I would normally create the graphics image as 640x480 (4x3 aspect ratio) in Photoshop, and as a final step, resize to 720x480 for use in a Vegas NTSC project (the equivalent resolutions for PAL images are 768x576 and 720x576). This means that you can maintain the aspect ratio in Vegas and not have to worry about geometric distortion if you want to zoom in or out on the image.

Richard Hunter

Actually, it would be better to create the graphics at 720x528 instead of 640x480 (or 655x480.) When Vegas resizes the 720x528 image, it is "condensing" the image information, when Vegas resizes a 640x480 (or 655x480) graphic to 720x480 it is having to create information that was never there to begin with. Reducing a large image to fit always looks better than enlarging a small image.

Also, keep in mind the aspect ratio difference between square-pixel computer images and rectangular-pixel DV images.

John

Vegas likes graphics to be 655x480 for NTSC. I think I saw that in Spot's book.
Good luck, Terry

Can someone please explain again why Vegas is the only video app in the world that wants graphics to be 655x480? Doesn't Vegas upsample these? Wouldn't it be best to design at 720x540, resize to 720x480 and convert graphic to DV pixel aspect ratio in the media pool?

Luxo,

Yes, absolutely, that's the way it should be done. Downsampling ALWAYS looks better than upsampling.

John

I feel sorry for you guys because you're getting a lot of conflicting info.

Lamont Dennis, you can do this in a few ways but the simplest way in Vegas would be to create your own stills on a 655x480 canvas (assuming NTSC) and leave the backgrounds clear. If you compose on a 655x480 canvas then Vegas will always get the size right and it'll make less work later on.

Personally, i'd want to leave the files in psd format with their layers, especially if you have text because you can always go back and change the text layers.

You've got a small problem with the 720x480 images from Avid because Vegas will try to "correct" them. There's a way to change the way Vegas treats the PARs of a particular file type (described below) but that may cause a problem with your own stills if they are also in psd format. Here are some options:

Option 1: batch resize all of the Avid stills to 720x528 before droping into Vegas. Sizing UP might avoid throwing picture data away. Vegas will then size the stills back down to the the right size.

Option 2: Batch resize all of the avid stills down to 655x480 before dropping them into Vegas. Vegas will adjust everything and the images might be just as good as if you sized up. (depends on how vegas does the resizing task)

Option 3: Set up Vegas to treat PSD files as 0.9091 PAR. Leave the avid files alone and drop them into the project. Compose your own stills on a 655x480 canvas and then save as png (you can't use psd format in this case). Drop them into the project as well.

Option 2 is the simplest, at least for me.

===================================================

Most NLEs want you to import graphics that are the size of the frame, 720x480 in NTSC DV. Vegas is flexible enough to take in graphics of any size but it will resize the image until it fits in the frame and fills the frame as much as possible.

Vegas also automatically adjusts the aspect ratio of the still image. Normally you woudn't need to adjust the aspect ratio before importing. Thus, if you compose on a 655x480 canvas and then resize it to 720x480, Vegas will fit that image into the frame and adjust the PAR again</>, which is not what you want.

It would be nice to have a clarification about this by the devs...

Which "this"?

Here are some links to go look at regarding PAR and frame size:

http://www.uwasa.fi/~f76998/video/conversion/

http://www.activeservice.co.uk/video/pixels/page2.htm

http://www.mir.com/DMG/aspect.html

http://www.lurkertech.com/lg/pixelaspect.html

Someone needs to explain why you cannot use the Photoshop CS NTSC DV settings (720x480) and specify a DV pixel aspect ratio (.9).

Seems like that is the right way to do it to me. This is new in Photoshop CS. I assume that Photoshop allows you to generate the image using square pixels and then scale the image to the .9 pixel aspect ratio at 720x480. In Vegas you just select .9 as the pixel aspect ratio in the image properties. Works for me.

In the book, Photoshop for Non-linear Editors, there is a section on integrating Photoshop with Vegas.

It is written by Sony's Dr. Dropout.

The file sizes he recommends are
NTSC 655 x 480.

He says the formula is pixel size of the destination frame times the pixel aspect ratio.

720 x 0.9091 = ~655.

I still maintain that creating the graphic at 720x528 and allowing Vegas to downsize it is preferable to letting it stretch 655x480 to 720x480.

John

John,

I agree with you, at least hypothetically, But I think this calls for a test.

Here's what I suspect will happen when you import that 720x528 image:Vegas resizes image to 655x480, then vegas stretches it out to 720x480.

I'm not saying this is what happens. I'm saying that this is what might happen. If so, you'd get nothing out of working at that resolution and then letting Vegas handle it. In fact, Vegas would be throwing away pixels and then adding new ones. That's the worst of both worlds.

How to test? I'd imagine that if you put some thin vertical and horizontal lines on the image you could see what happens after vegas "handles" it.

