A 12 bit linear system can record 12 stops at most, so take that as the upper limit. (I believe that a non-linear response or more than 12 bits would mean a complete system redesign, which I didn't hear being suggested. I suspect they just found one or more parts with improved specs to plug into the existing design.) If you look at the curve that the 11.3 stop claim was based on, you will observe in the bottom left region, the lowest 1 or 2 stops of exposure have a much different slope (flatter) than the rest of the plot. This suggests it is an area of significantly increased noise. If the improved circuit has lower noise, this will give you a much more usable image in the underexposed areas, without necessarily increasing the measured dynamic range number at all. In the end, you care about what the picture looks like and the DNR number alone doesn't tell us that.
Ok maybe I'm being thick headed but my understanding was that the 12bits was just the ceiling at which you see no more improvement from 12 stops. In other words 12stops @ 14 bits is just wasting two bits to noise. That still doesn't mean you can only record a maximum of 12 stops to 12 bits correct?
I'm hoping two weeks max !
Me too. I've got a lot of projects with RED-hungry clients this summer!
For those who want to know (and the fresHDV guys can attest to have been in a conversation with Graeme about this)....
Changing your white point affects your perception of color. You can see this in CRT displays. Moving the display's white point from D65 to 9300K makes reds appear less saturated. To compensate for this, a lot of consumer CRTs will oversaturate the reds and mess with the hue.
When you white balance an image, you kind of have the flip side to this. You white balance from whatever the scene is (for outdoors scenes, this changes throughout the day and depends on cloud cover) to your target white point. Target white point for video is D65, for DCI is P3 (I think; see etconsult.com), and for film is 5400K.
*White point = the color of white. Sort of like the color temperature of a light.
Ok, so when you change the white point you can numerically change all your values so that they look the same. There are different numerical schemes for this, such as the Bradford transformation. Bruce Lindbloom's site goes into some of the color science of it:
(and see the links to other pages too)
2- One of the problems with color science is that we don't understand everything... i.e. a lot of the color perception models are imperfect. I've heard a lot of good things about Fairchild's book Color Appearance Models 2nd Ed and it's probably a good reference for color science stuff.
I think that's the Coles notes version. You will be tested on this. ;)
"If we deliver on time, but the product has defects, we have not delivered on time."
- Philip Crosby Let's Talk Quality, 1989
Please take the necessary time to make this a camera for the history books. I've got my reservation locked. I'm not going anywhere. Appreciate all the hard work.
|« Previous Thread | Next Thread »|