Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Signal to Noise: Part 3 - Measuring your Camera

This topic has been archived. This means that you cannot reply to this topic.
10 replies to this topic

#1 mloffland

mloffland

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2510
  • Joined: 03 Sep 2004

Posted 06 July 2009 - 10:50 AM

Signal to Noise: Part 3 - Measuring your Camera

#2 RHK

RHK

    Lift Off

  • -----
  • Posts: 3
  • Joined: 25 Mar 2008

Posted 09 July 2009 - 10:43 AM

Craig, I've been trying to (slowly) follow your signal to noise articles because I'm tempted to try astrophotography with a DSLR camera so that I don't need a laptop with me at the scope for the duration of the sessions. I assumed that I could do post-processing of the images from the flash card with s/w on my iMac later. I was therefore surprised to see this statement in your most recent article: "Now, fire up the camera and connect in your capture software." This was just before taking a set of at least 30 dark frames. How does "capture software" play at this point? If we were using a dedicated CCD imager I could understand because it can't function without s/w running on a laptop, but with a DSLR I'm clueless. How and why does a DSLR need to interface with capture software? I do have a USB cable for the camera of course, but I've only used it for downloading photos directly to a laptop or desktop PC or connecting with photo editing software, like iPhoto or Photoshop Elements, but I haven't done that frequently. I usually download photos from the flash card through a flash card reader. Could you please explain your statement? Sorry if this is all obvious to the vast majority of you readers, but I can't move on if I don't understand. Thanks.

#3 Mike Clemens

Mike Clemens

    Frozen to Eyepiece

  • *****
  • Posts: 7753
  • Joined: 26 Nov 2005

Posted 10 July 2009 - 04:56 PM

Most people probably sequence their shots with an attached computer. If you can produce the requested flats, bias and darks of the appropriate duration without capture software, thats fine.

#4 Craig

Craig

    Vendor (Stark Labs)

  • *****
  • Vendors
  • Posts: 770
  • Joined: 16 Sep 2003

Posted 12 July 2009 - 01:16 AM

Mike is correct. I made the assumption that you'll be using capture software to do this. But, if you've got an interval timer for your DSLR, that will work just fine. I'd note that whatever loads those raw files should ideally load the raw bayer matrix directly and show you a monochrome image on screen and not a color image. Your sensor is really a mono sensor with an array of color filters on it and software is used to make it a color image. If you do your tests based on the color image, you're also testing whatever the software did to make the color image ("debayering" or "demosaicing").

Some of the software out there will load it as the raw data and others won't. FWIW, one can always use "dcraw" to get this pure sensor data.

Craig

#5 Agnotio

Agnotio

    Ranger 4

  • -----
  • Posts: 387
  • Joined: 29 Aug 2008

Posted 22 July 2009 - 01:45 PM

Hi Craig,

I'd like to say thanks for the series. I followed the whole guide, and even joined up AstroPhoto Insight so I could read your articles analyzing other cameras.

Some of the results I got are that the system gain of my Canon Xsi (450D) is 0.30 e-/ADU at ISO 800, with a full-well capacity of roughly 4850 electrons with the 14-bit ADC. The read noise was 4.6e- at ISO 800.

However, I couldn't measure the dark current because of the problem you mentioned in other articles: the scaling that Canon applies to the dark frames. The mean of my subframes decreased with increased exposure time while the standard deviation increased, just like you observed. This is really a shame because it means dark subtraction won't work optimally on Canon DSLRs with automatic scaling. I wish there was an option to turn it off. Although, I have to say I haven't noticed this effect from my suburban area; subtracting dark frames gets rid of hot pixels, which is all I'm really looking for as photon noise from skyglow is a much bigger problem than dark current for me. It looks like I will have to revise my habit of taking the read noise out of my image using dark frames though, as the scaling will throw a wrench in this process as well. I wonder if there's any way around it?

