NTSC Pennant?

[old_tv_nut]

Originally Posted by Pete Deksnis

Cool! I had not noticed cross color in the awning before your pointing it out. Neat simulation.

Pete

Note: most of it is in the I channel, and disappears with narrowband demods
I'm out of town and will post when I get home.

[old_tv_nut]

Originally Posted by W.B.

It seems, based on the colors as registered in this JPEG file, that 205° would be the RGB position for -I, 24° for +I, 275° for +Q and 95° for -Q, would that sound correct? (Being that RGB values are different from the phase positions in video language.) I had done my own research of position values in Photoshop and Corel Photo-Paint 11, and had come up with:
- 204° for -I
- 24° for +I
- 272° for +Q
- 92° for -Q

I'll have to check out the code when I get back home. This simulation was done in Photoshop by writing custom Photoshop filters to do the matrixing, chroma modulation, filtering, and demodulation, would you believe? I just love perverting software to do things I want that it never was intended to do.
By defining the +/- I and +/- Q subcarrier peaks to fall on pixel locations, modulation and demodulation can be simulated by multiplying by a pattern of
(+1,0,-1,0) or (0,+1,0,-1).

The +I channel should be at 33° electrically with respect to scaled R-Y and scaled B-Y, and the other three phases at 90° increments. You have to be careful that you are working with properly scaled R-Y (divided by 1.14 if I recall correctly) and B-Y (divided by 2.03 if I recall correctly) before converting to I,Q. The values you found may be correct when calculating angles in the RGB domain, that is, with straight R-Y and B-Y.

[W.B.]

Originally Posted by old_tv_nut

I'll have to check out the code when I get back home. This simulation was done in Photoshop by writing custom Photoshop filters to do the matrixing, chroma modulation, filtering, and demodulation, would you believe? I just love perverting software to do things I want that it never was intended to do.
By defining the +/- I and +/- Q subcarrier peaks to fall on pixel locations, modulation and demodulation can be simulated by multiplying by a pattern of
(+1,0,-1,0) or (0,+1,0,-1).

The +I channel should be at 33° electrically with respect to scaled R-Y and scaled B-Y, and the other three phases at 90° increments. You have to be careful that you are working with properly scaled R-Y (divided by 1.14 if I recall correctly) and B-Y (divided by 2.03 if I recall correctly) before converting to I,Q. The values you found may be correct when calculating angles in the RGB domain, that is, with straight R-Y and B-Y.

I found that +(R-Y) in RGB domain was 340°, while -(R-Y) was 160°; +(B-Y) was 250° and -(B-Y) [a.k.a. NTSC 'burst'] was 70°; and G-Y=0 values were 43° and 223°. I also found the +/-I and +/-Q values through taking the YCbCr calculations in color bars and changing the luma from 16 to 126.

I am aware that RGB values are different from what is registered on color vectorscopes.

[old_tv_nut]

Just want to point out to those reading the back and forth here about hue angles, that the Photoshop hue doesn't necessarily bear a direct relation to the NTSC chroma phase angle. The color produced on the screen for a given NTSC phase angle varies depending on the brightness and saturation of the color, due to the CRT gamma characteristic (non-linear response of light output vs. drive voltage).

This illustration "NTSC Color Television Standards" shows how the hue varies with saturation for a given phase.

+I is at 123 degrees (electrical) which clearly is not a straight line on the chromaticity diagram.

[old_tv_nut]

By the way, the non-linear effect on hue for a constant electrical phase is plainly visible in the image with Q turned off - compare the "yellow" color bar with the "red" one, or look at the difference between the "green" "cyan" and "blue" bars. All of these (like every other part of this image) have signals only on the +I or -I electrical axis, but clearly have different hues on the screen.

[old_tv_nut]

Here is the way the picture looks with Q turned off and a flat gray substituted for Y.



Note 2 things:

1) the hue is much more constant
2) the luminance is not constant, however, because the non-linear display causes an increase in luminance with increased chroma

[old_tv_nut]

Here's the narrowband decode version I promised:

[Phil Nelson]

Back to the pennant, does anyone know what the five little pennants are supposed to represent? The illustration's caption does not mention 'em, nor does the text on page 597 that references the illustration.

Phil Nelson

[old_tv_nut]

Originally Posted by Phil Nelson

Back to the pennant, does anyone know what the five little pennants are supposed to represent? The illustration's caption does not mention 'em, nor does the text on page 597 that references the illustration.

Phil Nelson

They were used as printer's devices to tick off highlights of the contents on the front cover of the IRE Proceedings.

[Phil Nelson]

Gee willikers, you guys know everything! Can I have your magazine?

Seriously, thanks for the pointer. That journal content appears to be available at http://ieeexplore.ieee.org/ . My brother-in-law is an IEEE member. I'll try to bribe him into getting those articles for me.

Phil Nelson

[Phil Nelson]

Got an interesting email today. This guy appears to have the original painting used for the IRE magazine cover. He bought it at an auction years ago and had no idea what it was (or why the row of little pennants) until he read the article on my website.



He says it's done with gouache on illustration board. On the back it is stamped "Associated Freelance Artists" from NYC, with another label that appears to order halftone color prints.

I asked him if he's able to provide a better quality photo, which would make a nice footnote for the article. Incidentally, I did eventually get my hands on a copy of the IRE magazine (thanks, John!).

Phil Nelson
http://antiqueradio.org/index.html

[cbenham]

Originally Posted by Phil Nelson

That seems likely. The plate in this book just reproduces the IRE illustration, which certainly isn't in TV format.

I think I have seen a recommendation from you for this book, in fact -- Basic Television by Bernard Grob.

I have been using this as a computer desktop background and thought someone might happen to have a better version. The scan I made looks kinda fuzzed when blown up to 1440 x 900 resolution. Still jolly, though. Phil

I found two versions in my files and I think they both came from attachments on Videokarma. Cliff

[ChrisW6ATV]

Here is another scan from that same book by Bernard Grob. I didn't realize it was discussed here until I re-read this topic after randomly picking up a book on my shelf, seeing the flag in it, and remembering this discussion.



If anyone wants it, I have an 8MB 5000x3000-pixel JPEG of this version, or the original 52MB uncompressed TIFF of the whole book page.

[West]

Wow! The things a person can learn from you guys! ...and I'm blown away by all the efforts expended on crafting these images. Interesting!

I and Q signals, computer generated subcarrier effects, NTSC flags...this is truly a world of it's own.