5d3 raw vs bmcc 2.5k vs scarlet mx part 1

in december some friends and i had the opportunity to do a quick test with 3 cameras that shoot raw. the hacked 5d3, black magic cinema camera 2.5k and the red scarlet mx.


the test:

we created a few scenarios that we could quickly shoot the cameras in to see how they would handle latitude, skin tone, and detail. to keep things as fair as possible we would try and shoot each camera with an equivalent focal length to the others (adjusting for sensor size), keep iso settings at 800, and shoot each with the same aperture. i’ve exported each to a 1080 resolution as that still seems to be the most standard viewing format. the raws are of course available at the end for you to play with. here are the results (in order of 5d3/bmcc/scarlet):

highlight and shadow (un-edited):

as_shot_5d3_000368 as_shot_BMCC2kEF_1_2013-12-19_1413_C0000_000485 M017_C001_12199T.0000508F

highlight and shadow (recovery):

full_recovery_5d3_000368 full_recovery_BMCC2kEF_1_2013-12-19_1413_C0000_000485 M017_C001_12199T.0000508F

skin tone (un-edited):

skin_tone_5d3_000297 skin_tone_5d3_000303

skin_tone_BMCC2kEF_1_2013-12-19_1448_C0000_000168 skin_tone_BMCC2kEF_1_2013-12-19_1450_C0001_000219

M017_C002_12192S.0000442F M017_C004_1219IG.0000253F

detail (un-edited):

detail_5d3_000349 detail_BMCC2kEF_1_2013-12-19_1517_C0000_000275 M017_C005_1219LE.0000327F


i’m not sure there are really a ton of surprises here. as an avid mark3 advocate i’ll readily admit the strengths the other two cameras have over the canon; primarily the latitude and the resolution. the big surprise was just how much that latitude seemed to help the image quality of the black magic cinema camera. i’m not a fan at all of the form factor of the bmcc, and i think for a ‘professional’ camera, the options are too limited, however that sensor sure yields a pretty image with plenty of wiggle-room.
the scarlet also i thought represented the mx sensor well, while not quite the latitude of the bmcc, it still produced lots of usable shadow detail, and the skin tones seemed just fine and plenty malleable, despite it’s detractors assertions they are unpleasing.
as for my darling, the mark 3, i still think when properly exposed creates (arguably) the best image-per-pixel of the bunch. that being said, a properly exposed image for the mark 3 has at least a few less stops of dynamic range with which to create a good image.
as for detail, raw serves all of these cameras quite well. obviously the bmcc and scarlet have a higher resolution, so cropping and punching-in will work better on those cameras, but original image to broadcast image each camera seems to be effectively the same in sharpness.

as noted in the title, this is part 1. a video showing each of the above tests is currently being cut together and when it’s finished i’ll post that as well.

click here for the original files if you’d like to play with them yourself, and probably show me where i went wrong.

3 Replies to “5d3 raw vs bmcc 2.5k vs scarlet mx part 1”

  1. Nice test, but i think there is a problem here!

    If you are comparing for dynamic range you should use each camera’s native iso and not the same for all of them.
    I guess that if you shoot all cameras at 200 or 100 iso(5d’s native) then 5d would be on top. If possible would be nice to see if that’s confirmed.

    also can’t exactly understand your skintone test. What are your conclusions on that? Seems to me that all cameras should be performing at the same level, especially after grade, but in your shots the exposure seems different so i can’t really see the point there.

    Thanks for the test!

    1. You are correct, if we tested the 5D at 100 we’d likely see some improvement on its DR. Also, if we shot it dual ISO it would have improved the DR. As someone who really likes the 5D3, this certainly was in the back of my mind, however we felt if we were to compare the cameras as cameras that could compete on the same level, there needed to be some standards. One standard would be an equal amount of available light for a usable image. ISO 800 is very standard for ‘professional’ digital cameras, and so we felt the 5D should be able to compete at this level… rather than handicapping it by 4 stops. I should have mentioned this in the original wright up. For our purposes, we felt the tests needed to be the same.
      As for the skin tone test, again you’re right. Any decent grade, I believe, could produce very pleasing skin tones. However, the test was to see ‘out-of-the-box’ how good those skin tones were. The MX chip has received countless criticisms about its ability to produce a pleasing skin tone, and this was to see how they would each compare with standard lighting and camera settings. I’m not sure why, but the BMCC operator opened up to a different aperture than the other two cameras, and while questionable, I don’t think changes the results.
      I appreciate your feedback. Once I’m done with the current project I’m working on, I’d really like to sit down and see what the 5D3 is capable of while using its best settings, so any ideas on testing methods for that would be welcome.

      1. It would also be nice to see a light sensitivity test to see if the 5d is 4 stops handicapped or less. Due to bigger sensor size, maybe iso 400 on 5d is equivalent with iso 800 on bmcc.

        Also, on the mx chip’s criticism, i feel that shooting raw or even digital, means some amount of color grading, so that is not an issue fo these cameras(you could even shoot it with 2-3 stops wrong exposure and save it later if you wanted to).

        Not a 5d user, but keep in mind that using dual iso, could possibly introduce some moire or aliasing. But still, i think, you ll be amazed by 5d’s results when using optimal settings.

Comments are closed.