Talk:Chroma dots

This article is written in British English, which has its own spelling conventions (colour, travelled, centre, defence, artefact, analyse) and some terms that are used in it may be different or absent from other varieties of English. According to the relevant style guide, this should not be changed without broad consensus.

This is heavy shit. The phenomenon of b/w images (and not only those, but it's most obvious in monochrome) showing moving color noise on analogue TV equipment has been known for half a century, it's due to splitting the RGB information into chroma and luma, record the color at only half the resolution (see YUV and chroma subsampling), and then send the resulting signal through a poorly isolated consumer cable which will result in crosstalk from the chroma wire to the luma wire, and vice versa (chroma-to-luma results in dot crawl, luma-to-chroma in color noise).

I have NO IDEA WHATSOEVER how they got the idea to use these well-known crosstalk artifacts to "reconstruct" "original" colors. They are no representations of any "real" colors whatsoever, they're just misinterpreted random interference between both wires where some electrons ended up on the wrong wire. Room_at_the_Bottom#Colour_restoration_of_the_original_television_recording babbles some about "color subcarrier" apparent in the dots, but the subcarrier is actually only present in the back porch, which is way-off screen to begin with, in the gap that's between the 576 *VISUAL* lines on the screen and the full 625 line signal. The "49-line" back and front porch aren't visible information whatsoever, they're pure code containing an abstract hue and timing specifier that's internally processed by the receiver, and they don't appear on the screen anywhere other than indirectly, by proxy.

I can only assume they used this random crosstalk color noise to make their best guesses as to "what should be there" according to what shapes they saw (human skin, a table, water), then paint the color in manually. See also placebo. --79.193.57.210 (talk) 23:58, 8 June 2010 (UTC)[reply]

Okay, been looking at their website at http://colour-recovery.wikispaces.com/ . Apparently, they're not even working with color noise as they seem to base their ideas on monochrome TV displays to begin with. Their theory is that what you see as color noise on a color TV will show as brightness noise on a monochrome TV. Unfortunately, those positively IGNORE the color burst and subcarrier alltogether, which is color TV's famous backwards compatibility, the color is stored within a portion of the signal where a monochrome receiver is not even looking for any information.

It's a pretty handy thing actually, as that's why you can easily get rid of color crosstalk from an originally monochrome film by throwing away both color channels of your YUV signal, which is what any monochrome receiver does by ignoring and it's especially easy with digital captures. For instance, in VirtualDub, you just choose the "TV" filter, and keep the Y channel only, and that way the former color crosstalk will not even show up as brightness. Note that just desaturating the whole YUV signal won't work. There's an intuitive visualization here: [1]

And that's just what happens with any monochrome receiver, the color crosstalk is not even visible as brightness patterns, as it's just not even processed by the receiver. But even *IF* it was, it's STILL utterly random. Color crosstalk shows up no matter whether your original footage, or subject, has any color or not, just look at the link above.

And that's also just what all the examples at the website of those guys look like. All you ever see on their site is them processing dated COLOR video footage which looks more or less alright BEFORE they process it, and after their process it looks like they heavily messed with the gamma, desatured it some (but not fully, it's still visible that it's original color footage), and then slammed on lots of color noise, and then rave and rave about how the ugly added color noise would be "full gamut", just look at http://colour-recovery.wikispaces.com/Full+gamut+colour+recovery , just look at which pix read "original video frame" and which "full gamut recovery"! I could do just that within 5 minutes in FinalCut Pro, making it look like a poor reception. --79.193.57.210 (talk) 01:18, 9 June 2010 (UTC)[reply]

When boiled down to their essentials the preceding entries seem to be saying that colour recovery is impossible, when all the available evidence suggest that it IS possible because IT'S BEEN DONE. Jeez. Lee M (talk) 02:12, 2 January 2012 (UTC)[reply]

Needs a visual example

This article could use an illustration. There are many types of artifacts in analog video. 174.20.161.190 (talk) 20:20, 23 August 2018 (UTC)[reply]