Westinghouse K840CK15 Restoration

[old_tv_nut]

Actually, the TK-41 image orthicon cameras were carefully filtered to match NTSC specs without any color matrixing, so broadcasts were per NTSC until the mid 60s or thereabouts. Meanwhile, the CRTs had changed to sulfide green (moving skin tones toward red) and a blue that was much less cyan than NTSC blue (moving skin tones toward yellow). At first, TV makers didn't change the matrixing in the receiver, went to cyanish white point to save the red gun, and left it to viewers to find a compromise adjustment for acceptable skin tones. Later, TV makers started changing the matrixing to get a better hue range from red through skin hue, orange, yellow and green, but this resulted in full amplitude reds being overly bright. When Plumbicon cameras came, they were matrixed to something sort of like NTSC at first in the US, but correctly for modern phosphors in PAL equipment. Of course, camera makers wanted the color to look decent on studio monitors in the US too, so there were compromise matrix designs going on. The result in the US was like the old joke about the factory whistle being set by the train arrival and the station clock being set according to the factory whistle. At the same time, US TV makers were introducing one-button auto color circuits that essentially moved anything close to skin hue even closer - but at least you could turn it off. SMPTE developed a monitor standard that could have the in-monitor matrix turned on or off. Off for setup and calibration, on to get the correct hue distinction from red to green, but with the over-bright reds. Color matrixing didn't get really sorted out in the US until HD came along and the whole world settled on a single standard, a very close approximation to the European/PAL and SMPTE C, which was also adopted as computer sRGB.

Here's something I wrote on the match of TK-41 cameras to NTSC (15GP22 CRTs) - essentially, the slight discrepancies could be made up easily with small adjustments of the receiver color and hue ("tint") controls. It's based on actual RCA documentation of the early camera color response, which was fortunately one of the few things saved from the trash when XSarnoff LabsX (correction - RCA Camden)closed. Additional edit: The TK41C prism assembly was kindly measured for this paper by Jay Ballard at my request.

http://www.bretl.com/viewing1950scolor.htm

The need for narrowband color trimming filters in the three channels was one reason the TK-41 was so much less sensitive than monochrome cameras. In the earliest experimental cameras, the green spectral response was too much toward blue to match the NTSC green phosphor. This meant that yellow, orange and skin tones put less light into the green channel than they should, and those colors came out too orange or reddish. Observers commented on this, which contributed to the idea that RCA's system did not have as good color as the CBS system. However, this was soon corrected, and the cameras were then actually considerably more accurate than color film when used with the NTSC phosphors.

The later model TK-41C with improved dichroic filters (in a prism block instead of the original thin mirrors) increased the optics efficiency quite a bit because [edited] the trim filters were attached to the prism block and were apparently more efficient (perhaps dichroic). (The prism block also eliminated ghost reflections from the mirrors' second surfaces.)

Looking at the greens on the Wizard of Oz today is very confused, because in restoring the film for DVD and Blu-Ray, the color has been adjusted for current displays, as well as attempting to divine what the original really looked like. The result is that, while the greens on your 15GP22 are impressively saturated and not yellowish, they are being artificially made that way by applying an sRGB signal to NTSC phosphors, and there is no way of telling if that is really accurate or not. Measurements of Technicolor dyes indicate that color saturation that high could only be obtained with quite dark, dense prints - BUT technicolor WAS always printed with high contrast and density, which is what produced the high saturation. So, it's probable there are compensating errors here, and the saturated greens on the 15GP22 are mostly legitimate. Whether they are "correct" or not, they certainly look pleasing.