Path: exodus.Eng.Sun.COM!NewsWatcher!user From: poynton@poynton.com (Charles Poynton) Newsgroups: sci.engr.advanced-tv Subject: Re: Re contrast vs resolution Followup-To: sci.engr.advanced-tv Date: Wed, 06 Oct 1993 20:05:04 -0800 Organization: Sun Microsystems Lines: 34 Message-ID: References: <9309291425.AA10933@spot.src.honeywell.com> NNTP-Posting-Host: desk As several posters have reported, black-and-white photographic negs and prints can have amazing contrast, several hundred to one. CMYK printing is generally much worse than a good B&W print, maybe 100:1. It is an extraordinarily good cinema that exceeds a hundred to one. The best screening room I have ever been in was measured at 160:1. But the EXIT lights in a typical cinema scatter enough light on the screen to bring it down to a little less than a hundred. CRTs are restricted by scattered light and ambient light. If you measure a CRT displaying a very small white blob, say a centimeter across, on a black flatfield in a pitch black room, a well-adjusted CRT can get 400:1. But if you make the blob much bigger, the light it emits starts to scatter into the dark area of the screen and the contrast goes down. In a very dark room, a well-adjusted CRT can get say eighty to one; in a typical dim living room in the evening, 30:1, and in a fluorescent-lit office, perhaps eight-to-one. To follow a poster's "off-the-track" comment, PhotoCD digitizes exposed film that has already been subjected to the S-shaped transfer function, so the mathematics is not easy. The digitization is 14 bits. But during a pre-scan, the transfer station slides a histogram up and down the image data and decides on a suitable dynamic range -- Dmin and Dmax, if you like -- for an eight-bit nonlinear code. The coding law is that of CCIR Rec. 709, the HDTV transfer function. It is a 1/0.45 power function with a linear slope of 4.5 near black. More detail in another of my TNs. If you guys can print PostScript and take a 100 KB mail message, I'll post it. C. Charles Poynton poynton@poynton.com