So, if NTSC standard resolution is 655 x 480, 1310 x 960 pixels would be a good starting point.

Laurence Kingston

What about compression? I know that since the final mpeg render is going to recompress the images, it seems natural to avoid compression as much as possible on the original scans. On the other hand, compression makes the files a whole lot smaller and easier for a hard drive to handle, and at the relatively low resolution of 655 by 480, I doubt that compression artifacts are really an issue. I know images I take with my digital camera seem easier to work with than my scans, and they are compressed quite a bit. Is there some general wisdom I can live by here?

Laurence Kingston

Actually compression artifacts are more noticeable with smaller size pictures. Of course, with smaller size pictures you don't need as much compression. I use Micrografx Picture Publisher to edit images and when saving JPEG files it uses a compression scale from 1 (very little compression) to 100 (much compression). I've found i can usually set this at 20 and have pictures that look very good. Most 654x480 pictures will run between 20K and 100K bytes at this setting. If i really want to avoid artifacts i can set this at 5 and get a file size between 40K and 200K and the results are almost identical to the original scanned image.

What scale you have for compression depends on the software you're using. If your picture files are ending up over 250K for a 654x480 picture then you could probably stand to compress a lot more (an uncompressed image at this size is only 920K anyway). If your pictures look distorted and have noticeable artifacts then you should use less compression. Do some comparison tests with different compression levels and compare the results of both the stills and the rendered videos. You should be able to judge the proper trade-off when you see the results.

Hi Laurence,

My 2 cents ... If you can, NEVER work with a compressed file. A .jpg photo is a compressed file. If it is a question of HD space, I can understand. But otherwise, a .tif, .png, .psd, .tga file will be better threw editing.

Gilles

PNG is a compressed format. TIFF, PSD, and TGA can also be compressed.

I understand the reasoning behind not using compressed files: that you are uncompressing and recompressing the images and that you get artifacts of artifacts. On the other hand, 655 x 480 is pretty darned low resolution. It seems to me that a compressed image with three or four times more pixels than the NTSC screen can show is hardly going to suffer from artifacts that are so much smaller than the NTSC can reproduce. It also seems to me that a high resolution picture with somewhat heavy compression would allow quite a lot of flexability in terms of zooming way in whereas a high resolution uncompressed image would tax your hard drive too much to get smooth real time renders of the crossfades. I could do a bunch of experimenting to get the best compromise numbers right, but I figured somebody else must already have done this (thanks Chienworks). I'm less interested in theoretical ideals than I am in real world rules of thumb I can go by.

Laurence Kingston

Stick with PNG. It is compressed but its lossless compression unlike JPEG.
VV would only need to uncompress once anyway. What you don't want to do is say use PS to edit an image save as jpeg, edit that, save as jpeg etc. You will start to get losses.

If you're going to zoom in on a still then yes, more res is what you need. as I said as a rule of thumb if your zooming in 50% then you'd need double the res of a frame.

Video is very low res compared to stills, but the brain makes up for that because the image is moving, each frame reveals more of the detail as the camera / subject moves.

Composite video is extremely low res in terms of color, don't know the exact figure in pixels but its about 20% of the luminance component. Next time you watch TV have a close look at say some grass, all you're actually seeing is monochrome grass with a green wash over it.

I wish I had 20% grass so I didn't have to drag out the lawmower as often.

If you don't drag out the lawnmower you'll soon have 120% grass!

As Farss has said, there's a difference between png compression and jpeg compression. PNG is lossless and won't create artifacts. It's a good choice with Vegas.

In theory, compressed stills require more CPU effort to uncompress them. Given that, compressed stills aren't your best choice. However, Vegas works great with PNG files.

Rob Mack

While .png is a good format, I would reommend using the files in a native.psd format (if you have photoshop). VV handles PS files better than many applications, and using layers saves having to manually generate an alpha channel. Further, VV will also maintain transparency levels, making a lot of effects even easier.
I think one of the greatest benifits of using VV for stills is that it doesn't create it's own version of the image file on import... it always pulls from the master file information. What this means is that after you edit your still and save it, leave it open in PS. Place it on the VV timeline, and then go back to PS and make a change (add / edit text). The changes are reflected in the file on your time line as well.. perfect for doing CG work. FCP and Premeire recognize psd as well, but they also import each layer seperately. I'm sure this may have advantages in some instances, but in others I think it would just take up too much space on the time line.

I have used thousands of images in png and jpeg format for apx. 60 videos created within the last 2 years that include stills. I almost always pan across the picture since most scans do not match the dimmensions of the video so I need to go higher res than the screen. I scan a standard size picture at 200 dpi (I only increase this if I have a small picture) and save it in jpeg with apx. 80% compression and get no pixelization--ever. You might be able to optimize the compression but I don't get picky with files that small (relatively (and with 5 harddrives (120 - 200 gig each)).

I would not use tif (really bog the system down), GIF (not enough color), or psd (photoshop--why take it into photoshop when most don't need to be touched up, and you can't give them to folks without it) for most apps, since I like to include a CD with the jpeg images to my client (whether gratis anr paid).

The discussion above is fine, if you want to understand the theory, but this should work fine for you as a rule of thumb (that's the descriptive difference between a marketing guy and an engineer (even though I have an undergraduate in engineering)).

It seems to me that as far a efficiency goes, the bottleneck is hard disk throughput, not image decompression speed. One thing that made me curious is that I was getting better performance out of my digital camera stills than I was out of scanned photos, yet if anything, the digital photos looked better in the final video.

Was that a 200 dpi scan for a 6 x 4 snapshot?

Laurence Kingston

I would second what Chanimal says. Just do some experiments with photos scanned at different resolutions, and then experiment with one photo saved with varying degrees of compression. I suspect you'll find that the only time you see compression artifacts is when compression is pushed to the extreme.

Same for the photos scanned at varying resolutions--it will be obvious pretty quickly just how far one can zoom in to a still with a given resolution.

This will no doubt change when we all begin doing work in High Definition, but that's more than a few years away.

i take it the Vegas cant handle Jpeg2000 format just yet?