5D Mark III Night Scene ISO Test

Firstly, the experimental setup...5D3 with 24mm L II


Test method: Everything was left on auto, except for aperture at f/8. In hindsight, I should have stopped down more, but I didn't install the shutter release for bulb mode and 30s f/8 iso200 was a good exposure...Shutter speed was determined by camera metering, which I must say is VERY ACCURATE compared to my 50D. Photos are out of camera jpg. I have the raw version, BUT I don't have the raw codec...

Why this test? ISO test in bright day light usually don't tell much. From my limited experience playing with camera(and it really is very limited), typical iso test in bright day light or studio controlled environment seems to be 1-2 stops better than it really is when you need that high iso. That means, the iso performance could be great in great lighting, but much worse in low light. In a way, this renders the high iso performance useless, because one will never shoot high iso with good light and high iso performance in POOR LOW LIGHT CONDITION is where it counts...Of course, proper exposure can significantly help.

So here are the night scene iso test. The 100% crops are taken from the full 22megapixel photos. This is the scene from my bedroom...



In crop 1, the focus will be smearing of detail in the sign board and noise in the background. Good or bad, you decide.
100% Crop 1
ISO200


ISO400


ISO1600


ISO6400


ISO12800


100% Crop 2
Bright and well lit part of the photo...perhaps representative of what you will get with proper exposure...Maybe the details to look for here are the noise in the sky and the sharpness of the building...

ISO200


ISO400


ISO1600


ISO6400


ISO12800


Crop3
It is a medium lit scene in the whole photo. A possible way of comparison is the detail of the brick wall. At high iso, noise control is probably smearing away all the detail of the brick wall.

ISO200


ISO400


ISO1600


ISO6400


ISO12800


Despite the iso capability of 102400, I feel that ISO 12800 has exceeded my minimum requirement for image quality. Therefore, I stopped at ISO 12800. Honestly, I will probably stick so ISO 6400 max for indoor general usage. For landscape, one probably has got a tripod. So perhaps native iso 100 might be the best option.

5D Mark 3 ISO 100% crop and effect of crop on DOF...

Previous post too long liao, lets continue here...all are 100% crop, for whatever reason, the blog can't display full 700 pixel width, so click on the photo to see full 700x700 on flickr.

5D3 ISO 1600...not sure if it is thin dof or oof...


5D3 ISO 6400


50D ISO 1600


50D ISO 3200


EPL2 ISO 1600


EPL2 ISO 3200


Honestly, at pixel level, I feel that the 5D mark III is a crazily huge improvement over the smaller APS-C and M4/3 sensors. However, it is to be noted that the 5D3 and 50D samples were downsized to 8mp and 9mp internally by the camera. I am not so sure what is the effect of downsizing here. The reason for shooting at lower resolution on the 5D3 and 50D is such that the 100% crop are showing roughly the same thing/area of the scene.

The flaw in this ISO test(and probably many other ISO test on review sites) is that the subjects have proper illumination and the 100% crop is not taken out of some shadow region. From my limited exposure to photography, the ISO performance of the shadow region usually tells a different story. Weaker ISO performer are typically exposed in the shadow noise and detail.

As for the DOF test, I tried to maintain a similar subject magnification...

EP-L2 @25mm f/4.4...sorry I don't have a fast lens for M4/3...


50D @50mm f/2.8


5D MARK III @50mm f/2.8


5D MARK III @50mm f/1.4


For DOF and background blurness, there really is no good or bad here. It just depends on what you are doing and what you need.

Maybe I have poor art sense, but here is how I will prefer to shoot the heatsink fan...
EPL2 14mm f/3.5

Upgraded camera- Canon EOS 5D mark III high ISO sample VS EOS 50D VS Olympus EP-L2

I think I might be officially out of flying RC. 700N has repeated broken clutch liner, not clutch, but the liner. E620 lipos are bloated. Been playing and spending a bit on camera and computers. Anyway, got the 5D mark III 1 day before official launch in Singapore, 16th March.

Initial thoughts is that af is very accurate even compared to the cross type af points on the 50D. However, my EP-L2 with the 14-42mm kit lens seems to be even faster than the 5D III with canon L lenses.

Test methodology:

Canon 5D MARK III- On medium, fine JPG(8MP?), Tamron 28-75 @50mm f/6.7
Canon 50D- On medium, fine JPG(9MP?), Tamron 28-75 @35mm f/6.7
Olympus PEN EP-L2- On large full 12MP JPG, 14-42mm kit @25mm f/6.3

All full size photos were uploaded to flickr as is, out of the memory card. Whatever flickr wants to do to them, I have no control anyway. Subject is just a thermaltake heatsink fan for cpu that I just grabbed off my work table.

100% Crop is done by DPP 100% crop 700pixelx700pixel, saved with JPG picture quality of 7.

Metering, white balance, NR etc etc etc, were all done in camera on auto. Of course, there will be so much inconsistency and uncontrollable variables in the test. Honestly, I wasn't expecting too much difference considering that the EP-L2 has very powerful in camera post processing. The post processing is so good that comparing the RAW with JPG, I thought the in camera post was pure magic.

Anyway here it is, ISO shootout for the 5D Mark III.

50D ISO 800


50D ISO 1600


50D ISO 3200


5D MARK 3 ISO 800


5D MARK 3 ISO 1600


5D MARK 3 ISO 6400


EPL2 ISO 800


EPL2 ISO1600


EPL2 ISO3200


Honestly, there isn't much difference at web size. Larger sample can be viewed at the flickr page...there is no ISO 3200 sample for the 5D3 because I thought it might be a piece of cake for it...NO actually that is just a more elegant way of saying I forgot to shoot at ISO 3200 for the 5D III.