Frage von Gubbel:Hello community,
my question is actually already in the title:
I would like to know if there are significant losses if I have a YUV - RGB video in editing and re-spend as YUV. In VirtualDub work as all the filters in the RGB color space ...
If yes, the YUV brightness and two color difference signals are stored / transferred, the "colorize" the SW-Picture then after.
For RGB would be black and white, if all three RGB values are equal. The color results from the relative differences between the color channels.
In YUV can theoretically synonymous "black" and "white" a "Color" have (Y channel is black or white and UV-channels have a value yet), not in RGB (same values for all three channels = no difference)
I notice for example if I have a color shift in RGB refined (eg Flaxen VHS in VDub), in addition to black edges no longer exists Color is available in black areas because no information is gone. If YUV is a. ..
So it looks like with the losses?
MfG
Alex
Antwort von tommyb:
Each color space conversion involves losses with.
YUV to RGB and back again as well - only you do not see with the naked eye. It is measured on each case, however, since filter "hang in between" would be the result of the measurement anyway no preference.
Antwort von Gubbel:
Ok, thank you ...!
How about with my "theory" because of the color shift? Is that with the "color values for Schwar and white"?
Lg
Alex
Antwort von tommyb:
Having thought about it briefly and deeply so true your theory.
Antwort von Gubbel:
... Coooool ;-)
In other words, that stuck in YUV but actually more information?? A 4:4:4 - YUV and RGB have a 24 bit but the same data rate, right?
If the color spaces really the same?
Actually, however, the "losses" only to be rounding errors in the conversion when converting ...
MfG
Alex
Antwort von WoWu:
@ Gubbels
I have with the move but did not quite understand what you want to do there, but you have both conversions are not only synonymous but rounding losses losses s.den implementations because such coefficients such as red and blue are very difficult to implement (ZBR = 0.2116 or B = 0.0722). So here you get a very low efficiency.
It would be better, would you deal with the RGB color space and YCoCg.
Since you are lossless, because everything can be made about addition and shift operations.
But I think your starting source is fixed there.
If you can avoid the conversion, please let dear.