Rethinking 4/3 Depth Of Field

Started Apr 11, 2013 | Discussions thread
Ken Strain Regular Member • Posts: 449
Re: Rethinking 4/3 Depth Of Field

Just another Canon shooter wrote:

Mjankor wrote:

The 1/180 and 1/200 "equivalent" exposures don't look very equivalent. The D800 shot looks awful.

Would you mind shooting a few more images.

Using exactly the same shutter speed settings (so people can't complain) and equivalent ISO and aperture, shoot the OM-D at 1600 and the D800 at 3200 and 6400.

Then possibly post the full pictures, rather than just crops?

Also, something colorful with detail. Most of the D800 noise is chroma, in the reds. But the D800 image has more saturated reds in the first palace.

Another variable is the lenses used. The zoom is likely to be darker than the prime.

Absolutely right; those two comments point to the heart of the matter.  My aim was to find out, without investing too much effort, how hard it would be to make a balanced comparison. The answer is that it is far from simple.  Exposure control is not, however, the main problem:  I'm sure the 1/6th stop shutter-speed difference is minor compared to "non-equivalence" of the processing.   I have custom dcps for two cameras, but would not trust Lightroom (which I currently use for normal processing) to match the processing.  (Also, the dcps were made for different lighting.)

Even with equality of tone-curves and so forth, there is still the question of how to resize the images for comparison, as that involves choosing the low-pass transfer function (blurring and/or sharpening).  I can see that being a point of disagreement whichever choice is made (e.g. should I have up-sampled?).

I tried to pick broadly "equivalent" lenses in terms of image quality.  The (not very satisfactory) best information to support my choice of lenses was from Lenstip. There the MTF50 results are 75lp/mm on a 12Mpixel sensor for the 75mm at f/2.5, while the Tamron comes out around 36lp/mm on a 24Mpixel sensor.  Of course that is only a hint of what to expect on 16M and 36M sensors.

I do not see this process yielding any more information than can be gained by reading Dxomark, Sensorgen or the Bill Claff pages.  Equivalent processing would probably yield results in agreement with those sites, taking into account the slight difference in the lenses.

I do not think it worthwhile to continue.  Thanks for the encouragement though.


Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow