Olympus E-M5: Dynamic Range - RAW
April 27, 2012 at 15:48 I posted in February a blog on E-M5 dynamic range. At that time it was about jpeg images. Now we have two popular and well known converters supporting E-M5, which enabled me to take another look with raw files. These converters are Lightroom 4.1 (Release Candidate 2) and CaptureOne 6.4.
This graph shows the same E-M5 and E-P3 dynamic ranges (orange and red) for jpeg images as shown in my previous blog. They are actually measured from raw files converted into jpegs in Olympus Viewer 2. This software emulates in camera jpeg process, which means I can tweak the jpegs to show optimal jpeg quality.
Blue and green lines show what can be achived from E-M5 and E-P3 raw files. I opened raw files into Lightroom and CaptureOne and measured dynamic ranges at various ISOs. Both software showed almost the same numbers for E-M5. With E-P3 there was slight variation at some ISOs but there was no trend in favor of either software. What you can see here is the average performance of these software. The biggest and maybe most important difference was for E-M5 at ISO 200. With Lightroom I was able to read a 12.6 EV dynamic range and with CaptureOne it was 12.0EV. At ISO 400 they both gave the same 12EV.
When you look at this graph, please do not take exact numbers too seriously. Look at trends E-M5 versus E-P3 and raw versus jpeg. While saying that I think it is no mistake to have E-P3 dynamic range at 10EV, which is the same as DxO Mark result for the same camera. Also two well respected software giving practically same numbers (sans ISO 200) for E-M5 should be no mistake either.
I have now shot a few thousand images with E-M5. Those images show the same trend in real life situations compared to E-P3 as this graph.
-p-
CaptureOne, Dynamic range, E-M5, E-P3, Lightroom, OM-D, Olympus, PEN, RAW in Camera Bodies, Image Quality, Test
Reader Comments (7)
Interesting! I wonder what the JPEG dynamic range would be with gradation=auto. Since you have the raw file, I suppose you could test it in Viewer.
Auto gradation would not do any good because my target is grayscale. DR is not measured from a file but from a series of shots which range from total blackness into absolute whiteness. No automatics should interfere here by changing tones uncontrolled. You can be sure that I have done everything I can to extract the absolute maximum that can be found in Olympus jpegs. In real life situation you get always less.
-p-
Well, with gradation=auto the camera includes more of the raw image's dynamic range into the jpeg. At least that's what it seems to me. And this is backed up by DPreview's dynamic range measurement. It's possible that it will not show with your measurement method but for real-life shooting it should offer a jpeg dynamic range that is closer to your figures for raw shooting.
There have been a number of reviews suggesting that the in camera iso settings for the E-M5 (above 200) may be inflated by more than one stop. Could you address this as it would have a considerable effect on dynamic range comparison? e.g moving all points to the left by one stop for the E-M5 would halve the dynamic range advantage with respect to the E-P3.
Bob: E-M5 and E-P3 have a similar ISO-speed calibration. It can be seen how similarly exposure metering works in both bodies and also how E-P3 dynamic range stays very nicely in the middle inside E-M5 dynamic range at every nominal ISO. I have no equipment to test how Olympus way of defining ISO is related to ISO standard. As said these two cameras are alike.
-p-
Edit: DPReview´s E-M5 review came out and they state that E-M5 ISO is "optimistic" by 1/3 stops. The same as E-P3 - and many other cameras.
Great work! I like to ask you about the testing process. How was it taken?
Thanks!
I used the simplest method there is:
- Shot a greyscale at widely varying exposures (at each ISO) from absolute black into absolute white.
- Chose a step in that scale and found out how far (in my exposures) into black this step can be seen differing out of noise and how far into white it can be seen differing from white in said converters. The latter is relatively easy to measure, the former is not easy to measure or see. Because of that I had to do lots of comparisons between shots and with actual images to keep a standard of what I consider as detail.
There is no unified definition of how much and in which way a detail must rise above noise to count. That´s why DR measurements in the internet differ so widely. Also there is a difference between hardware measured and perceived difference. I tried to keep to what I think as the last detail before noise in an actual photograph, because that´s the only thing that counts for me.
-p-