Steve Ingraham’s Point and Shoot Landscape

So what about image quality already…

The physical layout of the camera is important, but of course, it is really all about image quality. If the image quality of the H9 had been consistently excellent I would not now own an H50 and you would not be reading this comparison.

The following image samples are of three kinds. Images that demonstrate global, large scale values are available on flickr (by clicking on the embedded image here) at 1024×768 resolution…scaled down from the original files for web display. The detail shots, which demonstrate various pixel level image quality factors, are 1 to 1 crops of relatively small sections from the original files at actual pixels. Finally, there are two sets of full resolution files in the flickr set. The shots of the blank clapboard wall, one taken at the wide-angle setting of the zooms, and one at roughly the 50mm normal setting, which are included for those want to make their own assessment of corner sharpness, purple fringing, chromatic aberation, lens distortion, etc. etc.

Except for the ISO settings all images were taken in Program mode, unmodified.

Purple Fringing

One of the failings of P&S digitals is purple fringing. PF is a senor artifact, bands of purple color along the edges (especially high contrast edges), and especially at the the corners and boarders of the image. It is caused by purple light scattering through the the tiny collector lenses that rest atop every sensor cell, and which serve to focus the incoming light on the sensor element itself. Picture the sensor as infinite array of egg cartons, with a silicon receptor in the bottom of each egg cell. Over this whole array, there is a sheet of plastic dome lenses in three primary colors, one little dome lens per cell. These dome lenses work well at the center of the sensor, but at the edges of the sensor, the light is striking the lenses at too low an angle, and some light, especially purple light, passes right through, scattering to adjacent cells, and producing purple fringes, visible mostly where light meets dark. There is also a general smearing of detail associated with PF, especially in out-of-focus detail, caused by the same light scatter. The image corners are most effected simply because they are furthest from the center of the sensor and the light is at the lowest angle.

The H9 suffered from noticeable PF and corner loss of detail, especially in low light when the lens was likely to be working at full aperture. It was really noticeable when shooting under forest canopy, looking up, or when there were darker trees against the sky in the background of shots, or in shots of buildings outlined against the sky.

I set up a couple of tests for the H9, H50, and the A200 with its kit zoom. What follow are 1 to 1, actual pixel, detail crops from full frame images. You can see them large by clicking on the image. The order will always be H9 on the left, H50 in the middle (or on the right where there are only two), A200 on the right. Clicking on the thumbnails will display the full version.

As you can maybe see from the examples, PF has been considerably improved from the H9 to the H50. There is still more than I would like to see in the H50, but it is an improvement…and even the A200 shows some (though the purple is a different shade, and may be confused with some chromatic aberration (another source of color fringing, this time related to the lens design). Also, in general, there is, to my eye, less smearing of detail in the H50 than in the H9. I’d call it a significant improvement, but far from a compete solution to the problem.

Again, these are worst case images. In most shooting I saw little PF in the H9 images, and only occasionally had to go in with software after fact to remove it. I expect the H50 to show even less PF in actual day to day use.

ISO performance

One of the reasons for going to a full DSLR is that DSLRs, with their larger sensors, are able to produce reasonable image quality at higher ISOs than P&Ss. Most P&Ss are adequate at ISO 100-200, marginal at ISO 400-800, and useless at anything higher, even though camera makes insist on building in ISO settings up to 3200 these days. A DSLR, on the other hand, will produce a decent image up to 800 ISO, and a usable image at 1600 ISO.

(Of course, a lot depends on your definition of “usable”. Most P&S will produce an adequate 4×6 print at ISOs up to 800. The colors will be a bit muddy at 800, and very muddy at 1600. However, for a full page spread in a publication, or even a full screen monitor display at anything over 800×600, you will see rapid deterioration in image quality at speeds above 400 ISO. DSLRs, as a group, produce smoother, less blotch results, especially in out-of-focus areas of the image, and look quite presentable on the printed page or even the biggest computer monitors.)

