Hi Chienworks. What you say is in line with the way I have been working with DV for several years, I always assumed the DV 4:3 frame aspect ratio is for 720 pixels per line. However the reference given by Rob also makes sense. Basically, in order to view that DV frame on a CRT monitor, you have to consider the sample clock frequency and this equates to 711 samples (pixels) per line. I now believe (but stand to be corrected) that to make things work out, you have to start with a 4:3 image at 646.something (711 x 10/11) pixels per line, and then pad it out to 654.5 for Vegas to stretch to the DV project size of 720 pixels. Whether it is worth the effort to do this in real life I'm not sure, but it looks more accurate to me from a navel-inspecting point of view.

Richard

Look harder and read further. Specifically, I remember a paragraph in one of those links that talks about why the sample rate was chosen and why there was a concerted effort to make sure that both PAL and NTSC sampling yielded 720 samples per line.

The number really is 720. Now I'll go back and read that doc. To tell the truth, it's been a while since I read them thoroughly an I just skimmed them before providing the links.

(time passes...)

Okay, it appears that I was worng (proper spelling, considering) about the window of an NTSC frame that is 4:3. It's 711 pixels out of the total 720, according to the referenced artical. 702 for PAL, again according to that source.



(more time passes...)

It's all pretty comic. The 711x486 number is talking about sampling of the real true entire frame (720x486). However, DV25 only samples 720x480 and a 4:3 segment out of that frame would be 704x480.

How much does this really matter? In practice probably not much but the essential point we started at was to say that 720x480 doesn't represent a 4:3 image. For our purposes, we don't need to know much more.

Rob

Hi Rob. If we are talking about analogue video derived from DV, the 720 figure is nominal and not precise. The actual figure is the number of samples you can fit in one line, between the H sync pulses. For NTSC DV type video, this is 711 samples (PAL is even worse, it is 702).

704 has also been commonly used as the number of pixels per line, but it is also nominal, chosen because it is the next multiple of 16 pixels down from 720.

I suppose one way to try and prove this is to create a single-pixel border, 720 x 480, at the outline of a DV-sized frame in a Vegas project, then output this frame to an NTSC monitor. If I am correct, then you will never be able to see the left and right hand edges of this border on a CRT monitor (even one with an underscan/overscan setting). Only if you increase the border thickness to 8 or more pixels will it become visible. (I haven't tried this out for myself, but this is what I would expect to see.)

Richard

Richard, so you're saying we should create an image that is 646x480, then pad it out to fit, which ends up with 654x480. So isn't your final result what we have been saying all along? Does it matter if the extra 9 pixels are black padding or more of the image? We're not worried about filling the visible part of the image. We're worried about making an image who's aspect ratio is still correct in the final video. In order to do that, start with an image proportional to 654.545454x480, whether that includes padding or not.

I'll also suggest that if you are viewing your video on a computer which shows the entire frame that you'll want to use the full 720 pixels rather than 711 with padding because the entire 720 will be visible! I know this is true. When i create animations at 654x480 and import them to DV, the playback on the computer screen shows the entire frame i've created stretching the entire with of the window. None of the image i created is missing, and there is no empty space on the sides. So, the 711 (or 704) figure doesn't even apply to digital playback.

Hi Chienworks. Yeah you're probably right. Thinking it through again, multiplying 711 x 10/11 to get the image size doesn't really work, presumably because the 4:3 aspect ratio for NTSC DV frames is not based on 480 lines. I guess there will always be some geometric distortion in practice. Back to studying my navel....

Regarding viewing on a computer, yes all of the 720 pixels will be visible. The sampling clock frequency only applies when there is a conversion to/from analogue involved. Actually, even if you are going to view only on a CRT, you should also fill the entire 720 pixel width of the DV frame. This is so that if you do any page peels, video rotation or PIP in the project, you won't see any unwanted black edges in the final output.

Richard

Give it a try.

I've done this sort of thing with a line at the border and I'm pretty sure I've been able to see it on underscan but I don't have a monitor here to check that.

As I understand it, getting all 720 samples recorded onto an analog tape is mostly a matter of response time in the analog devices. I think you actually can get the whole thing onto a tape with some hardware.

According to the article, 711x486 gives you a 4:3 image. 704x480 also gives you a 4:3 image and is more commonly used because A:) it divides by 16 and B:) It doesn't include the outer 6 lines-which aren't included in DV25 anyway.

