Given that the face tracking system (the two cameras facing you) can tell if you are falling asleep, (The system was originally designed for driver fatigue studies - looking for things like blink frequency/duration and head-nodding), I would have thought that the potential to save lives would be quite high - especially for those people who work on the roads for a living. (HGV drivers, bus drivers, travelling salesmen, etc...) - I could imagine (in the distant future), a similar device replacing the cardboard-disk tachographs found in HGVs. (How much more effective would it be to prevent someone from driving when they were actually tired, rather than according to a fixed number of hours-on-the-road and so on?)
BTW - I have to admit to some bias, as I work for one of SeeingMachine's distributors. (SeeingMachines is the Australian company behind the FaceLAB system - go check them out, the technology is pretty cool - Machine Vision, finally coming of age!)
I think it is more a matter of academic debate than a serious issue: If CIE has been good enough for the professionals for the last 60-70 years, it is probably good enough for just about any consumer product - we are talking minor quibbles that might matter to a vision scientist specialising in colour perception, but are unlikely to matter too much to anyone else.
You can actually buy graphics cards with 42 bits colour resolution (14 bits per channel) from Cambridge Research Systems ( Although I have to admit a certain bias in this plug.:-) )
With the unfortunate problem that the datasets used to construct the CIE colour spaces are wrong. Well, not wrong, exactly, but they were created using data from experiments performed in 1924 on a small number of elderly pensioners. (Or so I am led to believe). The result of this is that the CIE colour spaces are a bit off towards the blue end. (This is because as you age, your lenses yellow as you gradually develop cateracts brought on by UV light - this absorption of low wavelengths has affected the data, making it less representative for the average joe-in-the-street than it might have been.)
Re:RGBCMY is more marketing factoid than it isreal
on
RGB to become RGBCMY
·
· Score: 1
What do you mean, a basis set? A set of colours from which you can derive other colours? For this kind of discussion, we need to talk primaries.
As far as I am aware, negative colours in RGB space were introduced with sRGB, in an attempt to introduce a consumer "standardised" RGB colour space that would map to the gamuts of a wide variety of consumer display devices easily and with a minimum of development cost.
The real issue is mainly about how widely spaced the primaries are in CIE1932/CIE1976/SML/Macleod-Boynton space (or some other physiological colour space). If the primaries are so close together that another device can be more red than your maximum red, or less blue than your minimum blue, then negative colours are perfectly acceptable - but only in terms of some other display device.
The farther apart the primaries are, the larger the gamut, and better the display. More primaries can make a larger gamut, if the new primaries are outside of the gamut formed by the original primaries.
I am not an expert, but I am led to believe by those who are that your rods are mostly bleached out in normal lighting conditions, and only really become active/useful in mesopic/scotopic (read:-really very very dark indeed) conditions.
It is your cones that provide most of the useful information to your visual system in daylight & twilight.
There are two parts to the technology that these people have: The display with the high number of primary colours, and the gamut-mapping algorithm in their colour chip. The first aspect is a matter of engineering and chemistry - the second is a matter of psychology/psychophysics (or rather, colour science).
It is not going to be easy to take an RGB signal and generate from it a RGBCMYK signal that looks perceptually similar, but "richer" and "more vibrant" than the original signal. (Just think how fuzzy and imprecise these terms are.)
- In reality the company probably has a quasi-parametric statistical model (think PCA/ICA or some such) driving the gamut-mapping process; the model itself operating on a dataset collected from a legion of volounteers doing colour-matching exercises or some such.
The real measure of how good these displays are will lie in whether or not it is possible to bypass the chip and manipulate the primaries directly - this will be the true mark of a professional display.
(Unfortunately, it is highly unlikely that this is the case, as modern displays are getting more and more complex, do more and more proprietary processing, and allow less and less control over the final output.)
Slashdot Mirror
Given that the face tracking system (the two cameras facing you) can tell if you are falling asleep, (The system was originally designed for driver fatigue studies - looking for things like blink frequency/duration and head-nodding), I would have thought that the potential to save lives would be quite high - especially for those people who work on the roads for a living. (HGV drivers, bus drivers, travelling salesmen, etc...) - I could imagine (in the distant future), a similar device replacing the cardboard-disk tachographs found in HGVs. (How much more effective would it be to prevent someone from driving when they were actually tired, rather than according to a fixed number of hours-on-the-road and so on?) BTW - I have to admit to some bias, as I work for one of SeeingMachine's distributors. (SeeingMachines is the Australian company behind the FaceLAB system - go check them out, the technology is pretty cool - Machine Vision, finally coming of age!)
I think it is more a matter of academic debate than a serious issue: If CIE has been good enough for the professionals for the last 60-70 years, it is probably good enough for just about any consumer product - we are talking minor quibbles that might matter to a vision scientist specialising in colour perception, but are unlikely to matter too much to anyone else.
You can actually buy graphics cards with 42 bits colour resolution (14 bits per channel) from Cambridge Research Systems ( Although I have to admit a certain bias in this plug. :-) )
With the unfortunate problem that the datasets used to construct the CIE colour spaces are wrong. Well, not wrong, exactly, but they were created using data from experiments performed in 1924 on a small number of elderly pensioners. (Or so I am led to believe). The result of this is that the CIE colour spaces are a bit off towards the blue end. (This is because as you age, your lenses yellow as you gradually develop cateracts brought on by UV light - this absorption of low wavelengths has affected the data, making it less representative for the average joe-in-the-street than it might have been.)
What do you mean, a basis set? A set of colours from which you can derive other colours? For this kind of discussion, we need to talk primaries. As far as I am aware, negative colours in RGB space were introduced with sRGB, in an attempt to introduce a consumer "standardised" RGB colour space that would map to the gamuts of a wide variety of consumer display devices easily and with a minimum of development cost. The real issue is mainly about how widely spaced the primaries are in CIE1932/CIE1976/SML/Macleod-Boynton space (or some other physiological colour space). If the primaries are so close together that another device can be more red than your maximum red, or less blue than your minimum blue, then negative colours are perfectly acceptable - but only in terms of some other display device. The farther apart the primaries are, the larger the gamut, and better the display. More primaries can make a larger gamut, if the new primaries are outside of the gamut formed by the original primaries.
That was poor. Very poor. Very poor indeed. And offtopic. A bit like this, really.
I am not an expert, but I am led to believe by those who are that your rods are mostly bleached out in normal lighting conditions, and only really become active/useful in mesopic/scotopic (read:-really very very dark indeed) conditions.
It is your cones that provide most of the useful information to your visual system in daylight & twilight.
There are two parts to the technology that these people have: The display with the high number of primary colours, and the gamut-mapping algorithm in their colour chip. The first aspect is a matter of engineering and chemistry - the second is a matter of psychology/psychophysics (or rather, colour science).
It is not going to be easy to take an RGB signal and generate from it a RGBCMYK signal that looks perceptually similar, but "richer" and "more vibrant" than the original signal. (Just think how fuzzy and imprecise these terms are.)
- In reality the company probably has a quasi-parametric statistical model (think PCA/ICA or some such) driving the gamut-mapping process; the model itself operating on a dataset collected from a legion of volounteers doing colour-matching exercises or some such.
The real measure of how good these displays are will lie in whether or not it is possible to bypass the chip and manipulate the primaries directly - this will be the true mark of a professional display.
(Unfortunately, it is highly unlikely that this is the case, as modern displays are getting more and more complex, do more and more proprietary processing, and allow less and less control over the final output.)