Dynamic Range comparison of RAW v/s video mode on the Canon 550D

Does a DSLR output the same dynamic range in stills mode (RAW) and video? I decided to test for myself to see if that was the case.

Test conditions and steps (RAW):

  • Camera: 550D. Lens: EFS 18-55 3.5-5.6 II at 35mm (56mm on the APS-C)
  • Aperture: f/8; Shutter: 1/30 sec (for both still and video); ISO: 100
  • Measured DR of the scene (using the spot meter and the zone system): 14.5 stops. This is outside the ability of the camera, and allowed me to find out ‘hidden’ detail in the highlights and shadows, if any.
  • Stills mode: RAW. Resolution: 5184×3456. WB: Shade. Style: Neutral (flattest possible in camera). Color space: sRGB
  • RAW file was Trimmed and resized in DPP to an uncompressed 1920×1080 TIFF 16-bit. This was converted to an 8-bit JPEG in Photoshop, at maximum settings. This was done to remove the color advantage offered by RAW and 16-bit.
  • From the above results, we get the base file (MOUSE OVER THE IMAGE FOR DESCRIPTION):
    RAW trimmed and resized in DPP

    Test conditions and steps for the video:

  • Same settings as RAW
  • Recorded as 1920×1080 24p H.264 file in a MOV Container. Color space: Rec. 709
  • Still from the 10 second clip was recorded using Nuke to an uncompressed 1920×1080 16-bit TIFF file, while converting the color space to sRGB. As above, this was converted to an 8-bit JPEG.
  • This file, the base HD stills JPEG, follows:
    HDStill

    What do we learn so far?
    No difference. At these settings, the DR output is the same. It was to be expected. From here on, things get weirder.

    Post Processing the RAW file for additional information:

  • In DPP, the RAW file was trimmed and resized as before.
  • The Brightness (under the RAW tab) was turned to -2, the lowest possible setting.
  • The file was exported as a 16-bit TIFF file, and then an 8-bit JPEG as before.
  • The same thing was repeated by pushing the Brightness slider to +2 (the maximum possible), and exported to TIFF and then JPEG
  • Here are the results (First +2, and second -2):
    RAW trimmed and resized in DPP Brightness minus two
    RAW trimmed and resized in DPP Brightness plus two

    So there is good information hidden in both the shadows and the highlights in the RAW file. Most of this information is Noise-free. The question is, can the video match this performance? I went about testing this using both DPP and Photoshop, just to make sure. First, the DPP results.

    Post Processing the HDStill in DPP:

  • Under the RGB tab, the brightness was increased to +100 (highest) and exported as TIFF and then JPEG
  • Again, the brightness was reduced to -100 (lowest) and exported as TIFF and then JPEG
  • The results are as follows:
    HDStill Brightness plus 100
    HDStill Brightness minus 100

    I was sort of expecting this, mostly because I’m new to DPP. So I opened the HDStill file in Photoshop.

    Post Processing the HDSTill in Photoshop:

  • Using the levels tool, the gamma slider was pushed to the right and left until details emerged in the highlights and shadows.
  • Mode was changed to 8-bit, and files were saved as JPEG in the Maximum setting, as before.
  • The results are as follows:
    HDStill PS Shadows
    HDStill PS Hightlight

    The truth is plain to see:

    The Dynamic Range of an HDSLR in RAW mode is far greater than that in video mode, possibly by 3-4 stops.

    At least for me, I’m convinced by RAW, and will be using that exclusively for my photography from here on.