"Interpolation" by digital backs... is it really the image?greenspun.com : LUSENET : Large format photography : One Thread
Was talking with a criminal attorney the other day and he was paying me for information on photography in preparing for a cross examination. Seems the 'good guys' took a number of photos & scanned them at the highest resolution (interpolated) with a scanner & then made really nice courtroom display prints. Naturally, they made his client look really bad. I pointed out to him that these were all "interpolated" prints, not exactly what was on the negative. It worked well enough for him to bring it up in court with the 'expert witness' the cops & DA had and got him so flustered & confused trying to explain 'interpolated' that he planted enough doubt about the evidence that his client walked. Seems the jury (interviewed afterward) didn't like the fact that the digital prints were not what was on film & 'interpolated' meant to them that something must have been added that wasn't there in the first place. So, those who use the interpolation programs, just what does it add?
-- Dan Smith (email@example.com), December 04, 2001
Oversimplified, but clones more pixels averaged to nearest neighbor. Right? GeeZ Dan hope the bad guy stays in Utah. Nevada's right next door. If he's a real wacko he'll go to California. J
-- Jim Galli (firstname.lastname@example.org), December 05, 2001.
It depends. Some interpolation schemes help you get the maximum possible resolution out of the scanning hardware, others are pure marketing hype and actually make the picture look worse.
Trust your eyes and not the hired expert.
-- Struan Gray (email@example.com), December 05, 2001.
As Struan says; it depends on the interpolation algorithm.
In its simplest form, interpolation simply puts another pixel between two adjacent ones, thus doubling the linear 'resolution'. The colour or greyscale value of the newly created pixel is derived from those adjacent to it by simple averaging, or by more complex bi-linear or bi-cubic statistical methods.
Of course, the recreated pixel doesn't add any real information, but just smooths out the existing samples to give a more acceptable image at a larger scale. It's a bit like using a finer grain film: It doesn't make your lens any sharper, but it gets the best out of what's already there.
In some cases the results can be surprising, turning a Piccassoesque cubist nightmare into a Rembrandt. (which is a good thing, in my book)
There's really nothing innately sinister, underhand or devious about interpolation, unlike the legal profession, it seems.
-- Pete Andrews (firstname.lastname@example.org), December 05, 2001.
Pete said, "It's a bit like using a finer grain film: It doesn't make your lens any sharper, but it gets the best out of what's already there."
My finer grain film doesn't use an algorithm to add something that wasn't there to be begin with. Interpolation software does EXACTLY that; it makes a "best guess" to add FALSE information to the image. While I agree there is nothing inherently sinister about it, interpolation is still faking it. If you consider the software in use for interpolation, you should see that the software will and does serve its purpose well when dealing with continuous tone images, but what would happen when trying to "increase resolution" of an image of a multicolor LINE DRAWING that exceeds the resolution of the device being used to capture the image? This is where "faking it" fails.
-- Chad Jarvis (email@example.com), December 05, 2001.
And what happens when you use a film that hasn't got a fine enough grain to capture all the detail in an image? For goodness sake Chad, it was just a simile.
It's about time the mention of anything digital didn't raise a howl of protest on these photographic forums.
Climb on the digital bandwagon, or stick your head in the sand, as you wish, but don't nit-pick just because you don't like the subject of the thread.
-- Pete Andrews (firstname.lastname@example.org), December 05, 2001.
I'm not protesting. Actually, being a computer geek of sorts for the last 20 years, I fully understand and respect the technology. I was just pointing out where interpolation falls short, and I don't think technology (or lack thereof) makes the photographer.
-- Chad Jarvis (email@example.com), December 05, 2001.
Interpolation doesn't _add_ anything to an image, but it can _subtract_ thru a type of statistical error known as aliasing. The classic example is to take a (chemical) picture of a suspension bridge from a very narrow angle along the length of the bridge, such that the suspension cables appear to _almost_ blur into a mass of solid gray. Then scan the picture. Depending on the resolution of the scanner, the interpolation algorithm, and the distance between the cables, you will sometimes be able to see individual cables, sometimes you will see a solid wall of steel.
A very simplified explanation is that if the detail falls _between_ the pixels, the interpolation will cover it up with whatever color falls _on_ the pixels. A more technical explanation is that the higher frequencies are masked by the lower.
Kinda tricky to explain over email, if anybody needs more info just drop me an email.
-- Mike Kelleghan (firstname.lastname@example.org), December 05, 2001.
We used to make "internegatives" when I was a kit. You take a 35mm negative with grain at 8X10, make a 4X5 negative from it, and make a 8X10 print from that and the picture lookes clearer. But the improtant thing is you still have the same amount of information. If a sign was too small to read in the origional picture, you still won't be able to read it after makeing an internegative or "interpolating".
The problem is not "interpolating" or "digital zoom" or for that mater "focus free" cameras. It is that fact that these sorts of things are used to confuse buyers. I don't know why, because although you will have to pry my film from my cold dead fingers, Digital is the right choice for over 90% of Americans.
Aside: Is it ethical for a triditional wedding photographer to use a Medium format camera with a 35mm back?
-- Neal Shields (Shields@ftw.com), December 05, 2001.