| Video Noise January 2008
 When a Reference Isn't a Reference
I’ve just spent 25 hours trying to find any real-world
video content that confirmed several failed tests using a popular evaluation disc: the HQV
Benchmark test DVD from Silicon Optix. This disc is used by many reviewers to evaluate
how well video displays, disc players, and video processors handle various aspects of DVD
reproduction; many products have been praised or panned based, at least in part, on how
well they performed on HQV Benchmark.
I spent so long trying to find any material that provided
real-world confirmation of HQV Benchmark’s results because I’d been using
an incredible 60" plasma display for a month and thought it had been doing a
remarkable job of upconverting 480i standard-definition DVDs to the high-definition spec
of 1080p. But when I got around to evaluating this amazing (and expensive) display with HQV
Benchmark, it didn’t fare as well as I’d expected it too. It took longer
than necessary to lock on to speedway grandstands and eliminate moiré patterns, it
didn’t completely eliminate the jaggie edges in two of the three "fingers"
in the moving jaggies test, and it didn’t master unusual cadence changes. This
high-end display also miserably failed the horizontal text crawl presented on HQV
Benchmark as typical of the text crawls commonly seen in broadcast video. The text was
legible, but was missing half the horizontal lines; the characters, instead of being
solid, looked as if forced through a screen door. Yet in viewing many hours of real-world
movies and standard-def programming, I’d never once seen any evidence that this
display exhibited any of the processing problems revealed by HQV Benchmark.
So I went hunting for problems. I watched hours of crawling
text from news channels, sports events, entertainment "news" programs -- you
name it. Local channels, cable/satellite channels, broadcast, recorded, live -- none ever
produced anything but perfect horizontal crawling text on the display. Which made me
wonder how any reviewer could use the results of the HQV Benchmark test as any
indication of how a display will perform with real-world programming.
When I then went looking for all the motion- and
cadence-related failures I was expecting to see, based on what had been revealed by HQV
Benchmark, the same thing happened. I could find no examples of these problems, even
with such infamous DVDs as Star Trek: First Contact and Gladiator. I also
investigated, in great detail, other, less notorious DVDs, such as the difficult edges of
airplane wings in Jurassic Park III, again with no evidence that this
display’s processing was anything less than perfect with real-world source material.
I must conclude that HQV Benchmark contains signals
that rarely, if ever, are found in real-world DVDs or standard-definition programming.
Therefore, using this disc as a benchmark is probably inappropriate, and certainly
misleading.
I also have an A/V receiver that includes a Silicon Optix
Reon HQV processor. I re-ran the HQV Benchmark tests, running the signal from the
disc player to the AVR, and from the AVR to the display. Surprise, surprise! The HQV
processor in the AVR sailed through every test, performing with near perfection. The only
better results I have ever seen with HQV Benchmark came from a Silicon Optix Realta
HQV processor, the Reon processor’s more expensive sibling. Imagine that -- the
company that makes the test disc also makes the processors that pass every test with ease.
I don’t want to denigrate the HQV processors at all --
they are clearly very good processors. However, my advice is to ignore any criticisms of
video displays based solely on tests conducted with HQV Benchmark. If the review
provides no confirmation of the HQV Benchmark test results with observations of the
same problems seen when displaying real-world source material, then I’d have to call
that review inconclusive. Maybe the product is OK, maybe it’s not. HQV Benchmark
is interesting, but perhaps not applicable to the real world.
...Doug Blackburn
db@hometheatersound.com |
