There's three mixed up concepts in there.
Color Space is I think Gamut, the range of colors the system can reproduce, NTSC has a slightly larger color gamut than PAL, think of the total color gamut or space as a circle, for practical reasons different system may limit the scope to a smaller area than the whole circle, many good articles that a Google will find.
Color Sampling refers typically to the use of YUV rather than RGB systems, again for practical reasons less of the color data than the luminance data may be sampled.
Most of the problems of graphics rendition on video systems are related to out of gamut colors (probably not a problem in this case, as there's only one color it'll simply be pulled back into gamut) and transition rates, particuarly on composite video connections which may ring, overshoot or undershoot at the transition, that's why adding a small border at 50% of the two adjoining values helps, by smoothing out the transition. In general the best way to avoid this (apart from keeping levels legal and colors within gamut) is to use adjoining colors from the same part of the color wheel e.g. red and blue would be BAD.
Bob.
encoded video, and by this I mean analogue non-component formats (like SVIDEO and COMPOSITE) also have format pecularities. Remember the dog tooth jackets strobing away, or indeed red against a black background. Fast and sudden transitions of luminance can have a chrominance effect for Y/C or composite video.
Many of our broadcasts don't suffer this today as set top boxes have a component hookup (YPbPr-rca-phono or RGB-scart) or are built-in to the telly. So along with the DVB signal - PAL has been enhanced through the digital age. Not that it was struggling behind NTSC...
more later if nobody else steps in.
legalise is more than sizing and luma/chroma dynamic range
Grazie, is your problem that the red "bleeds" onto the black background, so that you see this blocky edge on the right side of red text?
I believe this is because of DV's 4:1:1 sampling. If you look at things horizontally, DV has 1/4th the color resolution than it does for luma information (think of luma as black and white). Technically the codecs and converters are supposed to interpolate or smooth out the color information so that it's not chunky. In Vegas, you can approximate this via chroma blur horizontal=2. You'll see the blockiness disappear.
But when you go back out via DV, the camera usually doesn't apply chroma interpolation.
If going out to DVD, it uses 4:2:0 color sampling. So chroma blur will help a little. 4:2:0 Color resolution is half that of luma horizontally, and half that of luma vertically.
2- The other part of it is that color in DV is stored as color difference components. An illustration of this would be to apply the color corrector filter and to lower the gain. The gain lowers the luma values by multiplying it by the setting. Notice that everything becomes more saturated.
What happens is that the chroma values are the same, but they only give the correct color for a particular luma value.
In DV's 4:1:1 sampling, think of everything as 4-pixel segments. The first pixel has the right luma value and the right chroma value. The other pixels use the first pixel's chroma value. But these other pixels don't have the same luma value, so they will look a little off.
*DV actually stores luma and chroma, which are like actual luminance and chrominance but not the same thing. Charles Poynton's book goes into this (poynton.com). His book will tell you everything as long as your eyes don't glaze over first.
Grazie is DV PAL, so is already 4:2:0 - which'll be the same as DVD-Video worldwide.
It could still be colour sub-sampling that is affecting him. Or the quality of cables or bad termination if a broadcast-spec monitor-TV is being used without the right additional shunt or termination mode set on the back or via the menu.
If it is an analogue thing then it may be a levels issue. Have you used the legalise filters Grazie? They do some of the range squishing that fits the broadcasting standards that also accomodate for some of the inadequacies of TVs used as monitors or monitors in need of fine tuning or calibration (field or lab calibration).
If the red always veers to the right and fades away, or appears to, then the chances are that this is a signal ringing out due to cables or poor proc-amp circuitry. The best way to check is to verify this between different setups. DV preview of the Vegas timeline is a good start.
Ringing and power regulation (picture dimensions changing with step changes to overall picure luminance levels) are a common thing to find in most consumer kit and some prosumer/pro kit. The LCD monitor combats some of these and brings some different linearity curves that you can only hope are engineered correctly.
If the smudging is anything other than rightwards, then something more NLE centric is going on. IMHO.
4:2:0 doesn't just affect red- so adjust the intensity, and adjust the background to see where the point of failure occurs. Also review the written specs for on-screen graphics to overcome the broadcast limitations of encoded (analogue PAL down a single co-ax) video and picture-on-sound issues (when RF modulated). Some of these rules are still worth considering when buoding up computer generated media segments like titles. Naturally acquired footage usually works quite well with PAL.
NTSC is also quite capable of rendering leak free video if you choose your palette, dimension and motion path carefully. I'm harping on about PAL so this thread doesn't push the problems that can occur with mixing 4:1:1 with DVD target workflows. That isn't the case for Grazie in the UK, 95% of the time!
Grazie,maybe try using the ChromaBlur filter on the DV AVI text overlay. If you have Ulta, input the the two fields and use the spill feature for the red bleeding.
For what you're talking about, Glenn's answer might be the most useful for you. I'll elaborate on it with an example.
Suppose you have a line of 5 samples (pixels). Each pixel has a definite brightness value. The first and fifth pixels have definite color values. (This is 4:1:1 sampling, what NTSC DV uses.)
Now, between the first and fifth pixels are three pixels that have no red and blue values, just luminance. When DV is decoded the missing red and blue values are calculated. If red is "16" at pixel one and "235" at pixel five, then pixels two, three, and four are going to be approximated based on those two known values and the known luma value of the actual pixel in question.
The short of it is that, in any lossy video compression scheme, the color values of some pixels will be "guessed" based on other nearby values. This results in more gradual color transitions than you might have if every single pixel were a known value. This phenomenon of throwing away color data at render time and then guessing the values later at decode time is what makes titles in DV look bad when the titles rely on color differences instead of luma differences. White titles on black will look much better than deep red or blue on black.
This is why titles are so time-consuming and why I save a veg of every successful title-color-blend I get. It usually takes me 10 or 15 minutes of tweaking before I find a successful combination of font, title color, shadow and/or border that works on my television (which is my rinky-dinky monitor tweaked with color bars).
For smaller fonts I stumbled onto Chroma-blur awhile back mentioned by Glenn Chan and it can save some marginal stuff.
P.S. offtopic, but I love the look Glenn Chan's S-curve for film gamma gives my video, a more rounded deeper detailed look than the original.
Apologies for not getting back earlier - the Earth does keeps spinning yer know? LOL!
Woah! - I have a lot to read, and a lot of experimenting to do now. Yes I'm PAL and I guess that's 4:2:0. In another thread there is a suggestion to use MPEG as the render to get a better red/black combo. Might try that too.
Paul? I like the idea of keeping templates of that which works. I wonder if a script would give this numb skull an easier way forward?
Unless I have missed something (which is possible), there hasn't been a single conformant NTSC TV manufactured for consumer use since the 1950s.
NTSC has a very ambitious color space, and after it had been implemented accurately by the early adopters (at great cost), manufacturers cut corners continuously until the square was well rounded, i.e. cheap.
The 709 color space in the new DTV/HDTV standard is much nicer, with rich colors that put at least any currrent NTSC TV to shame.
Grazie, I just save the project I'm doing under a different name, then get rid of all irrelevant media except the title and any needed background, then save it under my titles folder. I've come up with what I thought were some pretty cool titles after lots and lots of tweaking and I didn't want to lose those after all the work.
Later on, of course, I can pull it in as a nested veg above a new project, change texts, tweak some more etc.