Home Theater Forum - View Single Post

Cees Alons · 02-01-2003, 09:21 AM

The Dreaded Chroma Bug

The so-called Chroma Bug actually consists of two problems, one of which is a real bug - in some DVD players. Both problems are related to the use of 4.2.0 colour decoding in interlaced video signals, although the bug is only obvious in static parts of progressive images.

What you may see is this: distinctly coloured objects (against a differently coloured background) have scan-lines directly above and below their edges where the colour of the object is still present in places where it shouldn't. It's very difficult to see on smaller screens (27" or smaller), but easily spotted on the larger ones, especially large projection screens. Also, slightly differently coloured horizontal stripes may be visible inside coloured objects (in fact: colour lines are "switched", more or less).

You may never have noticed it yet.

(1) The real Chroma Bug is caused by DVD-players in which the colour-decoding part of the MPEG-decoder makes no difference between interlaced frames and non-interlaced frames. They should be using a different decoding algorithm for each, but instead use the interlaced technique for both (sort of laziness of the developing engineer). Thus, the problem is visible in progressive (= non-interlaced) frames, but in fact only in static parts of the image (moving parts have "busy" edges anyway, so the human eye won't notice it). The interlaced frames are decoded properly in these DVD-players.

Most newer DVD-players have it right, now, so the chance of seeing this problem gets less and less (there's a difficult technical complication caused by some encoders that set the progressive-frame-flag in an odd manner - but it goes too far to go into that here, and it can be solved by the decoding algorithm anyway).

(2) The other Chroma Problem is more severe in nature, although slightly less visible. It's inherent to the MPEG 4.2.0 encoding of colour fields and the interlaced decoding technique. A 4.2.0 encoding means less resolution in the chroma signal than in the luminance signal. That's not too bad: the human eye itself uses a different resolution there (the rods-and-cones thing).
But as a result, the colour image, when properly (sic!) decoded in the "even" frame (reconstructed to full resolution, so to say), still cannot be the same as the image of the "odd" frame. Thus odd and even lines may have differently coloured edges along (again: static) objects, which will be visible to the human eye on large projections. This is not a bug in either the encoder or the decoder (as the above mentioned problem is), but nevertheless: it can be solved by special circuitry, especially inside the de-interlacer.

Again, it goes too far to discuss this in detail here, but be assured that the Faroudja deinterlacer, to mention one brand, solves the problem rather well.

02-01-2003, 09:21 AM
Cees Alons *Cees Alons* Administrator Location: Amsterdam, Holland Join Date: Aug 1997 Local Time: 03:16 AM Local Date: 10-07-2008 Posts: 18,180	The Dreaded Chroma Bug The so-called Chroma Bug actually consists of two problems, one of which is a real bug - in some DVD players. Both problems are related to the use of 4.2.0 colour decoding in interlaced video signals, although the bug is only obvious in static parts of progressive images. What you may see is this: distinctly coloured objects (against a differently coloured background) have scan-lines directly above and below their edges where the colour of the object is still present in places where it shouldn't. It's very difficult to see on smaller screens (27" or smaller), but easily spotted on the larger ones, especially large projection screens. Also, slightly differently coloured horizontal stripes may be visible inside coloured objects (in fact: colour lines are "switched", more or less). You may never have noticed it yet. (1) The real Chroma Bug is caused by DVD-players in which the colour-decoding part of the MPEG-decoder makes no difference between interlaced frames and non-interlaced frames. They should be using a different decoding algorithm for each, but instead use the interlaced technique for both (sort of laziness of the developing engineer). Thus, the problem is visible in progressive (= non-interlaced) frames, but in fact only in static parts of the image (moving parts have "busy" edges anyway, so the human eye won't notice it). The interlaced frames are decoded properly in these DVD-players. Most newer DVD-players have it right, now, so the chance of seeing this problem gets less and less (there's a difficult technical complication caused by some encoders that set the progressive-frame-flag in an odd manner - but it goes too far to go into that here, and it can be solved by the decoding algorithm anyway). (2) The other Chroma Problem is more severe in nature, although slightly less visible. It's inherent to the MPEG 4.2.0 encoding of colour fields and the interlaced decoding technique. A 4.2.0 encoding means less resolution in the chroma signal than in the luminance signal. That's not too bad: the human eye itself uses a different resolution there (the rods-and-cones thing). But as a result, the colour image, when properly (sic!) decoded in the "even" frame (reconstructed to full resolution, so to say), still cannot be the same as the image of the "odd" frame. Thus odd and even lines may have differently coloured edges along (again: static) objects, which will be visible to the human eye on large projections. This is not a bug in either the encoder or the decoder (as the above mentioned problem is), but nevertheless: it can be solved by special circuitry, especially inside the de-interlacer. Again, it goes too far to discuss this in detail here, but be assured that the Faroudja deinterlacer, to mention one brand, solves the problem rather well.