(Time passes...)

okay, I did some tests using Photoshop CS (incidentally, I don't usually use CS because my shop is using 6.0 and I need to stay compatible). Here's what I did:

-Make 720x480x0.9 PAR image in CS and then lay down about 30 1px lines horizontally and vertically, each with 1 px spacing

-Make 720x528x1.0 PAR image in CS and then lay down about 30 1px lines horizontally and vertically, each with 1 px spacing

-Make 655x480x1.0 PAR image in CS and then lay down about 30 1px lines horizontally and vertically, each with 1 px spacing

-Finally, I manually resized the 720x528 image (in Photoshop) down to 720x480.

What you'd expect to see in Vegas is good clean lines on the 720x480 image and banding on the other images and in fact that's what you get. You also get banding in the image that was resized in photoshop, of course.

However, I think that if Vegas were taking my 720x528 image, scaling it to fit, and then correcting the aspect, we'd see banding in both the horizontal and vertical lines. I'm very pleased to report that we don't. We only see banding on the horizontal lines and this says to me that Vegas is actually very smart about the way that it corrects PAR and resizes.

The first thing to point out is that Photoshop CS gave the best results. I was able to work in CS with aspect correction turned on, place lines 1 px apart, and have it look right in Vegas. This is probably the way to go as long as you set Vegas up to treat PSD files as having a 0.9091 PAR.

If you don't have CS, you can pick your poison and be content to know that one image dimension is probably as good as another. For 655x480 images, Vegas must create about 65 new pixels on each line. For 720x528 images, Vegas must throw away 48 pixels from the vertical dimension.

Objectively, 720x528 ought to be a better still image size. I'm not sure if I could tell the difference subjectively, though.

I think that there may be good reasons to stick with a vertical constant of 480 px. For one thing, pixels are just samples over time and vegas works on a sample level, methinks, so it might be that Vegas will just perform better with 655x480 images. If that's the case, it ought to perform even better with images in 0.9091 aspect.

I can't post my test images here and I'm too lazy to figure out how to post to my ISP, but I can mail you some GIFs of the images, if you'd like to see what I was working with.

Oh, yeah. So what would be the best thing to do with exporting stills from vegas? Use the clipboard and keep the 720x480. Otherwise you'd be throwing away pixels and then recreating them later when you add the still back to the timeline.

Rob Mack

Rob, if you wanna send your images to testbench at vegasusers dot com i'll zip them up and put them on the testbench download page for folks to peruse.

Okay, Bombs away!

Rob

http://www.vegasusers.com/testbench/files/rmack-vegas_still_image_test.txt
http://www.vegasusers.com/testbench/files/rmack-vegas_still_image_test.zip

Interesting thread. Bob, I looked through some of the references you linked to in an earlier post, and have a question for you.

Since the DV frame aspect ratio is 4:3 at a line pixel count of 711 (approx), and the pixel aspect ratio is 10/11, why do you use 654.5 (720 x 10/11) as the line pixel count for square frames instead of 646.4 (711 x 10/11)?

The banding in the resampled images that you posted is very revealing, I shall make sure to look for a less lossy conversion method in future.

Richard

Where do you get 711? A DV frame is 720x480.

Part of the confusion I think comes from the active area of a DV frame, try capturing from an analogue source and you don't get a full frame of DV.
But anyway here's the really simple method I use.
Set the preview window to display a full frame with Simulate Device Aspect ON, Save a frame. Open saved frame in PS, that's the size we work from, hasn't failed so far for 4:3 & 16:9 PAL or 4:3 NTSC.

Or you sould save a few steps, just remember the numbers at the bottom of the preview window!

Bob.

Bob, In any case, it's the size that Vegas wants us to use, and that's really what matters here.

Absolutely!

655x480 is the size that will give you the least hassle. It may not always yield the best quality for a freeze frame but if you're bringing stills in from scans I don't usually see that much difference.

In my daily routine I look at a lot of shots of electronic components and pick out shots for stills. I can tell you that I loose a lot of detail when I resize the grabs down to 655x480. For video, this sort of fine detail would make the picture vibrate so it's questionable how much of that detail you need to preserve.

Jaegersing, are you sure you aren't addressing this to me, Rob? I see that Bob hadn't posted yet. Let me know where you saw a reference to 711PX for a 4:3 window in the NTSC frame. My understanding is that it should be 704px.

Rob "don't call me Bob" Mack

As usual, an interesting thread. The one thing that amazed me a few years ago while trying to do the math on the optimum pixel count was to learn that DV is NOT 4:3. That's why you get the narrow black bars to the left and right of anything 4:3 you import.

To put a finer point on it, nothing in NTSC (and I asume PAL as well) is 4:3.

A 4:3 area is what you should always be able to see out of the entire signal. Analog signal, especially with tubes, needed more time in the signal before and after that 4:3 window in order to bring the signal down to 0V and then back up from 0V.

You can see these soft ramps at the sides of anything you capture from a VHS deck. It's allowable slop.

Rob Mack