In the grid that follows the first row are all ISO 100, H9, H50, A200. Then the second row is ISO 200 in the same order. Third row ISO 400. Forth row ISO 800. Fifth row ISO 1600. All cameras were set, where possible, to normal noise reduction.

A careful study of these images will probably give you a massive headache! Beyond that though, while all three cameras produce comparable results at ISO 100, and even 200, it is pretty clear that A200 substantially outperforms ether H at ISOs 400 and above, with the margin of improvement increasing with higher ISOs. To my eye, it seems that the H50, despite it’s reputed lower standard noise reduction setting, outperforms the H9 by a slight amount beginning at ISO400 and is clearly better at ISO 1600.

That said, when the full frames (remember the images above are tiny crops of image detail) are viewed full sized at 800×600 (roughly equivalent to a 4×6 print) it is hard to see any difference at all in images taken with the three cameras at any ISO under 1600. Differences would only become apparent in enlargements or on a large computer monitor.

I would rate the ISO performance of the A200 as pretty-dang-good. I would rate the ISO performance of the H9 has good up to ISO400 and acceptable at 800. I’d say the ISO performance of the H50 is just a smidgen better than the H9, (which makes it very good at up to ISO 400, and a bit better than acceptable at 800). I would definitely use the ISO 1600 setting on the A200 in a emergency. I might try, for lack of better, the ISO 1600 setting on the H50. I would not use the H9 above ISO 800.

Those are my conclusions. You are free to study the images and make of them what you will (including the promised headache).

Noise Reduction Artifacts

Clealy the ISO performance of small sensor cameras like the Sony H series is directly related to the efficiency of the noise reduction software. To some extent, the high ISO performance DSLRs also depends on the efficiency of their noise reduction systems.

One of the persistent complaints about the image quality of the H9 was that the noise reduction and jpeg compression routines were both too aggressive, and, in combination, tended to smear fine detail in certain kinds of images taken in certain circumstances. Large expanses of grass developed smudged looking areas. Distant rock lost its fine texture. Even faces sometimes looked cartoony, with the expected details blended out.

To my eye, certain images looked painted instead of photographed, over all or some of their area. I used the term rendered…areas of the image were rendered rather than faithfully reproduce.

Of course, like a rattle in the dashboard of a car that never rattles while the mechanic is in the car, I have had a lot of trouble so far finding a situation that would push the H9 into smear mode. Images that I thought sure would show the effect simply did not. The two images that follow are small sections of detail from the image immediately following them. I see some smearing in the H9…to my eye it looks like there is slightly more detial in the same section of the image taken with the H50. Interestingly enough, I see more smearing in the same section of the image taken with A200 than I do in the image taken with the H50.

The fact is that these sections of images…longish grass at a great distance…are pushing the ability of any digital system to record detail.

Sometime in the next week I hope to get back to a location where I took an image with the H9 which definitely showed the smear effect, and retake the image with the H50. If I can improve on what is here, I will revise this section.

Of course, the H50 now has three levels of selectable noise reduction. That will have to be tested. Watch for that comparison in part 3 of this review.

Revision, or addition at any rate. 5/24/08

Okay. Today I returned to Nubble Light to take similar images with the H50 to those I captured there a year ago with the H9. These H9 images were the first ones where i had reason to suspect that the camera was doing something strange in processing the image. My daughter, looking at 8×10 prints from one of the images said, “but it doesn’t look quite real…it looks almost painted.”

What follows are 1 to 1 crops from similar H9 and H50 images. I think they pretty much speak for themselves.

Overall Image Quality

(corner sharpness, resolution, color depth, etc.)

For your viewing pleasure here are a set of 6 images that you can view at full resolution to look for general image quality factors (like, but not limited to, the ones listed above). The first set were taken at the full wide setting of the zooms, the second set was taken at a rough equivilent of the longest tel on the A200 (70mm). The light was pretty dim, heavy overcast threatening rain, so we are working both the sensors and the lenses pretty hard here.

