CompuServe Messages

In recent messages I did notice a tendency for you to mix the numbers up regarding this resolution issue. Specifically you had recently confused the vertical scan line NTSC limitation with vertical division lines that determine the increment of horizontal information (hor. res.) I get the idea that you are rather well read in the technologies but somehow seem to be missing something very important here to relate the whole picture. Maybe it was the lateness of the hour or whatever but I'll start from the top with my understanding of the issue for other's benefit as well. This will be lengthy so I hope it doesn't get chopped by CIS. IN VIDEO, resolution is actually measured at the output monitor using a chart that is divided into alternating converging black and white lines. The test is: frame up the chart with an input device, usually a camera, and display and measure on the monitor visually. At the point where the lines are no longer separate you take a reading and this becomes the particular resolution in "lines" The chart has several locations where one can measure this usually at each corner and at center. There is a separate scale for horizontal and vertical resolution. This process tests the system rather than any particular piece of the system. Using a pattern generator the camera may be replaced. The RS-170A specification decrees a horizontal scan rate of 15,734 Hz and 525 lines vertical. It also specifies interlaced vertical lines which is 525/2=262.5. If we divide the horizontal by the vertical rate per field we get 15734/262.5= 59.94 Hz. or vertical scan rate commonly rounded to 60Hz. The vertical resolution is therefore limited to the number of vertical lines in the display. Further limiting this resolution is the number of lines that are actually used for picture display. The number is 486 or 248 per field. Therefore , the maximum vertical resolution obtainable is 486 lines for each picture. Horizontal resolution on the otherhand is not limited by a precise on – off of the phosphors of the display device but instead by the ability for the signal to represent a visual change as the scan lights up phosphors as it travels from left to right across the screen continuously. This changs is limited by many other factors as well but mostly by the the horizontal bandwidth of the video signal. With color luminance and chrominance must be addressed separately. For simplicity I'll use the higher figure of luminance which is 4.7 Mhz. (1.5 Mhz for chrominance). The number of lines may be approximated, assuming all else in the circuitry is "perfect" with the formula N=BW*2T [ BW= bandwidth, N=#of hor. lines, T= time of hor. trace. 63.5 us (us=micro seconds) less blanking or 53 us] This gives us a maximum picture resolution horizontally of 498.2 lines, or 500 lines which is theoretical. Note: you can calculate the horz. scan trace with 1/15734Hz. For recording, the horizontal luminance is boosted to accomodate even higher resolution. In order to produce a line on a video screen the trace has to alternate between black and white and back to white again. To produce this effect on a picture element basis (pixels) we would need a minimum of two pixels per line when adjacently placed. This equates to video lines of resolution = pixels/2 or 1000 pixels to generate 500 lines which is broadcast spec for NTSC as far as continuous image is concerned. Another issue at play is the measurement of resolution called rise time. This is generally used in character generators and graphics where by the measurement is determined for the time it takes the trace to rise from black to white as in graphics. The lower the figure the faster the rise the sharper the graphic. The figure of 0.035 us or better is considered to be "broadcast minimum standard" as for character generated graphics. The maximum number of increments of rise and fall ( two picture elements ) would be calculated at pixels= 53us/.035 or 1514 pixels. Using this relationship it can be mathematically shown that the horizontal resolution (for character generators) is approx. 1500 pixels to satisfy or equal specifications for broadcast character generators at 35 nano seconds rise time. In summary it must be understood that it is impossible to produce a line in video without rise and fall of the trace and to show a series of differences in the video detail it requires a minimum of two pixels for every line of detail. Otherwise the differences (detail) would be nonexistent. Greg is correct in his supposition that the formula is 2*pixels per line of detail but his relating 1008 to broadcast standard is somewhat inaccurate. Back calculating the ATVISTA 1008 pixels figure equates to a CG with a rise time of 52.5 nano seconds. FYI some Chyrons produce graphics in the 15 nano second range. I can only say that while the ATVista at 1008 pixels is probably state of the art in output it is not equal to a 35 nanosecond CG. for resolution. In addition one must be cautious in using specifications for cameras and VTRs when defining a standard in graphics. Of course, as the CG world changes, there is more and more blending of graphics and texture video maps to blend the two entities. But basically, 2 pixels for each resolution line of video is the rule on horizontal res. In conclusion, a 720 line specification VTR should receive graphics created with a minimum of 1440 pixels to max its resolution capability. In my opinion, your comments suggest that you do need to review the differences between pixels and video resolution as it relates to normal pictures and graphics. I thought the 720 D1 spec relates to resolution not pixels. I hope the above helps you to understand this relationship and is a start in your reading. BTW Greg did not suggest that svhs is higher res than D1. If he did then I missed it and I think he stands corrected. My sources for the above are various articles, and in particular an article on determining resolution by Cecil Smith.