Previous: Analysis of grain structure< > Next: Per channel scan performance

Index of this page

• Test results
  • Evaluation of resolution / resolving power and focus

Evaluation of resolution / resolving power and focus

The images below are crops of the full 8000 ppi resolution scans. As can be seen from these images, I think I got the focus of the chart spot on. The scans may look a bit soft, but remember: they are unsharpened and at "actual pixels" level, you are looking at a significant enlargement compared to the original test chart (see topic discussed here). In addition, they are a bit "over-scanned", a 4000 ppi scanning resolution would have been more realistic in terms of matching true captured film detail and the Alpha 900. I have therefor also included downscaled samples at 4525 ppi. At that resolution, these samples closely match the Alpha 900's image size. These last samples have received a bit of sharpening as well after the downscaling.

The reason why the original 8000 ppi scans appear "bigger" than the Alpha 900 shot, is because they ARE bigger. At 8000 ppi, these images are about 7100x10500 pixels (depending on exact scanned film area), so approximately 75Mpixel, triple the amount of the Alpha 900. Also notice the incredibly "clean", I would be tempted to say almost sterile, look of the Alpha 900 shot. At 100 ISO, pixel noise is pretty much absent in this digital shot, something that can not be said of higher ISO shots with this camera, according to some of the first reviews I have seen of the Alpha 900.

Notice that, contrary to the first public release of this test using a poorly calibrated Imacon 646 scanner, the new scan results clearly show that all films are easily capable of reaching the 2000 lines-per-picture-height mark, which translates to about 43 lp/mm. Since this is the highest resolution line pattern on this particular chart, all films and the Alpha 900 therefor outresolve the chart. In hindsight, it would have been better to a use a chart containing higher resolution line patterns, preferably up to at least 100 lp/mm. Still, the included photo samples on the chart do give some insight in the true resolving power of all the films and the Alpha 900. Read on for that!

Kodak TMax 100, 8000 ppi	Kodak TMax 100, 8000 ppi

Kodak TMax 100, 4525 ppi to match Alpha 900	Kodak TMax 100, 4525 ppi to match Alpha 900

Kodak Portra 160VC	Kodak Portra 160VC

Kodak Portra 160VC	Kodak Portra 160VC

Kodak Ektar 100	Kodak Ektar 100

Kodak Ektar 100	Kodak Ektar 100

Fuji Velvia 100	Fuji Velvia 100

Fuji Velvia 100	Fuji Velvia 100

Alpha 900	Alpha 900

Now viewing images at "actual pixels" tells already a lot, but not the whole story. By enlarging the line patterns to 800% in Photoshop, we are able to see the individual pixels. In the images below, you can see the end of the vertical line pattern that represents the 2000 lines-per-picture-height resolution. This resolution translates to about 40 lp/mm. Note that there are two series of images, the first ones represent the full 8000 ppi resolution samples, the second series the downsized images that match the Alpha 900's resolution.

Now the first thing you should notice is that, what appeared to be pretty solid black lines in all of the above shown pictures, turn out to be rather low contrast pixelated line and grain patterns in reality! Our human eyes, and the human brain, deceive us when looking at the "actual pixels" 100% images, and translate the repeating pattern of darker and lighter areas in a pretty solid "white / black" transition. Contrary to the first public release of this test with scan results of a badly calibrated "backup" Imacon 646 scanner, that wasn't capable of showing that the color negatives films in this test resolve the 40 lp/mm highest resolution line pattern on the test chart, the new scans of the overhauled Imacon 848 do show that all films are capable of resolving the 40 lp/mm line pattern with ease.

Lastly, notice the extremely clean image of the Alpha 900, clearly there is definition and contrast enough left for even higher resolution figures. The Alpha 900 gets very close to it's theoretical limit in combination with this lens. If we consider that it has a sensor of 4032 x 6048 pixels, and if each pixel row actually counted, than the maximum resolving power of the sensor would be (4032 / 2) / 24 mm = 84 lp/mm. With 40lp/mm guaranteed and room left for resolving additional image detail, that's not bad!

Kodak TMax 100	Kodak Portra 160VC	Kodak Ektar 100	Fuji Velvia 100	Sony Alpha 900

Downsized images that match the Alpha 900's resolution:

Kodak TMax 100	Kodak Portra 160VC	Kodak Ektar 100	Fuji Velvia 100	Sony Alpha 900

However, to put these 800% images in a more realistic context, I show you a 1200 ppi direct flatbed scan of the original test chart. So this is NOT a scan of a negative or slide, but a direct scan of the test chart itself! As you can clearly see from this image, the lines that were being photographed in these shots, aren't perfectly hard edged either in reality. In fact, you can see the offset printing pattern deteriorating the line definition. Be aware though, that although you might be fooled by this giant enlargement into thinking the print quality of this test chart was particularly bad, it actually looks sharp and perfectly normal in reality.

Canon 9950F direct scan of test chart Notice the offset printing screen pattern being visible in this image, and the overall "fuzziness" of the lines themselves as a consequence of this. Please notice that the actual real world size of this printing / line pattern is no more than just about 5 mm wide!

Previous: Analysis of grain structure< > Next: Per channel scan performance