PDA

View Full Version : Digital Surprise



Ken Lee
7-Dec-2004, 17:45
Another one of the more "counter-intuitive" elements of digital imaging: In a nutshell, the sharpest lens aperture does not always deliver the sharpest end result. I admit it: this is not directly related to Large Format, except for those using digital backs, for whom it may prove very helpful.



I am definitely not trying to start the one-zillionth trivial discussion on digital "versus" film, but I thought some would find this page (http://www.photo.net/learn/optics/pixel_peep/pixel_peep_part2.html" target="_blank) interesting.

paulr
7-Dec-2004, 19:49
The sharper aperture does give a sharper result. The problems caused by the sharp aperture in that example are from aliasing--added noise from pattern interference. But the edges are much more clearly defined, and there's more resolved detail, too. did you see the whole attached article?

Ken Lee
7-Dec-2004, 20:27
Thank you, you make an excellent point.

What I should have written, was that contrary to what we have grown accustomed in the world of analog photography, the sharpest aperture does not always deliver the best image quality. I suspect that many would find the color fringing and bleeding to be unacceptable, even if the image appears sharper.

Bruce Watson
7-Dec-2004, 20:42
Some of this will be solved by lens design optimized for digital sensors - the sensors don't seem to like light coming to them at angles, for example. Some of this will be solved by better algorithms to handle the peculiarities of the sensor matrix.

None of which means jack to me, until they can get the weight down. If they can give me a full frame 4x5 sensor with the entire capture system weighing the same as 10 film holders or less, then maybe I'll think about out. Until then, if I've got to hike with it on my back, it's going to be film.

paulr
7-Dec-2004, 20:49
That's definitely true.
These digital problems are new ones ... though the idea that the fuzzier lens might give better overall picture has had other incarnations over the years. In portrait photography, in work that's trying to be symbolistic or impressionistic, or in any case where someone wants to either suppress the flaws of the real world or to emphasize the broad strokes over the details. The vintage lens crowd has always fallen under this umbrella.

Glenn Kroeger
7-Dec-2004, 23:12
Hogarth:

They are never going to give you a full frame 4x5 single-shot system. They are going to give you a 40x50mm system that produces superior image quality (notice I didn't necessarily say more resolution or detail) but as the Bayer matrix reaches 35-40MP the overall image quality will arrive. Lack of grain and smooth tonal transition will rival 4x5. I am assuming color.

The hard part is seeing what the camera and lens systems will be that replace our current LF. Our wide angle lenses are great on film because they aren't retrofocus... but digital sensors prefer retrofocus designs to manage incidence angle issues. But the optical designers need to revisit lateral CA issues with many of these lenses. Movements will be correspondingly smaller so a mechanism to evaluate focus becomes a significant issue. I don't see any manufacturers that are really envisioning what a digital solution will look like in 10 years. I don't think its a 35mm SLR with a digital chip stuffed in the film plane.

Steven Dusk
8-Dec-2004, 04:23
Glenn, extreme wide angle lenses e.g. Rodenstock 55mm are indeed retrofocus to allow some flexibility for movements. While I agree that there will probably never be a 4"x5" sensor produced, a part of me hopes for one around 90mm x 90mm to be used on a baby Arca!

ADG
8-Dec-2004, 05:12
This may seem quaint, and I am not really this nieve,,, but when digital first appeared, i really thought that by 2005 I could go down to by local electronics warehouse and get a big piece of 'sensor' and glue it to a darkslide.
Solar panels were initially very expensive and this is where they have ended up.

Reality has turned out to be somewhat different (again)

Glenn Kroeger
8-Dec-2004, 06:40
Steve:

Yes, I dreamed of a 60x90mm sensor for an Arca 6x9 with 54MP...but alas I suspect the demand simply cannot justify the capital to develop nor the cash flow to purchase such a chip. Having been stunned by the quality that can come from a 22MP 37x49mm back, I can't imagine any commercial application that demands a larger chip. When the 36x48mm chips reach the pixel density of the new Canon 1Ds MkII, we will have 35MP.

For landscape work, there are big advantages. If I am realistic about my 4x5 work, the limitaiton in sharpness and detail is usually controlled by DOF and wind movement, not lens performance. A small sensor operating at an effective and virtually noiseless ISO 200 has some advantages.

paulr
8-Dec-2004, 06:42
"Reality has turned out to be somewhat different (again)"

and what ever happened to those moving sidewalks and rocket cars we were all promised?

Leonard Evens
8-Dec-2004, 07:49
That was an interesting article. It also may explain something I had noted about my scans. I use an Epson 3200 scanner, and I've been getting some color fringing at very high magnification. It is not a problem at magnfications I might actually use, but I wondered about it. The Epson scanners have staggered sensor arrays which supposedly act to minimize aliasing, but of course they can't eliminate it. Since I am scanning the negative and not the direct image from the lens, the situation is not exactly the same, but it seems reasonable that the color fringing I'm seeing is an aliasing effect. The Nyquist limit for my scanner is about 63 lp/mm, and it seems reasonable that my negatives may contain some detail at spatial frequencies above that. Of course, the optics of the scanner can't deliver much more than half of that, so anything happening around the Nyquist limit is moot.

paulr
8-Dec-2004, 09:44
I wonder if this is specifically why flatbed scanners tend to have double or so the digital resolution of their real optical resolution. It lets the scanner makers avoid this whole can of worms.