How We Assess Discs
What the heck are we talking about?
by Dan Ramer
Feb 23, 2007


When we asked for your opinions concerning our coverage of high definition discs, some readers took the opportunity to include more general comments about the site.  Among them were comments about the way we assess the quality of a disc’s presentation; those messages tended to fall into two distinct groups.  The first group questioned the validity of reviewing discs on high-end equipment that does not represent typical installations.  The second group wasn’t completely sure they understood what we were describing with some of the terms we use to characterize the presentation.  In a previous column, I promised that I'd offer a few words of explanation. 

The Equipment

Let me begin with the thought that a great transfer will look good on all manner of displays, but a flawed transfer may only appear mediocre on a very fine display.  So to serve all our readers - you own a broad quality range of displays - we must avoid using equipment that might mask problems.

What are the display characteristics that affect the image?  To state the obvious, size matters.  The larger the image, the more detail is revealed.  So just as a magnifying glass reveals more detail than the naked eye, so does a large image reveal more about the quality of the video than a small display.  Large images lend themselves to a larger viewing angle (the apparent width of the screen from your viewing position).  To simulate a large viewing angle for you, I sometimes capture an image from a disc, crop it, and magnify it to demonstrate what I see on an eight-foot wide screen.

A display's technology, quality, and setup matter as well.  A comprehensive discussion of all display types exceeds the scope of this article, but I’ll offer a couple of examples.  A direct view CRT display, a technology rapidly vanishing but still a huge segment of the installed base, has unique limitations.  Look closely at the face of the display and you’ll find that the picture tube has patterns of red, green, and blue phosphors.  The size of those small dots or stripes limits the resolution of the display.  In a similar fashion, fixed pixel devices, like plasma or LCD flat panels, might be limited by the number of pixels in their structures.  Displays usually have a sharpness adjustment, an artificial process that is supposed to create the illusion of sharpness but all too often simply add halos (explained below).  And a CRT display is likely to have scan velocity modulation active, which is a similar artificial process that also generates the same visual flaws. 

And then there is the issue of proper setup.  Without some kind of adjustment guide (like a test disc or the THX Optimizer), such simple things as brightness, contrast, tint, and color intensity may be off.  More complex adjustments like maintaining neutral grays throughout the video dynamic range (color temperature), also affect presentation, but it would be prudent to leave that specific adjustment to a trained technician with proper instrumentation. 

All of those display characteristics can mask or hide problems found in a disc’s transfer.  For example, if the brightness is too high, black levels will suffer.  If the brightness is too low, shadow detail will suffer.  (I’ll get to those terms soon.)

Similarly, the sound system is affected by the quality of the electronic chain (distortion, frequency response, available power), the quality of the speakers (distortion, frequency response, power handling), and the acoustics of the room.  With regard to the last, thousands of words have been published just on the subject of speaker room placement and placement of the subwoofer in particular.  That really exceeds the scope of this article.  Suffice it to say that a revealing sound system will not mask audible problems.

So the better the equipment used to review a disc, the more accurate the video and audio critique is likely to be.  When you upgrade to better components, you will see the strengths and weaknesses we describe.  And that seems like a perfect opportunity to transition to terminology.   

The Video

Sharpness may be expressed as small object detail and finely grained textures.  Small object detail relates to your ability to clearly perceive objects that were some distance from the camera lens.  So if you can recognize faces and perceive facial expressions in long shots, then small object detail is very good.  The term finely grained textures refers to your ability to see such little details as the weave of fabrics or skin pores.  Sharpness may be reduced by video filtering during telecine (the digitization of the original film) or with compression tools; such filtering reduces the size of the compressed digital video file on the disc, allowing more content to fit.  HD has the advantage of more available pixels, so unless something is wrong, HD will always have better detail and sharpness than standard definition.

Color accuracy may be discussed in terms of color intensity, hue, and smearing.  Intensity relates to how convincingly bright, primary colors are reproduced.  They should be vivid without seeming iridescent.  Hue may be best judged by the naturalness of flesh tones, something we all observe on a daily basis.  Smearing relates to a relatively rare phenomenon when color bleeds outside of the boundaries of an object.

Edge halos are the bright or dark outlines that accompany and are adjacent to high contrast transitions in the video, like the edge of a dark object against a bright background.  This cropped and magnified screencap is from Hidalgo; the halos are obvious (and there’s also a bit of mosquito noise - see below).  Halos are caused by two mechanisms: MPEG-2 compression and applied edge enhancement.  The second is now rare but may be identified by asymmetrical halos.  For example, if vertically oriented edges have worse halos than horizontally oriented edges, it is likely an indication of horizontal edge enhancement.  Halos impart an artificial, very unfilm-like appearance to the video.



 

The compressionist can diminish halos by manipulating parameters.  For example, compare these two cropped and magnified screencaps from the 2003 and 2004 releases of Predator 2.  Note the reduction of halos without a reduction in sharpness.

Brightness and contrast are strictly monochrome characteristics.  Relevant characteristics include black crush, white crush, shadow detail, dynamic range, and black level.  Black crush is a compression of the bottom end of the video dynamic range that obscures various levels of black and dark gray and transforms them into amorphous black blobs.  Shadow detail is related in that it refers to the visibility or obscuring of various levels of black and dark gray.  White crush is the compliment of black crush; it’s a compression of the top end of the video dynamic range that obscures various levels of white and bright gray and transforms them into amorphous white blobs.  Black level is an indication of how dark black appears.  It should not be gray or have any perceptible hue.  Dynamic range is related to the perception of the range of video amplitude from black to white.  A transfer may have great black level, no black crush, no white crush, but if the brightest video is a bit dim, the dynamic range is lacking.  The eye is more sensitive to deficient black level than dynamic range.

Two other digital or compression artifacts are macroblocking and mosquito noise.  Macroblocking appears as small square visible areas of the same intensity and color.  Here’s an exaggerated demonstration.  They are short-term fusions of adjacent pixels caused by compression errors.  These artifacts also have been called pixelation.  They are rare (hence the simulation).















Mosquito noise is visible as random pixels of low contrast around edges of high contrast.  Because the pixels are random, their dynamic appearance is like a subtle swarm of insects, hence the name. 


















The Audio

This is a little less mysterious.  Whether surround effects are audible or not, and the nature of the surround effects are straightforward concepts.  Discrete effects are those that are stationary, they emanate from one location.  Pans are sounds that move from one speaker location to another; the audible motion should be seamless and not change character as the sound moves.  Sound effects will have a dynamic range, from the most subtle to the loudest.  The score fidelity (lack of distortion and wide frequency range) and soundstage (good simulation of space) relate to how realistic the musical presentation seems.  Dialog intelligibility and fidelity relates to being able to understand the spoken word even in the presence of score and sound effects.  Deep bass is just that.  Deep enough and powerful enough, bass may be felt as well as heard, particularly if the frequencies involved are 25 Hz or below. 

All these audible qualities are no less subjective than our assessment of the video, but they are impossible to demonstrate from the website.  We might be able to post an image that shows a problem, but there is no practical way of demonstrating a problem in a 5.1 or 6.1 track without inviting readers into our homes.  So we try to do the best we can with suitable prose.

Final Thoughts

If you have any questions about our methodology or terms that I forgot to mention (or didn’t explain sufficiently clearly), feel free to write and I’ll update this article.  After it’s initial run, you’ll always be able to find it in our TECH section.