>原作者是 Henry
>抄錄自rec.desktop.video
Computers typically process graphics in RGB color space
(i.e. each primary color has its own channel), whereas
a digital camcorder does it in YCbCr space (there is a
brightness part of the signal and two color difference
channels with lower resolution.)
The thing is, the YCbCr format in the camcorder has some
headroom below "black" and above "white" for additional
processing (for example, to correct overexposured parts
of the image). On a scale from 0 to 255, black is defined
as "16" and white as "235".
Computer RGB formats do not have the same kind of concept
- they use the full scale from 0 to 255, where black is 0
and white 255.
* * *
Now we have a dilemma: when converting YCbCr video data
to RGB for computer processing, how can we retain the
extra headroom and footroom in the values? If we simply
clip them off and rescale the 16...235 range to 0...255,
we do not any longer have that extra data to work with.
Usually, this is solved simply by using the same limited
brightness range on a computer, too. Even though the image
will look dark and washed out when processing it on the
computer, it looks fine when put back on video tape. And
you get to keep that extra image data for editing. As
almost all video capturing/processing works this way,
it is hard to avoid. The same problem exists with playing
DVDs on a computer, too - this is why the DVD player
applications usually have contrast and brightness controls.
However, if your final target _is_ computer displays and
applications, not video devices, you should correct the
brightness range for better viewing on a computer monitor.
One tool for doing this is Avery Lee's VirtualDub and its
"levels" filter. See <
http://www.virtualdub.org/>.
VirtualDub can also frameserve to other applications,
such as MPEG encoders.
Unfortunately, you must re-encode your MPEG video if you
want to make any permanent changes to it, and this will
always reduce quality. Another option would be simply
using a media player application that has brightness
controls, such as Zoom Player:
<
http://www.inmatrix.com/files/zoomplayer_download.shtml>)
You could also be interested in Ben Rudiak's lossless
HuffYUV codec for storing intermediate video files,
should you need them at any point:
<
http://math.berkeley.edu/~benrg/huffyuv.html>.
For some background reading, see Charles Poynton's
excellent Color FAQ:
<
http://www.inforamp.net/~poynton/no...lour_and_gamma/
ColorFAQ.html#RTFToC29>
================================
以下是台湾 denysyou 的翻译，原文发表在台湾 dvworld.com.tw 的论坛上。原文地址：
http://www.dvworld.com.tw/forum/ ... d.php?threadid=2831
對不起幫忙翻譯一下,翻得不好請不要見怪
電腦典型處理圖形(或圖片)是在RGB(紅綠藍)的色彩顏色空間的範圍
(也就是每一個數主要的顏色有其對應的顏色空間或頻率),然而一台數
位攝影機卻是以 YCbCr來加以處理(這種顏色的空間處理方法是:在
顏色訊號會有一個明亮度的部份以及低解析度時所伴隨著的兩個顏色
差異區)
因此事實上是,在攝影機中的YCbCr的影像模式會有針對頂頭的淨空
區域在低於全黑的下方及高於全白的上方有著特殊的處理(例如影像
中的正確曝光部份).其範圍從0到255的區間,黑色被定義為16的數值
,而全白被定義為235的數值
電腦的RGB格式並沒有相同的概念－而是使用全部的範圍,也就是從
0到255,此時的定義全黑為0,而全白為255
所以我們就會有一個進退維谷的狀況：也就是當將屬於YCbCr的視
訊資料轉換成RGB的資料來提供給電腦處理,我們如何保留這些特別
的高層區和低層區對應的數值呢？假如我們僅僅是將他們略去,然後
重新將這個16到235的數值區重新調整成0到255的區段時,我們將從此
不再擁有這些特別的區域可以重新加以處理
通常,我們僅只是解決這個問題採用與電腦相同限制的明亮度
僅管如此,影像將會看起來比較暗或者較漂白,當我們在電腦上
處理這些影像,但是當我們將資料放回錄影帶上重新檢視時又
看起來很好.而我們在編輯這些影像資料時又可以得到這些額
外的資料.當幾乎所有的影像擷取/處理都是這樣運作的,因此
也就難以避免了.相同的情形也存在於在電腦上播放DVD的影
像,這也就是為什麼DVD播放機的應用軟體會有對比及明亮
度的控制了
然而,假如您的最後的目標是要做在電腦上播放或者是電腦的應
用程式,而不是視訊的裝置,你也許應該修正明亮度的範圍,也便在
電腦的螢幕上獲得更好的觀看品質.
一個有用的小工具可以作這樣的事是由Avery Lee's VirtualDub
及它的"levels(層次)"過濾器
可以在以下的網址獲得
http://www.virtualdub.org/
VirtualDub也同樣擁有畫面處理的應用程式如MPEG壓縮程式
很不幸的,你必須重新壓縮的你MPEG視訊資料,假如你想要任何
永久的改變時,而這也會降低它的品質.另一個很簡單的解決方法
是使用具有明亮度控制的媒體播放應用程式如Zoom Player
可以在以下的網址獲得
http://www.inmatrix.com/files/zoomplayer_download.shtml
或許你也有興趣使用Ben Rudiak的YCbCr不缺少的及霍夫曼YUV的
壓縮程式,可用在儲存視訊資料的中間媒介媒體
可以在以下的網址獲得
http://math.berkeley.edu/~benrg/huffyuv.html
還有一些背景知識的資料,可見於 Charles Poynton的卓越顏色問與答中
網址如下：
http://www.inforamp.net/~poynton/no...lour_and_gamma/
ColorFAQ.html#RTFToC29