O Greg Rogers αναλυει την ευκρινεια και την αναλυση..

takisot

Μέλος Δ.Σ.
Staff member
17 June 2006
23,923
Παλλήνη
Ενα εξαιρετικα επεξηγηματικο post του Greg Rogers στο AvsForum οπου αναλυει την ευκρινεια, την αναλυση, το MTF και διαφορες μεταξυ Ruby και dVision 1080p!



Ειναι ευκολα κατανοητο γιατι αυτος ο ανθρωπος ειναι ο καλυτερος reviewer προβολικων στον κοσμο! :cool:



Απολαυστε...


"The perception of image sharpness is generally determined by edge transitions. The more abrupt (faster) the edge transitions the sharper the image appears. The lack of sharpness is usually termed softness. Sharpness is a different quality than the ability to perceive detail, which is usually called resolution.

CRT projectors typically appear significantly softer (have less sharpness) than fixed pixel projectors because the brightness of the CRT electron beam spot falls off gradually (a gaussian-like distribution) toward its edges. The pixels on a fixed-pixel projector are basically discrete (have defined edges) hence they can inherently produce sharper image edges, i.e. better sharpness. However, if the discreteness (edge definition) of the pixels is reduced by lens optics, pixel device characteristics, or misalignment of multiple pixel devices the ability of the projector to produce sharpness is reduced. (Just as lens optics, CRT spot characteristics, and misalignment [mis-convergence] reduce CRT projector sharpness). Hence, there are sharpness variations amongst fixed-pixel projectors having the same native pixel density, just as there are sharpness variations amongst CRT projectors.

In the case of the dVision 1080p, the lens optics produced superb focus and virtually no chromatic aberration, which in addition to the discrete DLP mirror structure, and no misalignment of multiple DLP chips (since this is a single chip design), produces unprecedented image sharpness. The difference in perceived image sharpness between the dVision 1080p and the Ruby was not a close call. The difference was dramatic. The edges are sharper than on the Qualia 004 (which has better optics than the Ruby), which in turn is sharper than the Ruby. I devoted a special section (called Optical Performance) in my Ruby review to discussing its optical performance. Here is part of that section:

"However, edges are slightly softer on this projector [Ruby] than the Qualia 004. The gaps between individual pixels are more clearly visible on the Qualia, which is likely the result of a higher quality lens with a better MTF (modulation transfer function). The new, smaller panel may also be a factor. Regardless of the reason, high-definition images appear slightly softer and more like a CRT projector, although they are still considerably sharper than a CRT projector."
I might note that panel misalignment has an interesting effect on sharpness (and detail as well). If the panels are greatly misaligned (a pixel or more) it has LESS effect on sharpness and resolution because we then perceive the image as having discrete colored "ghost" lines. It looks terrible, but if you ignore that, it's not really affecting sharpness or resolution because we see multiple image lines. If the misalignment is greater than about 0.5 pixel but less than a pixel, we perceive color fringing, rather like chromatic aberration. It looks poor, and it does affect sharpness and detail, but we are more likely to complain about the color fringing effects. When the misalignment is sub-0.5 pixel we aren't likely to see color fringing from normal viewing distances but sharpness and detail perception suffer. I think the difficulty in aligning 1080p panels/chips on consumer products is going to make it unlikely that 3-chip/panel 1080p consumer projectors will achieve the sharpness/detail that is possible in single-chip 1080p projectors.

I'd like to discuss the artificial sharpness or false sharpness that arises when the pixel density of the projector is lower than required for the edge transitions determined by the video's rise and fall times, or when scaling or electronic detail enhancement or "sharpness" controls, add overshoot and ringing to the signals. But I don't have time to go into that now. However, those are examples of artificial or false sharpness that looks bad, and shouldn't be confused with real improvements in sharpness that arise from a higher native pixel density, superb optics, and electronics that avoid overshoot and ringing artifacts.

Resolution is a measure of the ability to differentiate spatial detail in images. But I normally try not to use the term resolution when describing image quality because it seems to confuse some people. Thanks to the computer industry, some people see the term resolution and immediately equate that to the native pixel format of the display. They seem to think that all 1920 x 1080 pixel displays have the same image resolution. But if resolution specifies the ability to perceive detail, then fixed-pixel or CRT projectors that display the same pixel format (i.e. 1920 x 1080 pixels) will likely have different resolutions. To make this more clear ... the resolution of two CRT projectors will likely be different, even if both projectors produce the same size images, using the same (1080p) scan rates. The resolution of two fixed-pixel projectors will likely be different, even though both have 1920 x 1080 pixel imaging devices.

The ability to perceive detail (resolution) gradually decreases as the size of the detail (increasing horizontal or vertical spatial frequency) is reduced. Therefore, we can best describe the ability to perceive spatial detail with a modulation transfer function (MTF), which measures contrast as spatial frequency is increased (rather like reducing the separation between a series of alternating black and white lines). The shape of the MTF curve (contrast vs frequency) is very different for CRT projectors than fixed-pixel projectors, because it usually depends more on CRT spot size and shape than lens quality. The MTF curve for fixed-pixel projectors is more dependent on lens quality and is ultimately limited by the native pixel format of the projector. The same physical factors that determine sharpness also determine resolution, but the perception of sharpness is mostly a function of the shape of the MTF curve, while resolution is normally defined as the frequency at which the contrast falls to some arbitrary level. For instance, the MTF curve for a CRT projector may start to fall off at a lower frequency but extend to a higher frequency than a fixed-pixel projector. As a consequence, the CRT projector may produce a higher ultimate resolution (finer detail) while producing less sharpness than fixed-pixel projector. In addition, larger detail may be much more visible on the fixed-pixel projector than the CRT projector because the slope of the MTF curves are very different. The same type of effects can occur between different fixed-pixel projectors because their MTF curves also vary depending on their optics and imaging devices.

In the Ruby review (Optical Performance section) I wrote the following about MTF contrast vs maximum spatial frequency (1920 pixels-per-picture-width) to compare the Ruby with CRT projectors:

"The AccuPel multiburst test pattern demonstrates that this projector can easily resolve the maximum 1920 pixels-per-picture-width resolution of the 1080i and 1080p video formats. Single-pixel black and white lines are displayed with exceptional contrast using digital signals, and nearly as well using analog signals. The multiburst lines are sharper and the contrast depth is much better than I have seen on a CRT projector."

The dVision 1080p produces even better MTF contrast in the maximum frequency pixel burst (1920 pixels-per-picture-width) than the Ruby, but its steeper MTF curve also improves the contrast for larger detail compared to the Ruby. For instance, the fine lines in the horizontal bar codes on soft drink containers were much clearer on the dVision 1080p than the Ruby during a 1080i D-Theater movie. However, the perceived difference in sharpness is even greater than the perceived difference in detail when comparing the dVision 1080p to the Ruby, although the difference in detail may become even more significant when higher resolution HD/Blu-ray DVDs are considered (they weren't out yet when this evaluation was done). "