payton
07-27-2006, 11:05 AM
hi forum!
after reading many books about digital video every book had a chapter about gamma. finally i thought i really understand it....but it seems i was wrong. im posting it here because i know in this AE forum are people that know pretty much about this technical stuff (like beenyweenies and myleniuim...)
in future i really want to understand gamme in depth because i hate it when i dont understand things that i need to use in daily work...
http://www.rassnik.de/gamma.gif
i made a chart (http://www.rassnik.de/gamma.gif) which should demonstrate how we perceive a (18%) grey card directly per eye or with a videocamera/monitor put into the signal-chain. i will start with the bottom one.
as far as i know right now...(things are a bit simplyfied to prevent blowing this post up too much)
the straigt way:
a) the all-famous 18% grey card, which we perceive as middlegrey, but thought linearly itīs not because...
1) ...our eyes have an included gamma correction of approximately 0,41. this brightens up the card and we...
b) ...perceive it as "middlegrey" (0,41 powerfunction to 0,18=0,5)
the long way around (upper one):
a) the grey card....
3a) ...is filmed by a videocamera. the camera adopts the behaviour of our eyes...
2) ...and corrects the signal with a gamma of 0,45. thatīs why:
I. without this correction done the half of our perceptual range would be encoded into round about 50 shades (with 8bit). banding will occure
II.the approx. inverse gamma of crt-monitors will be compensated. modern tftīs are acting like the old crtīs
3b) ...the 18% grey card should result in rgb values of R=127, G=127, B=127
5) ...when displayed at the monitor...
4) ...the gamma of 2,2 will darken the middle rgb values so that it will simulate the grey card. point 2 & 4 are erasing each other. its like (nearly) nothing happened till the grey card at the beginning.
6) ...the gamma of our eyes will brighten the displayed grey card at the monitor...
b) and we perceive "middlegrey"
the problem:
its written like this in all of these books. sounds pretty nice and logical in my opinion. it would be heaven on earth if there are not these gamma-testcharts that came directly out of hell...
these testcharts are not compatible with my statement at point 3b (and with my whole understanding of gamma)
take a look at this testchart (http://www.rassnik.de/gamma_chart.gif)
http://www.rassnik.de/gamma_chart.gif
if we are sitting in front of a pc this testchard should melt together at a gamma of 2,2. the greysolid of this gamma will be perceived as "middlegrey". in our case of a gamma of 2,2 the rgb values of 186 are "middlegrey"!!! thats complely different from my explained signal chain above. in my opinion rgb values of 127 should be perveiced as middlegrey.
if anyone could sort things out and help me understanding this i will be highly grateful.
now im hopping that a good gamma-discussion will start here with many people on board (but i dont think it will happen ;))
thanks for reading until here,
payton
inserted the pictures direclty into the thread
after reading many books about digital video every book had a chapter about gamma. finally i thought i really understand it....but it seems i was wrong. im posting it here because i know in this AE forum are people that know pretty much about this technical stuff (like beenyweenies and myleniuim...)
in future i really want to understand gamme in depth because i hate it when i dont understand things that i need to use in daily work...
http://www.rassnik.de/gamma.gif
i made a chart (http://www.rassnik.de/gamma.gif) which should demonstrate how we perceive a (18%) grey card directly per eye or with a videocamera/monitor put into the signal-chain. i will start with the bottom one.
as far as i know right now...(things are a bit simplyfied to prevent blowing this post up too much)
the straigt way:
a) the all-famous 18% grey card, which we perceive as middlegrey, but thought linearly itīs not because...
1) ...our eyes have an included gamma correction of approximately 0,41. this brightens up the card and we...
b) ...perceive it as "middlegrey" (0,41 powerfunction to 0,18=0,5)
the long way around (upper one):
a) the grey card....
3a) ...is filmed by a videocamera. the camera adopts the behaviour of our eyes...
2) ...and corrects the signal with a gamma of 0,45. thatīs why:
I. without this correction done the half of our perceptual range would be encoded into round about 50 shades (with 8bit). banding will occure
II.the approx. inverse gamma of crt-monitors will be compensated. modern tftīs are acting like the old crtīs
3b) ...the 18% grey card should result in rgb values of R=127, G=127, B=127
5) ...when displayed at the monitor...
4) ...the gamma of 2,2 will darken the middle rgb values so that it will simulate the grey card. point 2 & 4 are erasing each other. its like (nearly) nothing happened till the grey card at the beginning.
6) ...the gamma of our eyes will brighten the displayed grey card at the monitor...
b) and we perceive "middlegrey"
the problem:
its written like this in all of these books. sounds pretty nice and logical in my opinion. it would be heaven on earth if there are not these gamma-testcharts that came directly out of hell...
these testcharts are not compatible with my statement at point 3b (and with my whole understanding of gamma)
take a look at this testchart (http://www.rassnik.de/gamma_chart.gif)
http://www.rassnik.de/gamma_chart.gif
if we are sitting in front of a pc this testchard should melt together at a gamma of 2,2. the greysolid of this gamma will be perceived as "middlegrey". in our case of a gamma of 2,2 the rgb values of 186 are "middlegrey"!!! thats complely different from my explained signal chain above. in my opinion rgb values of 127 should be perveiced as middlegrey.
if anyone could sort things out and help me understanding this i will be highly grateful.
now im hopping that a good gamma-discussion will start here with many people on board (but i dont think it will happen ;))
thanks for reading until here,
payton
inserted the pictures direclty into the thread