In a way, it's not a point to belabor. DV25 is 720x480. The important point to remember is that that number doesn't represent a 4:3 frame and so one shouldn't try to make calculations as if it was 4:3.

Rob Mack

Exactly. You do, indeed need to fill out the entire frame.

The initial question was whether to use 720x480 stills or to just use whatever you have and throw it on the timeline.

The answer is that you can use 654.5454...x480, 720x528, or even 720x480x0.9091. You can also use any image size you want but Vegas will make certain assumptions about it that may well be wrong and create extra work. The simplest format to create stills in is on a 655x480 canvas. However, you can also use 720x528 and Vegas seems to handle it just fine. This size has the added attractions of not being a fractional size (like 654.5454x480) and the aesthetic satisfaction of using all the image pixels you possibly can. There is a very real probability though that if you use this size you'll take a hit to render performance. (Yet to be tested but it seems reasonable.)

The one place where it pays off to use 720x480 is when you export a frame to use as a freeze frame. It is better, quality wise, to put the frame on the clipboard and then use Photoshop to make a 720x480 image. Using Vegas' save image function will reduce the frame to 655x480 and definitely reduce the quality of your freeze frame.

Rob Mack

Now that the smoke has cleared...

NTSC DV has a frame size of 720x480.
The graphic that I use to 'Fly-In' & 'Fly-Out' is NOT 720x480. Its actual dimensions are 250x426.

To clearify:
I was sent a 250x426 graphic as a .psd created on a 720x480 photoshop canvas. All of the area outside of the 250x426 is transparent. Kind of like a PIP when you're watching the news and they have the NewsCaster talking and a PIP beside him with a picture or video playing. (Imagine flying a Business Card into and out of a video. - I don't want this to be full frame because my talent is still talking and visable at the time. I do NOT want to cover her. I just want the graphic/logo to 'Fly-In' when she mentions the company's name then 'Fly-Out' afterwards.)

My question is:
Should my 250x426 graphic be created on a 720x480 canvas that has every part outside of 250x426 transparent or just create my graphic at 250x426?

I am not concerned that a full NTSC DV frame is 654.5454x480 or 720x528 or 655x480 or 640x480 or 720x540 or any other size. I have determined that this is NOT the place to ask what is the FULL NTSC DV FRAME size, nor do I want to debate the FULL FRAME size of NTSC DV.

I would just like to know... Should 'Fly-Ins' such as the ones I am making be created FULL FRAME size or should they be created in their actual dimensions?

Thanks to all that have input.

...bump

Over 30 people had something to say before my clearification. What gives?

Exhausted?
No thoughts?
No suggestions?

I think I would just try it and see. I can't quite see what the big debate is over this. You have a graphic - just use track motion to position it how you want - add the transition - and bingo - your done. Seems to me you could have done this quicker than it took me to type this. Am I missing some part of the puzzle here?

I think there has been enough posted here to answer the question many times over. You can do it many ways. Pick one.

Understanding how it works might help you make the choice.

Personally, I'd take the graphic they provided and drop it onto the timeline. Then you can adjust it by either resetting the media's PAR or you can use the "Match Aspect" tool. I'd assume that they gave you the graphic that way because they wanted the graphic positioned in that spot in the frame, allowing that it would move in one axis.

Photoshop CS seems to make this all pretty simple as long as you know what's going on. The whole point of working on a full frame canvas is that you can place things where you want them in the frame. You can also export stills from Vegas to drop into this composition to help position text or graphics. Then you just turn off all the layers you'd rather not see in the final image.

Rob Mack

I once spent ages preparing around 100 stills for use in Vegas. I cropped each one to the exact number of pixels that would make it look the way I wanted on a DV video background. This was before I found out that Vegas resizes everything to the project frame size and the only choice you get is whether to maintain the aspect ratio or not. Resizing all of these still in Vegas using track motion was a real PITA and I don't think I got back all the exact same image sizes I wanted.

Anyway, if I had a similar thing to do again, I would probably take the PS files with their DV frame-sized background, and then just use track motion in Vegas to move them around. However, if I only had one or two stills to worry about, it wouldn't really matter what format they were in, as resizing a couple of stills does not take THAT long. For me, it would all depend on how many I needed to process.

Richard

Exactly.

Rob