Again, you are free to draw your own conclusions, based on your own examination of the images. This is not quite a fair comparison. The A200 shot is noticeably soft in the corners, but it is at its maximum zoom, while the H9 and H50 zooms are not even up to the middle of their range.

The most surprising thing here is the notable improvement in corner sharpness between the wide angle on the H9 and the wide angle on the H50. You might almost think it was a new lens…or suspect that Zeiss has been tinkering with the design between models. Of course, the silght improvement in Purple Fringing and image smear between models certainly contributes the apparent sharpness in the corners.

Overall, i find the image quality of all three cameras to be very fine…it happens I prefer the image quality of the H50 over either of the others…but that’s just me. You may have a different take.

5/22/08

Some people did not like the flowers for this image quality comparison so I looked for another target with more consistent detail and a wider range of color. See what you make of this: all shots at ISO 100. First shots at wide angle. 2nd shots at 70mm equivalent (max zoom on A200). Close ups at wide angle. Close ups at 70mm equivalent.

I have to admit, it is very difficult to see significant differences in any of these shots displayed at 800×600 resolution. Looking at the images at 1 to 1 shows, unfortunately, just how bad the kit lens on A200 is at this aperture. The H50 actually shows the least chromatic aberration of three and has the best corner resolution by a visible margin. In my opinion the final close up of the $20 bill from the H50 is nothing short of brilliant! In this test, the H50, to my eye, is the clear winner, with the H9 following close behind.

The Portrait

The way a camera handles flesh tones is, of course, very important to people who take mostly people pictures…and that includes quite a majority. No comparison would be complete without a portrait set. These images were taken outside, in open shade, with white balance set to auto. There were two sets, a close up, and a medium distance image. I am the close up in screen resolution, and a 1 to 1 crop of detail around the eye from the medium distance images.

Again, to my eye (and it is only my opinion) on the detail comparison the H50 edges out the H9, and the A200. Granted, the A200 once more may be showing the effects of the kit lens, more than actual image quality. (I’d love to try this one with the Zeiss zoom…but no one will buy me one.) The comparison of the full frame close up is much harder to call…especially between the A200 and the H50.

Macro

One of the things I like most about the H9 is its macro ability. It focuses to 2 cm. 2 CM. I have to be careful or I get pollen on the lens. Once, while trying to take an image of a rainstorm, the lens even focused momentairly on a water drop on the surface of its own outer element. That’s close!

I wasn’t expecting real macro performance from the A200 kit lens, but, surprisingly, it has a decent macro…and one that is, in many ways, easier to use than the H series. The macro on the A200 is at the tel end of the lens, which give you a much better working distance from your subject.

Shots for comparison (especially for those who are in to bokeh (the quality of the out-of-focus background).

But the H50 goes closer.

One change that should be noted between the H9 and the H50 is that the normal focus range on the H50 now includes the macro range. (I always left my H9 in macro anyway.) The actual macro mode on the H50 is supposed to give priority to the closest object…to avoid those times when the camera decided to focus on something in the background. I have not tested this feature enough to know if it works.

One thing Sony did, for sure, fix…

Though I am not much of an action shooter, it always disappointed me that Sony did not address the obvious shortcomings of Action mode on the H9. Here was a case where a feature simply did not work the way Sony’s own materials, ads, website, and manuals, said it was supposed to. The highest shutter speed I ever squeezed out of Action mode was 1/640, while the promotional material touted 1/4000.

Tests shooting birds in flght with the H9 and the H50 side by side show that Action mode now works! The H50 gave a shutter speed of 1/2000 of a second in the same situation where the H9 struggled along at 1/500th. And, the predictive focus seems to work much better as well. I was able to lock on to birds in flight much more often and more consistently with the H50 than with the H9. One for you Sony! (But you should have fixed the H9!!)

The best I got out of the H9 (crop of detail)

H50 from the same spot.

New Features

(even when designed to fix a flaw in the H9)

There are several new features on the H50. I will cover those in Part 3.