wrotniak.net: E-M1 Image Samples

The new E-M1, pre-ordered early September, finally arrived on November 14, mid-afternoon. It was delivered to my office, so I told the guys at work that I have a photographic emergency, and drove home to take some outdoors pictures before dusk.

I needed another afternoon to scrutinize the images on a good screen (certainly not on my Alienware M11x laptop, with a glossy screen completely useless for viewing pictures), so the whole process took a better part of the weekend; here are my first samples from the new camera, and some remarks. These should do — at least until I find some time for a more structured image sample set.

The samples in this page:

The Crofton Lake (ISO 200)
The row houses (ISO 200)
The built-in (almost) flash
Tree in the parking lot (ISO 200)

All images shown here were shot with one lens: the MZD ED 12-50 mm F/3.5-6.3 — a lens I learned to appreciate and enjoy on the E-M5 over the past year. (Samples from the 40-150 mm and 75-300 mm will follow shortly, and I am also getting the MZD ED 12-40 mm F/2.8 PRO, so stay tuned.)

Secondly, as this was my first day with the camera, I kept all tweakable parameters at default settings: Natural image mode, sharpness/contrast/saturation at 0, image stabilization at Auto (no tripod), single AF on central spot (locked if needed), matrix metering, noise filtering at Standard. Values of other parameters will be shown in captions.

Last but not least, I'm following here my usual way of showing samples: first the full frame, just for reference, reduced in size and re-sharpened, then a selected fragment (or fragments) of that, cropped and re-saved at very low compression, no other postprocessing. In most cases, the caption will contain the link to the original full image, as saved by the camera in the JPEG format.

Notes on sample interpretation

When viewing image samples in full pixel scale (when one pixel of the image maps into one pixel on your screen), do not expect the image to be razor-sharp, i.e. resolved on a single-pixel level.

Keep in mind that on a typical (96 DPI) monitor, a full frame from a 16 MP camera corresponds to an image sized at 48×36 inches (about 120×90 cm). This is huge — diagonally a size of a 60-inch HDTV! You do not view such images from the same distance at which you sit from a computer screen — a common rule of thumb assumes a viewing distance equal to the image diagonal; in this case, 60 inches or 150 cm. If your eyes are 20" (50 cm) from the screen, most of the image flaws (noise, granularity, unsharpness) will be exaggerated by a factor of three.

To put it differently: if you make a full 16 MPix frame into a 16×12" (40×30 cm) print, then the 330-pixel height of the fragment sample as shown below would show as 1.15" (2.9 cm). This means that a print of this size would appear twice as sharp as what you see on screen.

This has been my standard sample scene for the last ten years or so: houses across the lake from where I live. It was a late afternoon, and most of the scene was dull brown, can't help that. As I do not trust matrix metering with regard to protecting small-size highlights, all these pictures were shot at ISO 200, with an exposure compensation of -0.3 EV in aperture priority AE. Auto white balance was used here.

The sample at 18 mm looks very similar. The image, if anything, may be a little bit sharper at this focal length. Still, no chromatic aberration in corners (or anywhere).

What, however, worries me a bit is the presence of sharpening artifacts, visible in the corner sample shown on the right. On Olympus SLRs I used to set the sharpening to -1 or -2, and it would make sense here as well. (Still, remember that these samples are highly magnified, so even as it is, the effect is not a real problem.)

Note that at this focal length the full lens aperture is F/5.8, so the camera had to move from my chosen F/5.6; the exposure has been therefore reduced a bit. The lens, while a bit softer than at 25 mm, is still quite sharp.

At full zoom, very much like at 40 mm. Not bad at all, as you can see in the full-frame image. (In postprocessing I would move the mid-tones down a bit, and the result would be as pleasing as anything else we have seen here.)

A few minutes later, with the Sun getting quite low, I shot these houses a few blocks from my place. Quite a wide tonal range, a challenge for any camera.

All settings — as described in the previous section.

Here the -0.3 EV adjustment worked just fine (the bright areas were larger), but -0.7 EV would be also OK. Note that Auto WB balanced out the redness of late sunlight; usually for such occasions I prefer to use the Sunny setting. Perhaps just a matter of taste.

Same here; also note that the branches in the sky don't cause any chromatic aberration or sharpening problems. A good job.

This is the tiny flash, mounted in the hot shoe, which is included with the camera and powered from its battery. I used it for a picture of a bookshelf just to check how it works.

All settings, again, were as described above, except that the ISO was raised to 800.

The next day I had to be at work, but I took the camera along and managed to squeeze out a few frames in the parking lot; here is one of them.

Actually, the effective exposure compensation here was -1.17 EV, as after having the first look at the samples shown above, I decided to apply a permanent exposure shift of -1/6 EV.

First conclusions

No surprises: the image quality, even before tweaking the camera to my liking, leaves nothing to be desired. While I don't have enough samples to comment on color rendition, some other aspects already seem quite clear.

The 12-50 mm lens is sharp; actually it seems to use the 16-megapixel sensor resolution (I never thought I would say that).

The absence of chromatic aberration in corners, especially at wide angle, is quite impressive — and bit strange. There are some hints from Olympus that the problem is detected and addressed, partially at least, at the image processing stage.

