Canon Digital Rebel T2i (550D), T1i (500D), XSi (450D) and XS (1000D) Comparison

Long Exposure Dark Frame Noise & Temperature Testing

by Gary Honis

UPDATE: Initial tests of T3 (1100D) added near bottom of page.


Prior to this testing, I did dark frame and sensitivity of the four Canon models HERE and HERE. The objective of the testing done as described below was to determine how the four Canon models behave noise-wise over a two hour imaging session of continuous exposures. As Canon introduces new camera models with increasing number of pixels for the same sensor size and adds additional features including in-camera video capture, what is the effect on long exposure astro imaging?

Here are some key specifications for the four camera models tested:

Test Setup:

A temperature probe was taped to the top of each camera near the flash shoe as shown below. Each camera body was capped, the viewfinder covered with black electrical tape and a black cloth was placed over the camera. A remote timer was used to trigger the exposures. Testing was done at room temperature.

For all four cameras, in the Custom Functions menu item, long exposure noise reduction was turned off and high ISO speed noise reduction was disabled. Live View Shooting was also disabled in all four cameras and their camera back LCD displays were turned off. Highlight tone priority was disabled for the 450D, 500D and 550D. Auto Lighting Optimizer and Peripheral Illumination Correction was disabled in the 500D and 550D. Here are the settings used for all four cameras as displayed by Canon's DPP. I find that these settings are the best for reduced noise long exposure imaging:

Dark Frames and Noise - Uncooled:

For these dark frame tests the cameras were all powered with an AC power adapter and the same 4GB SD card was used to store the RAW images. 24 five-minute exposure dark frames at ISO 1600 were taken sequentially with each camera, using a 15 second delay between exposures. The Canon Digital Photo Professional thumbnail with information screen shots for the CR2 RAW files are shown below for the 1st exposure, 12th exposure at 1 hour and 24th exposure at 2 hours:

Notice from the above, that according to the histograms of the initial 5-minute exposures that the 450D, 500D and 550D appear better than that for the 1000D, with the 450D histogram showing the lowest noise level. As the cameras heat up, based on the exposures taken at one hour and two hours, the histogram of the 450D shows the lowest level of noise, while that of for the 1000D shows the lowest RATE of increase in the level of dark frame noise. The 500D and 550D show the greatest rate of increase. These results are in agreement with previous 5-minute exposure test results taken over a half-hour period posted HERE.

Keep in mind that noise performance will also vary among cameras of the same model as can be seen HERE. On that page I analyzed the initial 2-minute dark frame for 30 cameras and the results show that some 500D histograms are better than some 450D histograms but in general the initial exposure histograms are worse for the 1000D.

Camera temperature and thermal noise:

I measured the camera temperature using the thermometer probe on the camera bodies and also recorded the EXIF temperature readings from the RAW dark frame files over the two hour period.

The temperature probe readings show that the 1000D has the lowest increase in temperature while the 500D and 550D models have the highest rate of increase. The 450D model temperature increase is higher than the 1000D but lower than the 500D and 550D. The differences in temperature may be due in part to the material used for the camera model bodies. The thermal conductivity of some plastics, like certain metal alloys, can be very good. Plastic manufacturers are now producing thermally conductive plastics with thermal conductivities that exceed that of stainless steel. The advantage of thermoplastics is that they have the advantage of the lighter weight of plastic but with the high thermal conductivity of metals. It could be that the plastic used for the Canon 1000D has a lower thermal conductivity than that of the other camera models, acting as more of an insulator, but this is only a question on my part.

The software "Dark Library" was used to record the EXIF temperatures of the dark frame RAW files as shown below. Only Canon knows where the temperature sensing component is located in each camera model and their purpose may be for Canon's use only, such as automatically turning off the 500D and 550D models when the internal camera temperature gets too high during in-camera video capture. The trend of the EXIF temperatures in the graph below correspond well to the camera body temperature probe readings in the graph above. The EXIF temperature readings do show a quick rise of internal temperature as opposed to the gradual rise in the temperature readings of the body temperature probe as one would expect. The EXIF temperature readings seem to rule out the 1000D model using a plastic with a lower thermal conductivity. These tests correspond well to earlier long exposure temperature testing done HERE.

Uncooled Standard Deviation:

Using the software "Images Plus", all Full Frame RAW files were converted to TIF files and the histogram luminosity value for image pixel standard deviation was plotted in the graph below. The standard deviation values for the 450D model stays less than for the 1000D model, even though it rises to nearly match the 1000D after 2 hours of exposures. The standard deviation values of the 500D and 550D climb noticeably higher. The standard deviation for a 550D cooled in a beverage cooler while taking exposures over the two hour period is also included in the graph (pink line in the graph).

Using the software "Images Plus" (IP) the first, 12th and 24th five-minute exposure full frame darks at ISO 1600 were converted to TIF files. All files were then stretched using the IP Digital Development function with a Break-point setting of 5000 and cropped to 300 X 300 pixels at center. The resulting 300 X 300 pixel crops for all four models are shown below with the first exposure at top, 12th exposure in the middle and 24th exposure at the bottom position:

Instead of cropping a 300 X 300 pixel section from the full dark frames as shown above, another interesting display is shown below. As above, all files were converted to TIF files and stretched using the IP Digital Development function with a Break-point setting of 5000. The resulting images were reduced to 1200 X 800 pixels in size using Photoshop and then cropped to 300 X 300 pixels at center. The resulting 300 X 300 pixel crops for all four models are shown below with the first exposure at top, 12th exposure in the middle and 24th exposure at the bottom position:

For the 300 X 300 pixel crops displayed in the image immediately above, the image pixel standard deviation values were recorded and plotted in the graph below:

Check Back: Next testing I will do will be of all four camera models while being cooled by a whole camera peltier cooler based on my Level III cooler design. Dark frame noise and sensitivity testing is planned.

Canon 1000D (T3):

On 3/21/2011 I purchased a 1000D (T3) for modifying. My inital test was to take a 5 minute dark frame at ISO 1600 as was done in the above tests of the 450D, 500D, 550D and 1000D. Camera settings were adjusted to be the same as shown for the Canon 550D above. Below are the histogram displays using Canon's Digital Photo Professional software for the initial 5 minute ISO 1600 dark frame exposure of all 5 Canon models for comparison.

The image below compares the initial 5 minute ISO 1600 dark frames of all 5 Canon models. RAW dark frame files were converted to 16-bit TIF files using Canon's DPP software and the TIFs were then cropped at center to 200 X 700 pixels:

For the above cropped TIF dark frame images, image pixel standard deviation values for luminosity were recorded using Images Plus for all 5 Canon models and are displayed in the graph below:

The true test of dark frame noise is how the camera performs over time while capturing long exposures and as the camera heats up. I do plan to test the 1100D (T3) camera in the same manner as was done for the other four Canon models over a period of two hours.

Additional test results for comparisons of the Canon Digital Rebel T2i (550D), T1i (500D), XSi (450D) and XS (1000D) can be seen HERE.

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

For my low cost astro & infrared modification service for your Canon DSLR camera CLICK HERE.

Removal of IR Cut Filter for Astrophotography

.....Order DSLR Astro Imaging Guides on CD-ROM by Jerry Lodriguss HERE......



To My Astrophotography & Digital Imaging Home Page