Originally posted by gilboa
View Post
Announcement
Collapse
No announcement yet.
VESA Releases DisplayPort 1.3, Pushes 32.4 Gbits/sec
Collapse
X
-
-
Originally posted by log0 View PostThose 16.7 millions colors are not optimized for human perception. Our eyes are nonlinear devices, and very good at noticing brightness changes. And with 24bit you only have 256 shades, means you will see banding along smooth gradients, period. If you don't, you should eventually consider consulting an eye specialist.
IF you have a PROPER image with 256 shades of Grey (as an example) with a PROPERLY configured video driver/card to display FULL range (i.e. 0-255 per channel instead of 16-235 (witch is only 4bit, 16 levels of shade , witch is EASY to notice) for example witch by default ALL video drivers assume when installed) and IF you use a CRT (actually done also tests in LCDs), you CAN'T distinguish the EXACT position of a shade transposition.
IF YOU can see the difference is because YOU need to consult a eye specialist or something is INCORRECT with TEST conditions
(i.e. Video driver or card is NOT actually displaying image in full range but, as was PROVEN in the past with some cards/drivers, that some companies LIE when say it's in full range and actually was discovered image was be presented in 12bit, 16bit, 18bit or even 21bit (21bit is ONLY a little more than 2million colors or 128 shades of a color) witch is EASY for Human eyes distinguish....IIRC, Human eyes distinguish 8-10million colors.
Using only 12-21bit limits dramatically the number of shades (and colors) and then, yes, normal Human eyes can notice difference in the shades scale)
In fact, there's lots of people with slight problems in distinguish colors and color grades, i know several that have hard time to distinguish orange for example and call to same object painted in orange, red or yellow
Comment
-
Eye physiology is *really* complicated.
As a previous poster said, they are not simple linear detector. (And, shall I add, do no give results using absolute numbers).
There's a difference between:
- being able to recognise a random colour anywhere
- being able to visually compare colour in the same picture
- being able to notice "steps" in a gradient (and it also depends what colour is the gradient, we'll make more distinction between grays than other colours).
There's a reason why, radiologist:
- ...either used to use special monitor (B+W monitor which display more than 256 levels, and has a maximum brightness range which goes higher than regular monitors)
- ...or nowadays use software where they can quickly set different "windows" to watch different shade details.
(And before digital radiology, radiologist used transparent film suspended over a VERY bright light source).
10bit component instead of 8bit also make sense as their are RAW component, before applying any final transform (light & contrast, gamma curve, etc.) which could distort the signal by rounding errors
(see the mention of "badly confirgured" monitors).
i know several that have hard time to distinguish orange for example and call to same object painted in orange, red or yellow
Comment
-
Originally posted by DrYak View PostThere's a reason why, radiologist:
- ...either used to use special monitor (B+W monitor which display more than 256 levels, and has a maximum brightness range which goes higher than regular monitors)
- ...or nowadays use software where they can quickly set different "windows" to watch different shade details.
(And before digital radiology, radiologist used transparent film suspended over a VERY bright light source).
IF radiologists need(ed) to use special monitors with much wider brightness ranges than normal monitors to distinguish the different shades, it's because in a normal monitor, you can't.
Same applies when you talk about special software to set "windows" and possibly using special software tricks to enhance shade distinguish.
Without those, they couldn't notice the differences.
Just make a test with GIMP, with 4 square areas, 1 in white (255,255,255), other in red (255, 0, 0), other in blue (0,0,255) and other in green (0, 255, 0).
In 3 of them you gonna print a letter, number or symbol like "+" but in a color level only one level inferior (i.e. if in red square, print text in 254,0,0) .
Merge in the end all layers and make selection equal to none (you can also save file as BMP)...now just ask people in your family or friends if they can find out if ANYTHING is written in any of the squares or not, and if they feel there is, WHERE and WHAT.
I just made a test with several people here...NO ONE have a clue, but hey, maybe we are all "blind" :")
PS: NO ZOOM or any CRT/LCD settings change allowed during testLast edited by AJSB; 16 September 2014, 10:54 AM.
Comment
Comment