One result I got that was anomalous was that the log-transformed histogram of my read noise frame had shoulders almost like you got for the Canon 5D Mk II. Apparently, there is non-repeatable horizontal pattern noise there, because the FFT of the read noise frame still showed a vertical line instead of just random noise. Some disappointing results there, maybe you just put me in the market for a CCD :lol:

Anyway, Cheers.

#6 bill w

bill w

    Voyager 1

  • *****
  • Posts: 11279
  • Joined: 26 Mar 2005

Posted 22 July 2009 - 07:09 PM

finally made it through ;)
love the series
thanks for writing it up

#7 Craig

Craig

    Vendor (Stark Labs)

  • *****
  • Vendors
  • Posts: 770
  • Joined: 16 Sep 2003

Posted 25 July 2009 - 10:19 PM

Glad to hear some are working their way through it! Yes, you'll find that the Canon DSLRs aren't "textbook" in a number of ways. But, you do get a huge chip for a reasonable price and can take shots of the kids with that. Dedicated astro-CCDs tend to fall short in these ways (although I have seen shots of Alan Friedman's kids taken with his astro-cams).

Craig

#8 Barry E.

Barry E.

    Mariner 2

  • -----
  • Posts: 204
  • Joined: 27 Dec 2008

Posted 14 September 2009 - 11:52 AM

Great article, Craig. I finally had the time to sit down and go through this analysis with my modified Canon 40D and here's what I got:

System Gain: 0.049 e-/ADU
Full-Well Capacity: 3162 e-
Read Noise: 4.255 e-
Dark Current: 0.010 e-
Dark Stability: About 25 minutes
Bias: Stack of 100 frames FFT, showed 3 faint vertical lines/frequencies.

I'm looking forward to your next installment and some advice on how to take this analysis and do something with it.

#9 themos

themos

    Sputnik

  • -----
  • Posts: 40
  • Joined: 01 Apr 2008

Posted 13 October 2009 - 03:26 PM

For my Canon 450D at ISO400 (nominal)

gain is 0.568 e-/ADU
read error is 6.3 e-
full well capacity is 9143 electrons
dynamic range is 1449 or 32db

dark stability: unknown, std dev of 1-minute darks climbed from 13.9 to 17.4 and had not levelled off after 30 dark frames.

#10 Barry E.

Barry E.

    Mariner 2

  • -----
  • Posts: 204
  • Joined: 27 Dec 2008

Posted 13 March 2010 - 07:47 PM

Craig,
I know this is an old thread, but since I got responses from the Sampling thread, I'll try my luck here with another question.

Let's say I'm enjoying imaging with my Canon 40D during the nice cold winter (I assume that the cold weather means better SNR). Now, spring and summer are approaching, which means warmer weather. What will this mean for me?

I would like to prepare for the warmer weather by examining my winter images and calculating what I need to change for warmer temps. Let's say that I like the results I got from 40x300s @ ISO800 @ 35F, meaning whatever that SNR was, it's acceptable and a good target to shot for.

Now, can I somehow measure dark frames taken indoors (72F) and figure out what ISO, exposure length, and # exposures I would need when it gets to be 72F outside to get the same SNR I had during the winter?

#11 Craig

Craig

    Vendor (Stark Labs)

  • *****
  • Vendors
  • Posts: 770
  • Joined: 16 Sep 2003

Posted 17 March 2010 - 11:48 PM

In theory you could measure your dark current and then use this in the spreadsheet. In practice on DSLRs, it's not trivial. While it may be 35F outside, it's not 35F in the camera and the temp in the camera will rise (at least with the ambient at room temp) for a *long* time. Worse still, you can't use the actual signal levels that come off the camera to measure the dark current. You need to base it off the variance. Canon shifts and/or scales the image to keep the background level roughly constant as the temp goes up. The noise however goes up with it. You can see this in several of the reviews up on my site (where, at times, the background level goes *down* with increasing exposure duration).

Craig


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics