I should also add that if you're going to be using your picture as a still (no panning, zooming, etc.) then don't go crazy and scan it at 1200dpi, "just to get the best quality". You won't. If you end up with a multi-megapixel image then Vegas will have to resize it. If you are using better quality, it will be resampled and this can cause fuzziness. If you're using draft quality it will be point sampled and this can cause weird artifacts and make the image look rather harsh. If you're scanning a standard 4x6" print, there's no reason to scan above 150dpi as the picture is going to end up in the finished video only 480 pixels high anyway. The only thing you'll accomplish with a higher resolution scan is to waste disk space.

If however you want to zoom in, or use panning to have the picture move across the screen, then by all means scan at a higher resolution. It will take some practice and experimentation to find optimal scan sizes. Have fun. Play. :)

Here's a question... The lower "standard" for low resolution on a computer monitor is 640x480 (4/3 ratio). So WHY are some saying 655x480 which itself would seem like a distortion. Somebody explain the reason for those extra 15 pixels.

Well, i can answer your first question twice:
  a) 'cause that's what works to fill the frame (try 640 and you'll see the gap on the sides).
  b) 'cause Sonic Foundry said so.

As to your second question about why it's 655 instead of 640? I have no clue, other than that it does have something to do with aspect ratios.

Yea... but it don't make any sense. What I'm really asking is if the object is to avoid any distortion, then one of the two has to be incorrect. I would assume simply because it has become a stardard that 640x480 is a true 4/3 ratio, so again, I'm just curious about why 655x480 because if 640x480 is exactly 4/3 then adding those 15 pixels will add distortion to the width and make images ever so slightly fater then they should be. No?. True, a minor amount. Anyone know?

Hmmmmm. Now that i think hard upon it, i don't think i've ever heard a definitive statement that the ratio of a television screen is exactly 4x3. Maybe 655x480 reflects the actual size, and 4x3 is just an easy approximation for digital-based display systems.

I just did the following experiment using an original source JPG of 1072x600.

I used Photoshop to make three copies leaving the default 'constrain proportions' which resulted in three images of the following sizes.

1. 640x578
2. 655x592
3. 700x633

I then dropped each on to a empty track adding an overlay on a seperate track recording the image sizes. Playing the mini-video NONE of the images filled the preview window.

I next created a VCD files using the MC MPEG-1 template, burning same with Nero 5.5.
Playing on my TV through a Pioneer 333 DVD player IT MADE NO DIFFERENCE, none of the images filled the width of the screen.

I went back to Vegas Video and this time I clicked on the event pan/crop button changing from the default to the 4/3 standard TV ratio. This time the images was stretched and filled the frame in Vegas Video's preview window.

I burned a 2nd CD using the same template. ALL the images regardless of size filled the screen.

So again, WHY should I change any image to 655x480?

WVG: ummm, where did you get those sizes? None of the three you listed are anywhere close to the ratio of the original 1072x600 image.

Hi-

I believe the answer here is that VV3 is displaying by the DV standard of rectangular pixels. Digital Video uses rectangular pixels. Digital Video is 720 X 480 which on the surface appears to be a 1.5 aspect ratio, but is not because the pixels are rectangular. Look at your VV3 monitor and you can see your choices are 720 X 480, 360 X 240 or 180 X 120. All of them divide to 1.5.

Digital stills and scans use square pixels.

Since VV3 is digital only - it seems they decided to display as if everything were digital using rectangular pixels. Therefore, your square pixel images will not fit in their standard (square) aspect ratio. Instead they are converted to the digital ratio such as 655 X 480.

I am confusing myself now, hopefully this is correct. Someone please explain it if it is not.

Actually they are. I deberately started with an oversize image to ensure enough pixels prior to Photoshop doing the resize. The 'new' sizes are the results of specifying a width and letting Photoshop decide the height based on proportions.

Hope I'm not confusing anybody. <wink> Maybe the width "jumps" to fill the width of a frame at some DVD resolution, is does NOT for VCD resolution which is mostly what I do. So I'm asking what is the relevance of shifting a perfectly good and proportionally correct 640x480 image to 655X480 when it does nothing(without help) and you need to fudge it using the pan/zoom FX button to make it fill the frame as I proved in my experiement... at least for VCD's.

how does 1072x600 become 640x578? the aspect ratios aren't even close.

I appreciate all the interest in this thread. Its funny, though, that I'm not sure that I completely understand it.
The consensus seems to be that by cropping a photo to 655x480, the square pixel computer image will be correctly sized for rectangular pixel DV. (So, by inference, if I crop a photo to only 4:3, then it should have a slight distortion.)

While I understand the advise of using VV to crop the picture, it is actually easier, and creates smaller files, if I do the cropping in PaintShop. Also, the photos need so much touching up that I would rather crop to only the final size that I will actually need.

The pixel aspect ratio issue has been covered a few times, but here's a synopsis.

1) Graphics programs typically use square pixels

2) Broadcast formats use non-square pixels

If you are creating stills in Photoshop or whatever for use in Vegas, multiply the destination format width (in pixels) x the destination format pixel aspect ratio.

Ntsc= 720x480 pixels, pixel aspect = .9091

720 x .9091 = 654.552 (round it up or down to a whole value, to 655 or 654)

for a still image size of 655 x480.

If you look at the project presets, you'll see a table of pixel aspects- these will be different for NTSC widescreen, PAL, and PAL widescreen etc. Use these as your multipliers.

Ntsc= 720x480 pixels, pixel aspect = .9091

720 x .9091 = 654.552 (round it up or down to a whole value, to 655 or 654)

for a still image size of 655 x480.

------------------

So... we should create a still image with a size of 655x480, and it should look 'normal' on our *square-pixeled* computer monitor. that is, until we bring it into Vegas, where it will be stretched horizontally to fill the width of the preview window (which will be at 720x480), so everyone in the pictures will gain a few pounds. but THEN when we print it to tape, we'll be watching it on our video monitor or TV which has tall skinny *rectangular* pixels, so the people in the video will look 'normal' again.

and all we have to do to ensure that people are not too fat or skinny is to look at the *aspect ratio* in Project Properties (first make sure we've selected a template... NTSC DV in this example) and multiply that number by the destination format's width. use the result (rounded to nearest whole number) as the width of still images, and use the destination format's height as the height.

thanks SonicEPM for clearing that up... again... :D

- ben (cheesehole!)

Hi-

At first I didn't think cheesehole was right, but it seems he is. I did the circle test. Streched in VV3, correct on NTSC.

Question: What do you do if the output is to be both TV and Web/Computer?

For web playback you can: redo all the still images, live with the slight distortion,
or set the preview monitor to "display square pixels" and adjust the stills individually so that they look right to your eye.

if you are rendering for web use, your resolution will be something like 320x240 with an aspect ration of 1. so your pictures will look normal (circle will be a circle). the only problem is you may end up with some black at the top and bottom of your frame when you render to 320x240. you could just zoom in a teeny bit with track motion to get rid of the black.

- ben (cheesehole!)

>WHY should I change any image to 655x480?

You shouldn't. There are two defined standards at play here: A VGA screen is 640X480 and a DV frame is 720X480 (NTSC). Throw a round coin on your scanner and to fill the screen with it you'll need to draw a 640X480 rectangle around it. Now if you want that coin to look round on video you must resize it to 720X480 (or let the NLE do it).

I still have no clue where SF came up with the odd .909 pixel size. It isn't a part of the DV spec (or at least nobody's been able to find it yet). Maybe they're trying to include overscan areas?

>You shouldn't. There are two defined standards at play here: A VGA screen is 640X480 and a DV frame is 720X480 (NTSC). Throw a round coin on your scanner and to fill the screen with it you'll need to draw a 640X480 rectangle around it. Now if you want that coin to look round on video you must resize it to 720X480 (or let the NLE do it).

okay I'm not sure this was made quite clear enough so I'll say it a different way.

the reason for using 655x480 has nothing to do with the size of a VGA screen. forget about 640x480 it is completely meaningless to this discussion. the fact that 480 is the same number as the height of DV video is just confusing people. it has no bearing on this issue.

this is the reason you use a 655x480 size for your photos:

a PIXEL on a computer monitor is SQUARE. PIXELS from a digital photo camera are also SQUARE so pictures taken with a digital photo camera look right on a computer monitor.

a PIXEL on an NTSC TV is tall and skinny (a RECTANGLE). NTSC DV cameras use the same tall skinny pixels so that video shot with a DV camera will look correct on an NTSC TV, but not a computer monitor.

the proportions are different. therefore, images from an NTSC TV signal will look FAT on a computer monitor. (my wife was astounded by the size of her ass when she saw herself on a PC monitor! I assured her that it was because of the difference in aspect ratios between pixels on NTSC video and pixels on a computer monitor. she didn't believe me though.)

and it works the other way too. images that look fine on a computer monitor, will look SKINNY if displayed on a monitor with SKINNY pixels, like an NTSC TV.

in order to account for this difference in proportions, we must STRETCH our perfectly proportioned images on the PC just the right amount, so that when they go to NTSC DV video, they will shrink down to normal size, and no one's ass will be too FAT or too SKINNY.

>>>I still have no clue where SF came up with the odd .909 pixel size. It isn't a part of the DV spec (or at least nobody's been able to find it yet). Maybe they're trying to include overscan areas?

I'm not sure where this number comes from either, but it should be the physical width of an NTSC DV pixel divided by the physical height of an NTSC DV pixel. so, if we assume an NTSC DV pixel has a height of 1 unit, then it's width is .909 units. sorry if that's obvious, but it helps me to think of it that way.

also if you haven't checked out this thread:
http://www.sonicfoundry.com/forums/ShowMessage.asp?ForumID=4&MessageID=81847

there is more info, although it's a shame GG's posts have been removed from the thread.

"I still have no clue where SF came up with the odd .909 pixel size."

It is the conversion from 720 X 480 to the square pixel image size of 655 X 480. The math: 720 X .9091 = 654.55 (rounded to 655).