DrPablo
Offline
• • •
Upload & Sell: Off
|
 p.13 #6 · Where does the 3D look come from? | |
pere marti wrote:
How could this be possible?
The ability of a lens to discriminate two details is completely and entirely dependent upon how different they look. Let's think in terms of RGB values for a second. If you take an array of alternating lines, in which half the lines are completely black (RGB value 0) and half are white (RGB value 255), the contrast is such that you can easily discriminate tiny details, for instance 200 lpm. On the other hand, take an array of low contrast alternating lines, say half are RGB value 150 and half are RGB value 152, then it might not be until 100 or 50 or 20 lpm that you can actually tell them apart from one another. It's like trying to read a book in the dark as opposed to in bright light. Higher resolution lenses are always higher contrast, which means they're more capable of discriminating low contrast details. I'll post an example later when I'm at home.
But fundamentally, when you get down to the level of the finest details, resolution and contrast are synonymous. When there is sufficient contrast between two things, they are resolved. When there is insufficient contrast between two things, they are not resolved. So when it comes to testing a lens, there is a direct correlation between subject contrast and resolution. The highest possible resolution of a lens is with the highest contrast target. As you drop target contrast, the resolution plummets.
...And then you spoke about that extraordinary contrast of Velvia film. Couldn't this be replicated by appropiate S contrast curves?
Imitated yes, but replicated no. This is because the S-curve is applied post-capture, whereas the contrast of Velvia is intrinsic to the capture. Velvia is inherently very high contrast, which makes it great at capturing low contrast subjects and terrible at capturing high contrast subjects. So as compared with a low contrast film like Astia, it will have more separation of midtone details, and Astia will have more compression of midtone details. These distinctions simply do not exist at the time of capture with digital cameras, which means your post-processing is beholden to whatever the sensor can capture. Along similar lines Velvia (and all films) capture light with an intrinsic curve. Digital sensors do not -- they do a linear transformation of the recorded light upon exposure, and then you apply whatever curve you want in post-processing. So the relationship with the light in the original scene is completely different than the relationship the film has. You can imitate the effect of Velvia using post-processing, but this is NOT a capture effect, it's a post-processing effect applied to a digital capture that has a fundamentally different exposure-response curve than Velvia. The good news is this is what allows digital cameras to be so flexible, whereas with film I need to use several different films to get the effects I want. On the other hand the digital sensor is not going to "see" the incoming light the way that film will.
Here's that harbor shot of mine again. Astia on top, Fortia (similar to Velvia but more saturated) on the bottom. The shots were taken about 20 seconds apart from one another with the same lens, only difference was a 1 stop exposure change to account for the different ISO speeds. You can see the intrinsic differences in contrast, color, and rendering of highlights, shadows, and midtones. In particular look at the building on the left. The Astia image wonderfully captures the midtones, the Fortia image stretches it into highlights and shadows (and adds perhaps unneccessary drama). These are intrinsic light-response relationships, not processing effects from a common starting point.
The difference with digital is you would get one baseline RAW image, with all its intrinsic qualities and delimitors, and then stretch and compress various parts of it to make it look the way you want.
http://www.pbase.com/drpablo74/image/78608487.jpg
http://www.pbase.com/drpablo74/image/78608542.jpg
Edited by DrPablo on May 12, 2007 at 09:58 AM GMT
|
