PDA

View Full Version : is the limit the sensor or the lens?


montreal
31st of May 2005 (Tue), 21:40
Hi,

Not sure this can be answered but here I go anyway. Two questions:

1- I know that any image will be less sharp when blown up on screen (especially if it's blown up a lot). But let's say, hypothetically, that my handling of the camera is flawless (which it isn't, of course). How can I tell the difference between unsharpness caused by reaching the limit of the sensor vs unsharpness caused by reaching the limit of the lens?

2- I read on this forum that pretty much anything that's taken with a digital camera has to be sharpened in post-processing. Is that true? If so what is the reason for that?

I have a 300D (Rebel), 18-55 and 70-200L.

rdenney
31st of May 2005 (Tue), 22:24
Hi,

Not sure this can be answered but here I go anyway. Two questions:

1- I know that any image will be less sharp when blown up on screen (especially if it's blown up a lot). But let's say, hypothetically, that my handling of the camera is flawless (which it isn't, of course). How can I tell the difference between unsharpness caused by reaching the limit of the sensor vs unsharpness caused by reaching the limit of the lens?

If the unsharpness extends beyond about two or three pixels, it's definitely the lens or your technique causing it. If the unsharpness is less than that, it might be the lens or the sensor. If the blurriness is noticeably different at the edges and in the corners than in the center, it's probably the lens. If it's worse at the widest apertures, it's definitely lens faults, and if it worse at the smallest apertures, it's definitely diffraction (which is incurable and not a sign of fault except choosing too small an aperture).

Sensors have an anti-aliasing filter that reduces sharpness at the pixel level slightly to prevent interference patterns between the pixel grid and fine patterns in the image. At some point, pixel density will increase to the point where all lenses will be blurry enough to obscure any patterns that might interfere with the grid patter and then no anti-aliasing filter will be needed. A little bit of unsharp masking will restore the appearance of sharpness at the near-pixel level.

Note that the slight blurriness caused by the antialiasing filter is not likely to be visible when printed at 240 pixels/inch or more. Viewing the image at full resolution on a computer monitor is about 100 pixel/inch--not nearly enough to hide that fuzziness. A 300D or 10D will make an 8x12 print at 240 pixels/inch, so up to that size any apparent fuzziness is mostly likely a lens or technique fault. Larger prints make still look good, but they won't be the best.

Rick "who rarely got prints from 35mm bigger than 8x12 that were really good, especially after getting used to larger formats" Denney

montreal
1st of June 2005 (Wed), 13:25
Wow, thanks Rick!

That pretty much answers it all.

After a little more research I suspect that my LCD screen is another bottleneck... I'm not sure how to know it for sure but I suspect it's one of those bogus 16.2 millions colors (6 bits), not true 16.7 million colors (8 bits).

The colors are much more saturated when I look at the images at work... Although I don't think that monitor is calibrated very well... but then again neither is mine.

ron chappel
1st of June 2005 (Wed), 20:27
Some thoughts-

DSLR's give blury images straight from the sensor ,mostly because of the AA filter as Rick explained and also because of the bayer colour sensor.
(note-the bayer filter doesn't interpolate every pixel- as many tend to think.
The bayer pattern interpolates ONLY the colour data .....this does effect the sharpness of the image slightly but not even remotely to the extent that fovean would have us believe)

The reason most cameras are set by default to give slightly unsharp images is because they are meant for enthusiast use where it's expected the user would prefer to have complete control over the sharpening process rather than leaving it to the camera.
Once sharpening is done it can't be reversed!

montreal
1st of June 2005 (Wed), 22:31
Thanks, this is exactly the kind of info I was looking for. I'm actually gonna print this post.:)

Pekka
2nd of June 2005 (Thu), 05:03
See also http://photography-on-the.net/forum/showthread.php?t=2849

Andy_T
2nd of June 2005 (Thu), 05:51
Pekka,

you are right, that thread ought to be compulsory reading :lol:
It's vital to successfully using and understanding a DSLR.

Best regards,
Andy

Tom W
2nd of June 2005 (Thu), 06:01
Pekka,

you are right, that thread ought to be compulsory reading :lol:
It's vital to successfully using and understanding a DSLR.

Best regards,
Andy

It should be mandatory only if one understands that diffraction is not the only factor to be considered when stopping down. It's gradually increasing effect as you stop down must be weighed against the increased depth-of-field that stopping down gives you along with the improved corner sharpness of some wide lenses.

If you shoot macro, f/22 is a very necessary aperture setting sometimes. Diffraction does come into play at that setting, but the alternative of not getting the entire plant/bug/object in focus is a much less desirable outcome.

montreal
2nd of June 2005 (Thu), 09:15
So basically, to be sure I understood... on some non-SLRs, the sensor resolution exceeds the maximum resolution of the lens at certain apertures?

Tom W
2nd of June 2005 (Thu), 09:28
Yes, depending on the lens and the part of the lens that you're using. Most lenses, and especially wide-angle lenses have a sharpness advantage in center frame that diminishes as you move farther away from the center. So, the lens may outresolve the sensor in the center and not at the corners.

montreal
2nd of June 2005 (Thu), 09:32
So much to learn, so much to learn...

I'm not quite sure how to apply this knowledge to my 300d.

DwightMcCann
2nd of June 2005 (Thu), 17:07
So, we should all get 1DsMII's, eh?