Well, most of you can guess the champion without actually seeing one line of text … the D3s is the world champion in low light, and no one is really all that close. But that’s not why I include the D3s … I include it as a control so we can see what should be in the crops I am going to use to analyze these cameras at 1600 ISO.
Note: Many have commented that I have stepped on my … well, you know … by comparing the XZ-1 at 1600 against the Canon at 1600. Fair enough. I will be publishing comparisons at the appropriate ISO deltas later today. Stay tuned …
The new article is here.
Those of you who have been to this blog before know that I like to bring my own brand of
bullshit brilliant analysis to what I see. I will take a set of crops from the DPReview comparison engine and interpret them for you. You are free to disagree … if you do, though, I’d appreciate a comment added to this article so that I can answer you and correct the text if I agree.
You can read about this new Olympus camera at the DPReview site on this page. So I won’t bother repeating anything but to say that the camera is considered an enthusiast cam, which puts it squarely against the S95/G12 and the LX5. So that’s who I will use for comparison. Simple enough …
One more thing … the D3s does not have 1600 ISO crops available, so we’ll look at 3200 ISO crops. Close enough, since the D3s has about a 5 stop advantage over the compacts …
A penny for your thoughts … well, maybe not a penny. But a coin nonetheless. And the Olympus really mucks it up with classic smearing and chroma noise. The yellow blotching is disturbing, since this is not even shadow. The other two give a credible performance, mimicking but not matching the D3s.
I like the yarn area for testing the ability to render subtle tonal changes and very low contrast details (individual threads that make up the larger threads) which make or break the feeling of texture and 3 dimensions. The D3s is magnificent here (duh) but the Olympus is not bad. It clearly has a bit more contrast than the others, which helps in this test. But the heavy NR also creates a slight flattening effect that removes some of the 3D look. The G12 is a bit better in some areas and worse in others. So in the end this is mostly a tie between the two. I think the LX5 comes in third her, but only by a hair.
In the paper clip area, the Olympus sweeps to an easy victory. This is the perfect subject for its extra contrast and one thing you notice is the lower level of edge artifacts that are clearly evident in the other two. The D3s soft because of depth of field in case you were wondering.
An area that really surprised me in the clock face. The others all give a very credible rendering of the small tick marks. But the Olympus completely bungles this at the top of the face. They aren’t even visible … here we see the dangers of excessive noise reduction. So the Olympus comes in third by many lengths …
In the opposite corner, the heavier processing in the Olympus shows up as improved saturation, almost matching the colors and richness of the D3s. But the edges take on a hard look with many artifacts that the others do not have. I would say that this will be ok in print, but will cause problems for those who like to process their images. Definitely look for ways to turn down sharpening, NR, contrast and such if you plan to process.
Looking deep into the shadows, we see that the Olympus is again completely outmatched. Very excessive blue channel noise reminds me of the G10. Probably caused by similarly too-small pixels.
Another example where the Olympus processing is playing havoc with fine edges. The others are soft, but very credible looking. The Panny surprises here with excess noise in a non-shadow area. That would make me a bit nervous about using it in low light.
The globe is the clearest example yet of the dangers of excess noise reduction. This is just nasty … while the other two are very credible here, as with the watch face.
With the sculpture image, the excess NR in the Olympus removes enough detail to hand a fairly easy win to the other two. The LX5 is pretty off color here, but then it seems to do that. A weakness in my opinion.
The paintbrush on the left side hands the Olympus another tie with the Canon.Not a bad rendering, although some grain shows up. The bottle in the background shows excessive blue channel noise from the Olympus, but the Canon and Panny have really blotchy rendering of the bottle’s smooth surface, so all in all I consider this just a mess for them all :-)
Ok … enough. There is lots more to explore if you want to … go here.
- Bottom line – this cam is a challenge in low light.
- The Olympus has too much blue channel noise in shadows.
- The Olympus has too much noise reduction, which mucks with edges a lot.
- The extra contrast of the Olympus does look good on some subjects.
- The Panny has white balance issues.
- The Panny is better than the Olympus in most crops, but never better than the Canon.
So … for a general purpose shooter without long reach as a requirement, the G12 would be my choice. For a compact in the pocket, it’s sister cam, the S95.
Update: For web use, they are all fine as shown in the second article here. More importantly, that article compensates for the differences in lens speed at wide and at telephoto. So read that one for a more accurate take on the subject.
RAW does not change the results very much, as this crop shows … I changed out the D3s for the D7000 because the D3s only had 200 ISO RAW crops …
That blue channel noise makes processing the Olympus a nightmare. I know … I processed many a Canon G10 image in RAW.