Arc-seconds per pixels and EOS movie resolution

[keybaud]

I've been trying to understand the concept of under-sampling and over-sampling and, although I don't fully understand over-sampling yet (why would infinite resolution be bad?), I have been doing some calculations with my own cameras.

The recommended arc-seconds per pixel is 2.5-3.5 for DSO and less for planets (I've lost the image posted on these forums showing these values, so I'll look for it later).

With a MN190 (focal length 10000mm), which I don't own yet, and my cameras, the results are as follows:

EOS 30D: 1.32 arc seconds per pixel

EOS 70D: 0.85 arc seconds per pixel

This looks like the 70D would oversample, but I've read that it is possible to use movie mode to improve this (the 30D doesn't have a movie mode). Please can someone confirm that the following assumptions are correct:

1. Shooting at 1920x1080 uses all the pixels, but interpolates groups of them to create a lower resolution movie. Max sensor resolution width is 5472 pixels, so a 1920 movie is a reduction factor of 2.85. Now, when I drop the 1920 max resolution of the movie into the arc second per pixel calculator, I get a value of 2.42, but each pixel has been interpolated by the camera. My assumption is that although this is a better value for DSO objects, it is artificially created and may have issues of its own.

2. Shooting at 1920x1080, but using the 3x digital zoom, actually reduces the number of pixels being used within the sensor (I assume that it actually drops to 1920 and the 3x is approximate, as a true 3x will only use 1824 pixels, which just seems silly). This will also have 2.42 arc seconds per pixel, but it is now a 1:1 mapping. I assume that this is better than 1, even though less light has been captured.

3. It is possible to shoot at lower resolutions, but digital zoom only operates at 1920 modes, so running at 640x480 would just group even more pixels together and I could do that myself from the original 1920 movie. The only other benefit is increased movie durations, due to the smaller file size.

4. As the movie arc seconds per pixel is the recommended value for DSOs, should I be using movies instead of still images on the 70D or am I better off sticking a 2x Barlow on and using the 30D at 2.64 arc seconds per pixel?

Having spent the day reading up on back focus and arc seconds per pixel, my head is going to explode, so please be gentle with your replies!

For the geeks out there, I stole the formulae from: http://www.wilmslowastro.com/software/formulae.htm

[keybaud]

Some of my maths is wrong. I realised that on a 1:1 ratio 1920 movie the arc seconds per pixel can't change, because nothing else has changed apart from the number of pixels actually capturing the image. The error was caused by my not reducing the sensor size in the spreadsheet formulae, as this has now reduced to the number of pixels being used. This means that I can only get 2.43 if I use the whole sensor and the camera interpolates. The options are now:

30D

1.32 from all pixels with an image

70D

0.85 from all the pixels with an image

2.42 from all the pixels with a movie

0.85 from a subset of the pixels at a 1:1 mapping with a movie, but with a reduced field of view (factor of 2.85).

[narrowbandpaul]

With the over and under sampling the theory is, and it's just theory, that your seeing should span 2 pixels. Typically in the UK this could be from 2-4" and can vary over the course of a night. A few places on earth have sub arcsecond seeing.

So with this in mind a good scale would be from 1-2"/pix.

With this movie mode you describe you are effectively binning the sensor. Not throwing it in the trash, but grouping a 2x2 or 3x3 matrix of pixels together and calling it a single pixel. This can also be done in software. The movie mode will feature short integration times so will be very limiting. Shoot long subs at high res and then bin them in software if you think it's appropriate. The reason the camera bins in movie mode is to reduce the file size so it can capture at 30fps.

From the specs you quote the 30D has a better image scale with your scope and typical British skies.

Hope that helps

Paul