Adam, from a theoretical perspective, I can tell you that the information on the page you linked, about getting more detail by oversampling, is completely false. Yes it looks convincing, because of the way the sampled raw data is presented. In reality, oversampled data contains no more information than data sampled at the Nyquist limit.
Given that the signal is sampled at or above the Nyquist frequency f, and the fact, by definition, that the original signal contains no frequency components greater than 1/2 the Nyquist frequency, then the same original continuous signal can be recovered irrespective of what sampling rate (>= f) is used, if the data is analysed correctly. Correct analysis is not simply connecting the dots like the webpage author did for the waveform (and the star image examples are even more misleading). It is a well defined rigorous mathematical process that guarantees the complete recovery of the original signal. And it applies equally in one, two and higher dimensions, so it works for audio, video, or any other sampled data you can think of.
So don't feel like you need to go nuts with the CCD resolution. Just sample at twice the resolving limit of the scope and that will be as good as you are going to get. The rest is all up to doing the right thing with data you collect.
davo's threorem:
there is a fine line-> lotsa beer and steak at the snake valley camp will mean i will understand what steve and adam are on about, but will have forgotten the next morning or steve will be too drunk and none of his words will be understood??
ok merlin, got up this morning and had a lovely cooled scope, but the hole in jetstream did not coincide with me getting up, so, here is a venus and jupiter at 1250/6 = 208x with a 640x480 screen. ie prime focus toucam without any barlow or powermates
also is a 1250 / 6 and then mulitplied by between 5 to 7.7 for the tv 5x powermate ( i have no idea how far the ccd is away from the powermate . it is an unprocessed frame of the blue channel on a cropped 400x400 from the 5th feb
God I love this. Where else would you find debate like this?? I think these discussions give us an insight into minimum requirements and maximum outcome. If the seeing conditions allow then longer focal lengths will help, but there's no need to overly magnify the image where it is more difficult to focus and much fainter. If we can find the optimum settings then we can greatly assist each other in standardising the optical set ups at minimum cost and maximum ease of use.
Great stuff.
Hi Merlin66, I did want to ask you if you were happy with the direction your thread is taking - I don't want to hijack your thread!
janoskiss, thanks for your post - it has made me re-evaluate some of my earlier thinking and calculations - and chase more up to date information. Perhaps its a little too early to show poor old Nyquist the door!
I have to admit that this issue of what is the maximum useful focal length to use with a webcam is something that I've grappled with for a couple of months. I am aware of Nyquists theorem (which says you should sample a 2x the resolution limit) but I see many planetary images that I believe are taken at a higher sampling rate than this - not hard fact, yet, but I'm working on that.
One thing that has been problematic for me is defining the maximum resolution of an optical system - early on I think I was using 'older' definitions (Dawes limits, Rayleigh limits - not sure, have to go back and look). When using these as your definition of maximum resolution modern planetary images with webcams blow the 2x sampling away by a long way! Another definition of maximum resolution I have found is FWHM (Full Width at Half Maximum). I'm thinking that if you use FWHM as your definition of maximum resolution the high res planetary images we see may actually fit Nyquists theorem nicely. Again, I'd like to back that up with some hard facts.
Somewhere during the searches I did on double stars this was mentioned... BUT seeing conditions screw up a lot of the theory. I did see some interesting work on amateur speckle interferometery with doubles which allowed "recognition" of close doubles. How this could be applied to planetary disks is beyond me but.....
In another thread the use of filar micrometers is dead. The humble Webcam can do a far better job of measuring double stars maybe we should set up some close doubles for our experiments to see exactly what resolution we can achieve????
Would this be "more meaningful" than using planets???. Always a concern in the back of my mind that our image manipulation is adding detail that's not really there! How's that for a can of worms.
Always a concern in the back of my mind that our image manipulation is adding detail that's not really there!
This can be quite true, thanks to the deconvolution and sharpening algorithms that are used when stacking and processing webcam digital images.
Imagine a very close double, that in the raw data, appears to our eye not to be split. but all it would take is a slight gradient between the 2 stars to be sharpening , turning the greys to black and hence providing a visible "split" between the 2 stars that wasn't visible before.
Is that ok? I dunno, i guess it is.
Where we have to be careful is with planetary imaging, where we talk about "features" on a planets disc but are really artifacts that are introduced by the stacking and sharpening algorithm. This is very common on Saturn, with the whole "Encke minima/encke gap" argument, and also where the rings meet the planet, where colour from the rings can find themselves transferred onto the disc.
