The digital zoom heresy (or Why a limit on digital zooms?)

[normc]

Your in good company.... and I like your test. For myself I can sometimes get better quality using digital zoom by lowering rhe resolution on the camera I am sort of guessing that the camera shake or vibration is not as much of a problem?

[BillDrew]

Daniel T wrote:

This thread caught my interest, so I decided to perform a real-world test of my Canon A510's digital zoom. I used my camera's highest resolution and lowest compression settings, and formy subject I chose a page from a book (black text on a soft white background) at a distance of 10 feet. Shot #1 was taken using only the 4x optical zoom, and Shot #2 included the camera's maximum digital zoom for a total of 13x.

I downloaded both images onto my PC and compared them side-by-side. I displayed Shot #2 (the digitally zoomed image) at100%, and I used my image-editing program to zoom Shot #1 (the optical-only image)4x so that it's text was approximately the same size as Shot #2, thus basically performing the crop on my computer rather than in the camera.

Results: Image #2 (digitally zoomed) was a bit darker overall, but it showed noticeably better detail. Also, there werefar fewer digital artifacts (from in-camera sharpening?) along the edges of each character of text. I also noticed that the pixels on Image #2 were much smaller when comparing the two images in this manner (and I assume that my camera created the extra pixels via interpolation).

I have to admit I was surprised by the results. Prior to this test I had just assumed that the digital zoom was merely an in-camera crop that would only be useful for those without computers or editing software.

However, I'm just an amateur digital photographer, sothese results are hardly definitive. Perhaps somebody who knows what they are doing and has better-trained eyes would care to repeat this test under more controlled conditions?

I think you are on the right track, but need to do one more step to make a good comparison - resize the cropped image to the same pixel count as the digital zoomed image. Try several different resizing techniques - IrfanView (www.irfanview.com - freeware you should have) offers six. If you do that, I think you will find the two images are pretty much the same, or the cropped version better since you can use much more computer intensive resampling techniques than can be fit in a camera.

[Alan T]

One of the numerous reasons Irecently bought a Casio EX-Z750 as a replacement for my Casio QV-5700 (which I have kept) was its 7Mpixels, compared with the QV-5700's 5Mpix. This means I can zoom digitally on the computerOR in the camera, and maintain quality rather better. I lost a whole stop of lens aperture, though (f/2.8 instead of f/2), and that's why I kept the QV-5700.

As a former OM-10 SLR owner with 35-80 and 80-210mm lenses, I've always felt the need for a longer zoom, but I also want a compact digicam. I find 5Mpix is adequate most of the time, and I do really big prints with reasonable quality, so I have my new 7Mpix camera set most of the time to 5Mpix. This makes all post-processing and storage on my antique 400MHz Pentium II a lot quicker, because the files are about 1600KB instead of 2000KB. I have the digital zoom on the Z750 switched off, usually, but it's there when I want it. Times when I use it are when memory card space & battery power are at a premium, e.g., when in remote locations, or when I forget to take the charger on holiday.

So I still tell people digital zoom is useless, but don't fully believe that any more.

Happy shooting!

[Daniel T]

BillDrew wrote:

I think you are on the right track, but need to do one more step to make a good comparison - resize the cropped image to the same pixel count as the digital zoomed image. Try several different resizing techniques - IrfanView (http://www.irfanview.com - freeware you should have) offers six. If you do that, I think you will find the two images are pretty much the same, or the cropped version better since you can use much more computer intensive resampling techniques than can be fit in a camera.

Thanks,Bill, I took your advice. I decided to re-run the test, but with one change: I turned off (or reduced, I'm not sure which) in-camera sharpening by choosing the "Special Effects: Low Sharpening" option. The edge artifacts are no longer visible in any of the photos, and that took care of the worst problem of the previous test. I then repeated the test and I did as you suggested by resampling the cropped image so that it would have the same pixel count as the digitally-zoomed one. I used irfanview and tried all six resampling methods.

New conclusion: The best resamples for this subject matter were done with the Mitchell and the Bell methods, and both came out pretty darned close to the image that my camera interpolated when I set the digital zoom to 13x. It was basically a draw, although the camera-interpolated image may have been a tiny bit better in some spots. Certainly it was no worse.

I guess I'll continue to try out the digital zoom in different situations. So far it doesn't seem to have any noticeable downside except for the fact that I'm going to have to use a tripod more often.