Canon Digital Rebel T1i (500D)

Dark Frame Noise and Sensitivity Comparison

to XSi (450D) and XS (1000D)

by Gary Honis

Introduction:

As 2009 comes to and end, the Canon 500D (T1i) has been on the market for eight months, and examples of astro images with the DSLR are just now becoming a little easier to find online. The Canon 450D and 1000D have been very popular cameras for astro-imaging and maybe because they have been performing so well, those considering new cameras have been opting for the proven cameras instead? I had the opportunity to test three cameras sent to me for modification. For theses tests, I tried to have good control conditions. The cameras were all purchased from the same supplier and were tested before modification. Testing of all three cameras was done at the same time with imaging done indoors and at room temperature. Here are some key specifications for all three camera models:

Physical Size:

The 450D and 500D have the same exterior dimensions and other than the dial style difference and four tiny hole pattern for the 500D's speaker, would be difficult to tell apart. The 1000D is the same height and width but is a little shorter in length and has a smaller LCD display.

 

Dark Frames and Noise:

For these dark frame tests the cameras were all powered with the same battery and the same 4GB SD card was used to store the RAW images. Bulb exposures were taken using a Canon RS-80N3 remote switch for all three cameras. The viewfinders were covered to block any ambient light in a darkened room. No lens was used; camera bodies were capped. For the Canon 500D, in the Custom Functions menu item, long exposure noise reduction was turned off and high ISO speed reduction was disabled. Dark frames were taken at one minute, three minute and five minute exposures.

Three-minute exposure darks at ISO 1600 - full frame and reduced to thumbnail size:

Five-minute exposure darks at ISO 1600 - full frame and reduced to thumbnail size:


For the dark frame comparisons below, the original RAW images were converted to TIF files in Canon's Digital Photo Professional and then in Photoshop all three were resized to 900X600 pixels, then cropped to 300X600 pixels at center to fit on the display for comparison.

Three-minute exposure darks at ISO 1600:

Three-minute exposure darks at ISO 1600 as above, but stretched for comparison:

Five-minute exposure darks at ISO 1600:

Five-minute exposure darks at ISO 1600 as above, but stretched for comparison:

Check for banding:

For long exposure astro images, some DSLRs have been found to suffer from wide horizontal banding, most noteably the Canon 40D. I have not had the banding issue on the modified 300D and 450D that I have done most of my deep sky imaging with. I had a discussion with Gedas Vysniauskas who has struggled with the problem and from his testing of various cameras of his and of others, he believes the banding issue varies from camera to camera and not as much as from model to model. To test for banding in these three cameras, I greatly stretched the five-minute noise comparison image using Images Plus' Digital Development function with a setting of 2000 for the break-point as seen below and there was no evidence of banding. With the high level of stretching applied, these three images don't look anything alike......do they?


30 Cameras Analyzed:

When I do camera modifications, part of the testing before and after each modification, is a series of two-minute dark frame exposures at ISO 1600 to check for banding issues. In doing so, I have noticed that some cameras, of the same camera model, have lower dark frame noise than others. The analysis below provides data for ten 1000D, 450D and 500D cameras sent to me recently for modification. Canon's Digital Photo Professional was used to display the histogram of the two-minute ISO 1600 RAW (CR2) files. The software ExifTool was used to extract the camera temperature in degrees Celsius from each RAW file since the ambient temperature varied somewhat during the dark frame tests. These dark frames were the initial exposures upon powering up the camera.

I converted the two-minute ISO 1600 dark frame RAWS to TIFs using Canon's Digital Photo Professional and then recorded the standard deviation of the whole color image for each dark frame using Images Plus. The exposures are shown below in order of increasing camera temperature for each camera model:

The standard deviation values above are listed only to compare cameras of the same camera model. As camera temperature increases, one would expect dark frame noise to increase. In the above graphic I've marked the cameras that appear to be somewhat noisier with a red asterisk * and those that appear to have less dark frame noise with a green asterisk *.