Always a concern in the back of my mind that our image manipulation is adding detail that's not really there! How's that for a can of worms.
yes, i saw some images on cloudy night that simply looked wrong of mars. damien peach is still producing consistant ones of mars even now.
what is the formula for the limit for a 10" scope. visual they say 60x your 10" ie 600.
in great seeing two mornings ago, i was at 500x and i believe with a top planetary ep, i could have pushed that 600x limit, but in theory this seems correct.
With the 5x powermate i can theoretically get between 1040 and 1600x, so what is the theoretical limit in mag??
Just re-entering some text that was deleted some how during upload
When we are talking about the maximum magnification for any given telescope the general rule is 40x to 50x per inch of apature. Therefore a the maximum magnification for a 10" scope will be between 400x to 500x. But in practice as you will know these high magnification are unobtainable, due to may factors such as weather etc.
In my experience when observing planets visual I offend use medium magnification (about 200 -250X), I end up with a smaller, sharper image with out the blurring effect cause by excessive magnification.
I have spent some years now doing planetary image and like may have been on a steep learning curve. What I have found is that there are may factor that effect the final produce e.g. the weather (something that we can not control), collimation, temperature, focal length, Image Scale, focus, frame rate, gain, processing etc.
For planetary imaging, Image scale is an important factor, since you will taking pictures of small targets (Mars, Jupiter, Saturn, etc.) you will need to magnify these objects in order to capture sufficient detail. For imaging, magnification is a function of focal length and the focal length will determines image scale.
So the first thing that must be considered is what amount of magnification should be used when imaging the planets? Ideally, an image scale of about 0.25 arcseconds/pixel is recommended. This should reveal the most detail possible under good seeing conditions without being magnified too much. The required focal length depends on the size of the pixels in your CCD camera. For a typical webcam with a pixel size of 5.6 microns you are looking at a focal length of 4600mm to achieve an image scale of 0.25"/pixel.
If you wish to work out the magnification of your imaging set up us this formula;
Focal length of telescope / 50 = Magnification.
Example 1 – Prime focus
A Canon EOS 350D plus an ED80 will work out to a magnification of 12x
600/50=12
Example 2 – Negative projection (Barlow)
A 5x Powermate plus 10” f10 SCT will work out to a magnification of 250x
2500x5=12500 12500/50=250
As mention above, an ideal image scale of 0.25 arcseconds/pixel is recommended, but in the real world of planetary imaging the image scale can be a lot smaller and as image scale is determined by the focal length, the focal length is going to be larger. Therefore the typical focal length when imaging with a webcam can range between f20 to f45. However, longer or shorter focal lengths can be used with excellent results depending primarily on the atmospheric conditions.
<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com[img] /><o:p>[font=/><o:p></o:p></font>
<o:p>For further reading I strongly recommend that you read Damain Peach article on "An Examination of optical issues on real Planetary Images" at http://www.damianpeach.com/simulation.htm</o:p></FONT></P>
Last edited by anthony2302749; 08-02-2006 at 09:24 AM.
Reason: Missing Text
Seeing conditions. can be, somewhat over come by using high frame rates. The average webcam is limited in this area due to its design (USB 1.1, compression etc). The current tread is to use a Camera, such as the Lumenera LU075, which can operate at high frame rate, are USB 2.2 or Firewire and can down load uncompress files with out loss of data. Bird would be able to give a better explanation on this topic. I use a Lumenera LU075C camera and can image at 37fps+ this mean that I can collect more frames in a minute than the average webcam and therefore will have a greater chance of collecting more frames free of a blurring effect cause by the weather.
In my last post I commented on the need to properly define the maximum possible resolution for an optical system. Only after that is defined can we calculate the proper focal length to provide 'critical sampling' ( 2x the maximum resolution as per Nyquist).
Attempts have been made to define maximum resolution based on the visual observation of double stars (Rayleigh limit and Dawes limit). These definitions have suited visual observers of double stars but are not suited to high resolution digital imaging of planets. In the "Handbook of Astronomical Image Processing" Richard Berry and James Burnell nominate the diameter of the Full Width Half Maximum (FWHM) as "a realistic measure of the smallest detail contained in an astronomical image". The diameter of the FWHM is the region within the bright core of the diffraction disk where the "light has fallen to half its central intensity".
I have adapted a formula given in Berry and Burnell to give the focal length for critical sampling given the aperture of an optical system, the size of the pixels in your webcam, and the wavelength of light you are interested in. I have attached a PDF file with the formula.
This, I think, is an answer to the question "what focal length should I use with my webcam to achieve critical sampling". Not quite the question that was asked, but a step closer!
Oversampling
If you image at focal lengths below those calculated by the formula attached you are undersampling. If you image at focal lengths above you are oversampling. The comment from HAIP regarding oversampling is that sampling at 3 to 5 times the maximum resolution "insures that none of the information present in the telescopic image is lost because the image is broken into discrete samples".
