-
Opening an discussion about NTSC: 648×486 vs 720×540
This is one of those things that everyone seems to have an opinion on, myself included:
Here’s the situation: you are making artwork in photoshop (for example) or rendering out 3D for an NTSC 720×486 (0.9 PAR) comp. Working in square pixels is much easier than working at a non-square Pixel ASpect Ratio, as things (like circles) look the way they will when appearing on your TV.
What resolution do you work at?
A LOT of people recomend 720×540. However, if you drop a 720 x 540 image into an NTSC 720×486 (0.9 PAR) comp, the image is too big. It needs to be scaled down to 90% it’s size. On the other hand, if you work at 648×486 (Sq. Pix), and drop it in that same comp, then AE automaticzally adjusts it so it fits perfectly.
So what’s the point of NTSC 720×540. It doesn’t fit into any of the standard non-square comps, and to my knowledge, broadcast never works with square pixels, so it’s not like there’s a special broadcast format that can only use 720×540.
Some people might argue that making it bigger is always better because you can scale down, but I disagree. If you are only hiding 10% of the pixels (By shrinking it to 90%) then you are actually distorting your artwork to some degree. I’ve found that high contrast images that were shrunk down, tend to shimmer as they scale move accross the screen – because pixels that have been obscured (by hiding a percentage of thos pixels)are now being revealed and then obscured again as the layer changes.
So what’s the point of 720×540?
Anyone?
Please?
Thanks.
Aharon
—————————————-
Aharon Rabinowitz
aharon(AT)yahoo(DOT)com
http://www.allbetsareoff.com
—————————————-
Creative Cow Master Series DVD
particleIllusion Fusion Volume 1
available @ http://www.pIllusionFusion.com -
12 Replies
-
I assume we’re talking about static images here, since animating the image in AE renders (hah!) this discussion somewhat moot, in my opinion. 🙂
Well, we know that when going from 720×540 to D1, scaling happens, so pixels are altered. I guess the question is: are any pixels being altered when going from 648×486 square to a D1 non-square comp? If so, how much?
A test might be to create an image using the clouds filter (or some other fractal generator) in PS at 648 square, and one at D1 non-square, using a recent version of PS. with identical settings to the first. Then the two images would be dropped into a D1 comp in AE, with the upper one in difference mode. If we assume that the D1 image is unchanged, any changes to the 648 image would appear as non-black pixels, found visually or by rolling the cursor over the image and watching the info palette.
I’d do the test, but I haven’t upgraded my PS in a while.
If the 648 image doesn’t match the D1 image through the difference mode test, then we need to compare the 648’s pixel changes to a 720×540 image’s pixel changes to evaluate the quality losses of 648 vs. 720×540. I’d imagine that the quality loss would depend on the content of the image (vertical vs. horizontal lines), but I’m speculating.
Does anyone want to run the test?
Steve
-
I think this applies to both still images and video footage you create.
The point of my question is this –
The only alteration to the footage that is 648×486 when droped in a 720 comp is that the pixels are stretched. But any square pixel image is stretched when dropped into a non-square comp. And that’s not really an alteration or loss of quality in the image. If it were, it would be an industry standard to NEVER work in square pixels, since there would be a generation loss right at the begining.
It’s been my experience that woking at 648×486 and dropping it into a 720×486 comp yields results as good or better than 720×540. And less work – since you don’t need to resize. There’s no reason that you should create art or video, and then have to squish it down, if there is a better alternative.
To my knowledge, there is no broadcast system that uses square pixels (I could be wrong about this), or at least not one that uses 720×540 – so what’s up with 720×540? Why do people use it, or say to use it.
I feel like at some point someone said it was the right thing to do, and then people just continued to do so, without thinking about a better alternative.
Thanks.
—————————————-
Aharon Rabinowitz
aharon(AT)yahoo(DOT)com
http://www.allbetsareoff.com
—————————————-
Creative Cow Master Series DVD
particleIllusion Fusion Volume 1
available @ http://www.pIllusionFusion.com -
Your’re right — this applies to still images, 3D CG or video footage, as long as the footage/image remains in one position, one scale.
So what process does the 648 image go through? I can’t see the discrete pixels remaining in the AE comp as 648 discrete wider pixels, so I’d expect that some sort of recalculation is required to turn 648 square pixels into 720 square pixels, since that’s all a computer can display. But if an NTSC monitor is displaying 720 discrete pixels, then those 648 need to be recalculated into 720 anyway … ?
My point is, though 648 may be better (haven’t tested), it may still be subject to some pixel interpolation.
Regarding the industry standard of never working in square pixels, I think we computer graphics guys have never had a choice untill recently. Now, if it can be shown that working in square pixels degrades the image due to the 648-to-720 recalculation/interpolation, then yes, it should now be made an industry standard to never work in square pixels if the image-generating app allows it.
I agree that using 648 in AE, with no resizing-to-fit is easier. I rarely create images in PS, but create them at 648 for that reason. Most of the time I work in AE in D1 comps with PAR correction for previewing.
As for quality loss, maybe 648 is better than 540, but I cannot do a real test, as I stated.
At any rate, whenever I mention creating square-pixel images, I’ll now say 648×486 for simplicity. Maybe I’ll refer the user to a page with a test …
Cheers,
Steve -
“Brian Maffitt at Total Training recommends that resolution, and the Good Herr Maffitt has never steered me wrong. I trust the gent.”