Sensitivity

Test object:

For the sensitivity tests below, the cameras were all powered with the same battery and the same 4GB SD card was used to store the RAW images. Bulb exposures were taken using a Canon RS-80N3 remote switch for all three cameras. The viewfinders were covered to block any ambient light in a darkened room. A Sigma Zoom lens was used at 200mm and F32 with the cameras set at ISO 1600. For the Canon 500D, in the Custom Functions menu item, "long exposure noise reduction" was turned off and "high ISO speed noise reduction" was disabled. Exposures were taken at one minute, three minute and five minute lengths. Liveview focusing was used at 10X zoom for all three cameras.Imaging was done in a dark room with a 5 watt incandescent lamp in one corner of the room. The lamp was covered with a diffuse material to limit the amount of illumination for the testing. This was done to crudely simulate astro imaging of a faint deep sky object. Dark frames were not subtracted.

Full Frame screen dumps from Canon's Digital Photo Professional "thumbnail with information" view:


For the image comparisons below, the original RAW images were converted to TIF files in Canon's Digital Photo Professional and then in Photoshop all three were resized to 900X600 pixels, then cropped to 300X600 pixels at center to fit on the display for comparison.

One-minute exposures at ISO 1600:

Three-minute exposures at ISO 1600:

Five-minute exposures at ISO 1600:

Five-minute exposures at ISO 1600 calibrated with one dark frame subtraction:

Photoshop levels and curves was applied to the image above for display with better color separation:

Conclusion:

Based on the above ISO 1600 dark frame tests, my opinion is that the 500D camera tested displays less noise than both the 450D and 1000D cameras tested. The DPP "thumbnail with information" view comparisons for sensitivity testing shows a very slight edge in brightness for the 450D. The five-minute exposure sensitivity test images show a slightly better overall image for the 500D, both with and without a dark frame subtraction.

My personal opinion is that other than the obvious difference of lower noise in the 500D image, all three cameras performed well in the comparison testing , provide surprisingly similar results and marginal differences in the sensitivity images are barely noticeable.

What do you think?


For discussions on DSLR modifications and cooling for astro imaging, please consider joining the DSLRmodifications Yahoo Discussion Group HERE.


A question was raised on the DSLRmodifications forum about the possibility of banding in the 1000D image with a request to "way over stretch" the image to check for banding. I dark frame calibrated the 1, 3 and 5-minute test exposures taken with the 1000D and applied a severe stretch in Images Plus with a Digital Development function set at a break-point of 300. Here is the result of all three "way over stretched" images as an animated .GIF file:

Note that any horizontal bands change in position with each exposure. I think that the horizontal lines that appear in these extemely overstretched single exposure images would not appear in any final astro image after stacking numerous dark-calibrated frames.


A request was made on the DSLRmodifications forum to provide some quantitative information on the dark frame test results of each camera. Using the "photometric information" function of the software Images Plus for the luminance of each entire dark frame RAW image, the data was recorded and an excel spreadsheet was used for the following graphs.

 


A question was raised on the DSLRmodifications forum about interpreting the quantitative information of the graphs above, obtained from Images Plus software. Using Photoshop CS instead, a similar presentation is made below. The original RAW dark frames for each camera and exposure length (1, 3 & 5 minutes) were conveted to TIF files and then cropped to 2500 X 2500 pixels. The "Standard Deviation" and "Mean" values for the entire cropped image as displayed in Photoshop's Histogram Window for luminosity.


Canon 10D, 300D, 450D, 1000D & 500D Dark Frame Comparison:

How has the long exposure noise of the Canon DSLR cameras improved over time? The same three cameras above (450D, 1000D & 500D) introduced in 2008 and 2009 were compared to the Canon 10D and Canon 300D, both introduced in 2003. For dark frame testing of the 10D and 300D, the cameras were powered with the same battery and the same 1GB CF card was used to store the RAW images. Bulb exposures were taken using a Canon RS-80N3 remote switch. The viewfinders were covered to block any ambient light in a darkened room. No lens was used; camera bodies were capped. . Dark frames were at ISO 1600 and five minute exposures as was done above for the 450D, 1000D & 500D dark frame tests.

Five-minute exposure darks at ISO 1600 - full frame and reduced to thumbnail size (no processing other than conversion from RAW and resizing):

Five-minute exposure darks at ISO 1600 as above, but stretched for comparison:

Note that the right edge amplifier glow of the 10D and 300D has been eliminated in the newer models.


Removal of IR Cut Filter for Astrophotography


Google


To My Astrophotography & Digital Imaging Home Page

Starmaster Telescope Eyepiece Astronomy Astrophotography Meade Celestron Televue Eyepieces Telescopes

Starmaster Telescope Eyepiece Astronomy Astrophotography Meade Celestron Televue Eyepieces Telescopes