While Hitchhiker post is spot on when imaging a point source such as a star/deep space object it does not relate to planetary imaging.
Planetary imaging is an art in itself, Planets are extended objects, and the Dawes or Rayleigh criterion does not apply here as these limits refers to point sources of equal brightness on a black background. For example, a 254mm aperture telescope has a dawes limit of 0.45" arc seconds. The dawes limit is really of little use the Planetary observer, as it applies to stellar images. Planetary detail behaves quite differently, and the resolution that can be achieved is directly related to the contrast of the objects we are looking at. A great example that can be used from modern images is Saturn's very fine Encke division in ring A. The narrow gap has an actual width of just 325km - which converts to an apparent angular width at the ring ansae of just 0.05" arc seconds - well below the Dawes criterion of even at 50cm telescope. In `fact, the division can be recorded in a 20cm telescope under excellent seeing, exceeding the Dawes limit by a factor of 11 times!. (Quote from D Peach)
From my experience in planetary imaging, magnification is important, remembering that magnification is a function of focal length and the focal length will determines image scale. So the reason many planetary imagers use such high f/ratio e.g. f28 to f50 is to get a large enough image to fill as much of the CCD chip as possible. This way there are more pixel use to record finer detail.
Also for imaging planets there is no problem with critical sampling,
except the rotational speed of the planet itself. For example if you are to image Jupiter you will need to keep the video capture time to 90sec as the image will begun to blur after this lenght of time due to the rotaional speed of Jupiter.
You can also under sample if the video capture time is to short e.g there is not enough frames in the avi file to produce a decent image.
Final I should add an image of Mar. The original file was 10000 frames from that 2000 where used to produce the final image. I used a 10" LX200 @ f50 plus a Lumenera LU075C camera @ 37fps.
anthony, I agree with everything you have said. I was going to make a comment in my last post that all of this discussion was treating planetary images as if they were a collection of point sources rather than a continuous image - but my posts all tend to be too long anyway!
All of this has proved to me that, for planetary imaging, experience and experiment are worth 100 times more than any amount of theory.
I guess the answer to the question "what is the best Focal Length for webcam imaging" is "go out and find out"!
With that, I will go back to processing my images of Jupiter taken with my LPI at a focal length of 6,250mm!
I see that you are using a 10" LX200 @ f10 plus a 2.5x Barlow, so your effective f/ratio is f25. The magnification is around 125x, you probly can increase your f/ratio to f30 or high. Important tip with barlows, buy one that has optimum aberration control such as the UO Klee 2.8x ot the TeleVue Powermate. Other things to consider is collimation, temperture of the scope, weather etc.
I see that you use the LPI, it is a good beginner camera. When you have mastered some of the basic I would recommend upgrading to a camera similar to the Lumenera or the camera that Bird is currently using. They are well worth the money and you will have so much fun.
Hi Anthony, I have just bought a Televue 2.5x Powermate (used it for the first time on Wednesday morning). Up 'til now I have been using a cheap 2x Barlow that has been rolling around in my 'spares' box for about 15 years.
Right now I am struggling with the issues of seeing, collimation and scope cooldown. Planetary imaging is really about attention to detail!
The LPI is a great camera for learning but, as I learn more, I become more aware of it's limitations. The larger pixel size (8 micron vs 5.6 micron for a ToUCam) means less image scale for the same focal length, the lower sensitivity means longer exposure times and less frames than the ToUCam brigade. I'm still very happy with the LPI (especially as I got it for 'free' with my scope) - at this stage my results are not limited by the camera, but by my skills in other areas. Once the camera becomes the limiting factor I will look to upgrade to something better.
Once the camera becomes the limiting factor I will look to upgrade to something better.
i would double check with http://www.telescopes-astronomy.com....am_philips.htm when a prime focus adapter will be available for the 900NC. I would say that the $30 extra, this will still be a great imaging tool. If you can still pick the 840k, then the improved drivers from the 900nc work in it. My gut feel is that hopefully there is a slight speed advantage on board with the 900nc, so that even though it is not usb2.0, you can still take advantage of the higher data frames.
Done a bit more homework. In Capen/Dobbin book "observing and photographing the Solar system" a bit dated now but still has some great info on all the planets...
They agree visually the resolution on planets is controlled by the contrast in the image and a resolution of 1/8 to 1/10 Dawes limit is achievable.
For photographic the resolution is not only the contrast but the seeing v's detector ie film grain size and sensitivity ( or today the CCD pixel size)
I think therefore for hi resolution planetary work we should assume the telescope system can give a resolution of 1/10 Dawes and then work backwards to the EFL to present this detail to the CCD pixels where at least two pixels cover the best resolution.