Obviously, at ISO 200 there is no chance of seeing any meaningful noise (and the sole ISO 800 sample [8] is not enough to make any judgements).

Matrix metering, consistently with my previous experiences, does not seem to protect highlights from clipping (or almost-clipping) if these highlights are relatively small. This means that in similar, high-contrast scenes, using an exposure compensation of -0.3..-1.0 EV remains a good idea. While I still have to find a camera offering a better highlight protection, I think this is technically quite feasible.

The default sharpening seems to be, as always, a tad too high for my taste, if not by far. I will be having a closer look at that.

The Crofton Lake (ISO 200) This has been my standard sample scene for the last ten years or so: houses across the lake from where I live. It was a late afternoon, and most of the scene was dull brown, can't help that. As I do not trust matrix metering with regard to protecting small-size highlights, all these pictures were shot at ISO 200, with an exposure compensation of -0.3 EV in aperture priority AE. Auto white balance was used here.

	[1] 12 mm, 1/500 s at F/5.6 Click here for full frame
Obviously, even the -0.3 EV compensation barely helped in avoiding the burn-out in the bright house siding. Still, not really bad, but usually I would prefer -0.7 EV here. The detail is good at this wide angle (except the very corners) and, surprise, I do not see almost any chromatic aberration (or color fringing) there, and that is impressive.

	[2] 18 mm, 1/500 s at F/5.6 (full frame)
The sample at 18 mm looks very similar. The image, if anything, may be a little bit sharper at this focal length. Still, no chromatic aberration in corners (or anywhere).

	[3] 25 mm, 1/320 s at F/5.6 (full frame)
Here the AE system moved the exposure up by 2/3 EV here (as compared to [2]), quite opposite to what I would expect judging from the subject. Well, perhaps that's because of less sky in the frame? But this is where matrix metering is supposed to show its intelligence (obviously, it doesn't).
What, however, worries me a bit is the presence of sharpening artifacts, visible in the corner sample shown on the right. On Olympus SLRs I used to set the sharpening to -1 or -2, and it would make sense here as well. (Still, remember that these samples are highly magnified, so even as it is, the effect is not a real problem.)

	[4] 40 mm, 1/320 s at F/5.8 (full frame)
Note that at this focal length the full lens aperture is F/5.8, so the camera had to move from my chosen F/5.6; the exposure has been therefore reduced a bit. The lens, while a bit softer than at 25 mm, is still quite sharp.

	[5] 50 mm, 1/200 s at F/6.3 (full frame)
At full zoom, very much like at 40 mm. Not bad at all, as you can see in the full-frame image. (In postprocessing I would move the mid-tones down a bit, and the result would be as pleasing as anything else we have seen here.)

The row houses A few minutes later, with the Sun getting quite low, I shot these houses a few blocks from my place. Quite a wide tonal range, a challenge for any camera. All settings — as described in the previous section.

	[6] 27 mm, 1/500 s at F/5.6 (full frame)
Here the -0.3 EV adjustment worked just fine (the bright areas were larger), but -0.7 EV would be also OK. Note that Auto WB balanced out the redness of late sunlight; usually for such occasions I prefer to use the Sunny setting. Perhaps just a matter of taste.

	[7] 30 mm, 1/800 s at F/5.6 (full frame)
Same here; also note that the branches in the sky don't cause any chromatic aberration or sharpening problems. A good job.

The built-in (almost) flash This is the tiny flash, mounted in the hot shoe, which is included with the camera and powered from its battery. I used it for a picture of a bookshelf just to check how it works. All settings, again, were as described above, except that the ISO was raised to 800.

	[8] 50 mm; aperture priority (-0.3 EV): 1/60 s at F/9.0, ISO 800, auto WB
Well, no surprises, everything very nice. No visible light fall-off (none expected at this focal length), impressive resolution, white balance on the spot.

Tree in the parking lot The next day I had to be at work, but I took the camera along and managed to squeeze out a few frames in the parking lot; here is one of them.


	[9] 50 mm; program exposure (-1.0 EV): 1/200 s at F/8.0, ISO 200, Open Shade WB
Actually, the effective exposure compensation here was -1.17 EV, as after having the first look at the samples shown above, I decided to apply a permanent exposure shift of -1/6 EV. First conclusions No surprises: the image quality, even before tweaking the camera to my liking, leaves nothing to be desired. While I don't have enough samples to comment on color rendition, some other aspects already seem quite clear. The 12-50 mm lens is sharp; actually it seems to use the 16-megapixel sensor resolution (I never thought I would say that). The absence of chromatic aberration in corners, especially at wide angle, is quite impressive — and bit strange. There are some hints from Olympus that the problem is detected and addressed, partially at least, at the image processing stage. Obviously, at ISO 200 there is no chance of seeing any meaningful noise (and the sole ISO 800 sample [8] is not enough to make any judgements). Matrix metering, consistently with my previous experiences, does not seem to protect highlights from clipping (or almost-clipping) if these highlights are relatively small. This means that in similar, high-contrast scenes, using an exposure compensation of -0.3..-1.0 EV remains a good idea. While I still have to find a camera offering a better highlight protection, I think this is technically quite feasible. The default sharpening seems to be, as always, a tad too high for my taste, if not by far. I will be having a closer look at that.