Posts by Sorry that handle is already taken. 3553 publicly visible posts • joined 19 Jun 2009
The monochromatic orange was an extreme example to illustrate the problem based on the human eye's red/green overlap.
In practice, if you have two objects that look to the human eye like they are the same colour under a particular light source, but have very different spectra (i.e. metamers), a 3CCD or Foveon will make them look different, and there's no way around that. Meanwhile a well designed bayer filter array doesn't have this problem.
The Foveon "46MP" sensor captures 15MP of spatial information, and results in images with about the same level of detail as a 30MP bayer sensor, once the bayer sensor's done its demosaic thing.
So for the same output level of detail, the data storage and processing requirements are 50% greater for the Foveon sensor.
The colour filters on a bayer sensor can be designed to replicate the colour response of the human eye, in which there is a large amount of overlap between red and green. A Foveon sensor can't do this because of the way it separates the colour. Imagine a monochromatic orange light. A bayer sensor will pick some up in each of the red and green channels, which will tell the camera that it's recording orange. A Foveon sensor will record it in either the red channel or the green channel, depending on the wavelength, and can't know that it's all gone wrong. It's like a 3CCD, and those are going out of fashion too.
Additionally, despite the marketing material that suggests that Foveon records red/green/blue in its three layers, it's actually recording white in the top layer, yellow in the middle and red in the bottom. It then introduces large amounts of noise as it performs the maths required to turn this back in to red/green/blue.
And that's completely ignoring the complete dishonesty in marketing it as a "46MP" camera.
You probably don't actually want Foveon, although Sigma does make some good lenses.
All DVDs are 720x480 except for the ones that aren't.
Also, they're only "p" if they're progressive scan.
I should have clarified that by pointing out that regardless of how long you're looking at the mirrors, you're not focusing on the mirrors themselves, you're focusing on the objects in the mirrors, which are optically much further away than the mirror itself. Additionally, the point of the rear vision mirrors (and head checks) is to momentarily check for the presence of objects that may become obstacles. Continually changing focus from at a gaming device at less than arm's length and a television screen some metres away is very much different, and hence I agree with the original commenter in that such exercises may well cause eye strain in a statistically significant proportion of the population.
Also, as far as the speedometer is concerned; firstly, the speedometer isn't the mirror, so it's irrelevant to my original comment. Secondly, unless the numbers have moved since you last looked at it, it doesn't take long to check where the needle is without necessarily having to study it in detail.
Okay, so we've determined these things are pocketable if a) you accept the limitations that using a single pancake prime dictates and b) you're wearing a jacket.
And that a µ4/3 system is smaller than a "full frame" DSLR system, but nobody has ever suggested otherwise.
I'm not entirely unconvinced by the luggage-friendly system size argument however.
Human response times don't vary very much around the mean. Pro athletes have an advantage in that they are better at anticipating what is going to happen next.
For example a batsman in cricket will read cues from the bowler's body language to determine what the delivery is likely to be. This has been shown in testing of elite cricketers with shutter goggles that blind them at the moment of the ball's release; they are largely still able to defend the delivery based on the information they've acquired up until that point. Among the best bowlers are those able to mask these cues, confusing the batsman.
Anyway the idea that pro gamers and athletes enjoy one quarter of the reaction time of an average human is pretty much nonsense.
Sure, VA signals have been around for decades, and many of those systems count vehicle throughput. The back end of those systems however is a legacy of their age, so they don't pass around a lot of information and the network is often partitioned into small groups of sites with pre-determined coordination settings. My experience with signal control systems is limited though and I'd be surprised if there aren't more advanced systems outside my experience already in use.
I'm interested in the statement "For example, the system would detect that traffic on one section of road is slowing down or has stopped, and divert it to alternate routes to prevent the backfill causing gridlock." Are they suggesting that the system dynamically and forcibly changes drivers' route choices? How do you do that? The seppos aren't going to like that one.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