And he knows a lot. I have the DVD’s myself. But I don’t agree with everything on there as being the best way to do something. Not that they aren’t extremely helpful, and an essential to any training library.
“I would rather squeeze pixels than stretch pixels, which is what you do if you work in 648×486, the other popular square-pixel resolution… even if it does fit automatically into 720×486.”
Why is squeezing better than stretching? just because the pixels stretch automatically, doesn’t mean they’re being distorted. Think about it – NTSC DV Widescreen and non-widescreen have the same exact resolution 720×480 (check the comp settings in AE)- They just have different PAR’s – clearly things are being squished and stretched, and yet we work at the same resolution for both. So there really must not be distortion when the pixels are stretched to fit in the proper aspect ratio. They have to be unstreched or squished to get back onto your TV and look proper.
“I just highlight my 720×540 layer, use the command-option-f keystroke combination (mac) and I’m done.”
Right but what you are actually doing is changing the aspect ratio (thereby stretching the pixels) and then shrinking it to fit. If the change in PAR is distorting things, you’ve now done it twice. And you’ve obscured some of your image by hiding pixels.
Of course, the 648×486 resolution is also built into AE, but that leads me to my third reason….
Makes you wonder why it’s there, right. Once I realized that the square pixel equivilent was 648×486, I saw that AE had that setting – I didn’t even notice it before that.
Oh well… The fact is, I’m a big talker, but how much do I really know. Thus this posting to begin with.
—————————————-
Aharon Rabinowitz
aharon(AT)yahoo(DOT)com
http://www.allbetsareoff.com
—————————————-
Creative Cow Master Series DVD
particleIllusion Fusion Volume 1
available @ http://www.pIllusionFusion.com -
I don’t understand why everyone doesn’t just avoid squishing or stretching these pixels by just working at 720×486 D1 in the first place in both after effects and photoshop. You can always turn PAR correction on to make sure that things are looking okay and to preview your work. I know some people don’t like working with the correction on (I’m not sure why because I don’t notice that big a quality loss), but you can always work with it off and give a quick preview every once in a while with it on. I’m probably fairly new at all this stuff compared to a lot of people on the boards, but I thought I’d just find out why people don’t work D1 in the first place.
David
-
“I don’t understand why everyone doesn’t just avoid squishing or stretching these pixels by just working at 720×486 D1 in the first place in both after effects and photoshop.”
The reason I don’t is Because the PAR correction views in both AE and PS suck. Everything is jagged. You can’t really see what it will look like.
I don’t know about you, but when I have 2 days for turn around on a project, I want no surprises. Not being able to see my artwork clearly is something I try to avoid.
Now if they would work out the kinks on that feature, I might be more inclined to do it that way.
—————————————-
Aharon Rabinowitz
aharon(AT)yahoo(DOT)com
http://www.allbetsareoff.com
—————————————-
Creative Cow Master Series DVD
particleIllusion Fusion Volume 1
available @ http://www.pIllusionFusion.com -
Which is the same way I feel about using 720×540 as a resolution. It doesn’t make sense that it’s used, and yet most of the experts say use it. I have not heard a decent explanation as to why, though.
As far as the future goes, I’m looking forward to watching TV in square pixels while my robot drives my flying car for me…
—————————————-
Aharon Rabinowitz
aharon(AT)yahoo(DOT)com
http://www.allbetsareoff.com
—————————————-
Creative Cow Master Series DVD
particleIllusion Fusion Volume 1
available @ http://www.pIllusionFusion.com -
I switch off PAR correction, then look at the NTSC monitor for final output, then switch PAR back on, and so on. Somtimes I work in D1 without PAR, but the NTSC monitor gives me the final look.
Ideally, a clean-looking non-square pixel setup on the computer would be nice, but it’s not there yet, obviously.
Steve
-
“A 720×540 comp has the potential for containing more information than its 720×486 counterpart.”
By that token, why not work at 800×600, or larger?
What do you mean when you say “the potential for containing more data? Can you explain? What are you going to do with that extra data – doesn’t it get tossed out anyway?
“So using the principle that it’s better to toss away information than it is to invent new information, I chose 720×540. “
But I feel like more is less, in this case. You end up having to throw away pixels, instead of setting it up right and not having to lose any data at all.
I guess it’s a preference – I’m trying to understand how the whole thing got started, and if there is an advantage or disadvantage in doing it one way or another. I don’t mean to be confrontational, at all.
Thanks.
—————————————-
Aharon Rabinowitz
aharon(AT)yahoo(DOT)com
http://www.allbetsareoff.com
—————————————-
Creative Cow Master Series DVD
particleIllusion Fusion Volume 1
available @ http://www.pIllusionFusion.com -
After googling pixel aspect ratios and reading a bunch of articles on it and going back and reading the Dr. Strangepixel article, I think I’m starting to get the hang of it all.
However, there is one thing I don’t understand. Why on our computer monitors do D1/DV pixel compositions appear stretched out horizontally instead of squished horizontally? If rectangular pixels are longer horizontally than square pixels, than wouldn’t the rectangular pixels have to be squeezed horizontally in order to fit into the square pixels that make up our computer monitors, and therefore squeezing the picture?
Sorry if this seems like a stupid question. I’m just one of those people that even if I know how to deal with a situation (such as how to handle square and rectangular pixels), I have to know the “why” of how it all works.
David
Reply to this Discussion! Login or Sign Up