in particular the smaller 2-0, still accommodates the
largest planetary body when used on most amateur scopes. The extra FOV of the 2-1, and even
more so the larger 2-2, may allow for a moon or two to be included, but otherwise isn't worth
much. Lunar and solar photographers, of which I am not, may find the extra FOV to be a real
help, but when one considers the Qe, full well, read noise, the resulting dynamic range,
exposure times, and frame rates, the better option is still the 2-0.
