Originally Posted by BasicBlak
I'm referring to a failure of basic performance parameters (like 3:2 or even 2:2, for instance) that often get shrugged off as irrelevant to real world conditions. So called enhancements are another matter entirely, but they too can sometimes apply.
In general, I can't say I necessarily disagree with any of your points. However, if one purchases enough software, one will eventually come across real-world items that pose any one of many "problems" that may have manifested during a test-disc phase, TV enhancements notwithstanding. To me, I'd think that would be the whole point in a review pointing these issues out in the first place. If anomalies that are discovered during a review/test didn't matter, there'd be no reason to do a review at all, let alone any of us paying any attention to them. Seems to me that many of us make purchase decisions more or less based on how well a given TV performs ultimately in these reviews; consciously or not, we often buy things based on performance reviews (along with, of course, manufacturer reputation, and/or personal preference). We want the TV, DVD player, etc with the fewest test disc issues in case said issues do
eventually show up during our ownership, which is very probable if, like I said, you watch enough material. So in my personal opinion, a product failing an assortment of benchmarks during testing does
potentially matter with real-world material and deserves to be taken into consideration. To that end, I feel SonyCrusader's original statement is quite valid. Just my two cents....
I dont disagree with what you're saying either, I actually intended to quote SonyCrusader directly and quoted your response to him by accident. But he specifically used ME's behavior on a test pattern to make his point, and I think applying ME in to a motion resolution pattern to be a particularly bad way to measure its effects.
The reason why ME will fail miserably specifically on a motion resolution test is that its basically a worst case scenario for ME that doesnt exist in the real world at all. Motion resolution patterns have distinct alternating pixel structures that are rarely if ever found in video. ME has to measure the motion of objects in the video, and in order to accurately smooth out a motion resolution pattern and keep the pattern intact, it would have to do this on a pixel by pixel basis, or else the exactly pattern and thus the measurement is going to be destroyed. But such an algorithm can hardly work on a pixel by pixel basis, or else basic video noise would cause serious artifacts on real world video.
Just as video compression relies on analyzing blocks of pixels to determine motion and save on data, interpolation essentially does the same thing in reverse. Which works reasonably well on real video, but can cause artifacts in some scenarios.
But because a worst case scenario shows up on a test pattern, thats no reason to dismiss it outright. Another example - if one were to take a sharpness pattern on a DVD, and display it on a HDTV, the result would be blurred lines rather than sharp edges, due to the interpolated pixels in the scaling. Technically, the pattern is being displayed incorrectly. To do so would require point sampling, blowing up the pixels into larger squares rather than blending them. But if one were to apply point sampling to video, it would make you cringe just to watch it.
Still technically a flaw perhaps, but a desirable flaw in the real world.
ME may have it's issues, but I doubt future interpolation algorithms will be able to improve it much further. Its still basically creating something out of nothing, so it will never look as good as real high frame rate video.
Although I still think a large part of the dissatisfaction with interpolation on these forums is due to the association of high frame rate video with low quality programming such as soap operas, and I dont think that will ever change, no matter how good the interpolation becomes. Not until film is shot at a native 60fps or more will interpolated video actually seem like an improvment to most here.