On Mar 15, 2012, at 3:15 PM, Godfrey DiGiorgi wrote: > On Thu, Mar 15, 2012 at 2:41 PM, Larry Colen <l...@red4est.com> wrote: >> Let's rephrase that question: >> >> "Would you like a camera with two stops more sensitivity, and three or four >> times the resolution? > > Don't mix up the subject of resolution and sensitivity. A Bayer mosaic > sensor with a given dimension photosite array has exactly the same > spatial resolution as a non-mosaic sensor with that same set of > dimensions. The only way you get more resolution is to add photosites > and reduce antialiasing requirements. A Bayer mosaic sensor > interpolates chrominance values per output pixel, not the spatial > values of points in the scene it captures.
Interesting, I misunderstood. I thought that each final site was an interpolation of three or four adjoining sites. Since it wasn't something I had the opportunity to do anything about, I didn't study it as deeply as I might. In a similar vein, I'm not as well versed on optical design as I might be as if I'd taken a course in it college. > >> What if you had to give up color to get it?" >> >> Color doesn't come for free. In order to get color, you have to throw away >> a lot of light at each sensor site. If all you ever photograph are >> technically easy subjects like models in the studio, barns, or cats in >> plenty of light, then incrementally better performance isn't a big deal. If >> you find yourself pushing the performance envelope of your camera and having >> to convert to B&W, not for specific effect but because the noise is less >> obnoxious in B&W than in color, an added stop of performance can be worth >> it. > > Here's the deal: > > The key to B&W photography is that it is a non-linear capture of a > color universe abstracted into a set of monochromatic tonal > relationships. One of the vectors is the translation of chrominance to > luminance. When we were working with B&W film, we chose films with > particular spectral characteristics and then added filtration in front > of the lens to tweak those spectral characteristics to match the > particular end tonal relationships we wanted. For instance, red and > green at the same luminance look the same when cast into monochromatic > values. But to capture their perceptual/emotional difference > correctly, we need to differentiate them. So we use filters that > separate them (red filters makes red things brighter, green things > darker, and vice versa). Exactly, or used films with different characteristics. > > When we moved into the digital capture realm, B&W become a rendering > exercise instead of a capture exercise ... it's dependent upon how we > evaluate and translate the chrominance values into luminance, not how > the sensor captures the data that we work from. We pay a small price > for this in terms of putting filters on the photosites, reducing > sensitivity by some amount (not two stops, given the same photosite > array dimensions) to capture the whole range of spectral data, and by > doing so we gain massive amounts of rendering flexibility. I figured that each filter cut out about 2/3 of the spectrum, which would be about a stop and a half, but I wasn't letting a strict adherence to fact interfere with my hyperbole. > > With a sensor that only captures monochromatic values, we're back to > considering B&W at capture time ... so we have to pull out the filters > again in order to adjust chrominance to luminance translation at > capture time, cutting sensitivity by at least as much as the Bayer > color mosaic does. If, of course, you choose to not accept the broad band color balance. If I'm using a black and white sensor, it'll be because I need every single photon that I can capture. Plus, I always have the option of dropping a color filter in front of the lens anyways. > > I'd rather have the flexibility. I've worked with monochromatic > digital capture cameras way back in the digital dark ages (1980s) and > have NO interest in going back there. I want the ability to manipulate > chrominance to luminance translation at rendering time, not at capture > time, with a larger, more robust captured dataset. One could apply the same argument to the lightfield camera. > >> I probably wouldn't drop the dosh on the Leica version though, even if I had >> Godfrey's toy budget. > > LOL! > > Sorry, the Leica isn't a toy. I don't consider it as such. A K5 or > K-01 might be a toy to you ... to me the Leica M9 is a serious > instrument for making photographs, as would either the K5 or K-01 be > if I went to purchase them. Until I'm making my living, or at least substantial income, from photography, if I'm being honest with myself, my cameras are toys. My cameras contribute significantly to my mental health, not so much to my fiscal health. > By making that statement, you are implying > a) that I have a lot of money and b) that I toss money around without > discretion, Larry. And it was exactly that implication that got the laugh I was aiming for. See note above about facts and hyperbole. I trust you to be a big enough man to take a little ribbing now and then. Besides, you're so darn cute when you pout. -- Larry Colen l...@red4est.com sent from i4est -- PDML Pentax-Discuss Mail List PDML@pdml.net http://pdml.net/mailman/listinfo/pdml_pdml.net to UNSUBSCRIBE from the PDML, please visit the link directly above and follow the directions.
