[Sam Moulton]

Welcome to video…. Interlaced video only has the illusion of being sharp. Too much detail and you get moray patterns or flicker. The minimum horizontal width for any detail is actually 2 pixels. Even progressive rendered video is interlaced on a TV or CRT monitor, it’s just that both fields are identical slices of time so you are still limited to the amount of detail you can display. The trick of good design is to make it look sharp, even though it really isn’t.

[Sam Moulton]

do you mean yuv?

[Sam Moulton]

first you need to figure out how many pixels you need for your print job. Say you want the print output to be 8.5 inches wide and your printer recommends 240 PPI. Multiply 240 by 8.5 and you get 2040 pixels.

Now make your comp settings 2040 pixels wide and square pixels. If you have video or other raster graphics in the existing comp you may be out of luck because the originals just don’t have enough pixels. You’ll have to do the best that you can.

Once you get things working in AE, export the frame as a Photoshop file and then open it up in PS and change the PPI to 240 but don’t resample the image so that the pixel dimensions stay the same.

[Sam Moulton]

Turn off auto orientation is the easiest

[Sam Moulton]

not sure what you are trying to do here. You could add an expression to POI and just pickwhip to the position. This gives you some really odd movements of the camera.

If you just want to manually point the camera somewhere then turn auto orientation off.

[Sam Moulton]

Three things jump right out at me about your project..

First item. The illustrator artwork was 20 inches by 20 inches. Actually that was the document size. The artwork extended way past the edges. Why so big? You end up with a comp that’s nearly 4000 pixels wide. Lots of unnecessary calculations going on here.

Second item. I noticed that the antialiasing was set to faster instead of more accurate in the interpretation of the footage. I know that there are no gradients in your artwork, but I find that this just works better. If I knew how I”d set the interpretation rules so this was the default.

Third item. It has been my experience that CR works much better when you scale up than when you scale down. I guess this goes along with the first item.

[Sam Moulton]

You’re making this too hard.

640 X 480 – square pixels right?
720 X 480 – rectangular pixels 4:3 right?

drop the 720 X 480 finished comp in a 640 X 480 comp. You’ll loose 4 pixels on the left and right side and that’s all. The 8 missing pixels won’t show up on tv when you view the DV sized comp anyway so there should be nothing critical in there because of safe action/safe title considerations. If there is DV footage in the 720 X 480 comp you have black bars on the left and right side anyway because every DV camera, make that every video camera I’ve ever seen doesn’t put picture all the way to the edge anyway.

[Sam Moulton]

don’t know, keylight has been used for several recent blockbuster feature films….

I’ve never had much trouble with keylight. Most of the time success is just a couple of clicks and maybe a second application of the effect or a garbage matte to clean up a bad area.

[Sam Moulton]

648 X 720 was a typ0 should have been 648 X 480….

I know all about the adobe doc. There’s no place, no where, no time, that you should monkey with different X and Y scale values in standard square or non square comps. AE is PAR aware and if the footage is distorted on final render then the interpretation is wrong.

Check out this tutorial

https://forums.creativecow.net/cgi-bin/page_wrapper.cgi?forumid=2&page=https://www.creativecow.net/articles/gerard_rick/pixel_madness/index.html

or better yet, drop a HDV sized comp 1440 X 1080 @1:33 in a HD comp 1920 X 1080 @ square pixel, or a D1 square pixel solid (720 X 540 @ square pixel) scaled exactly to 90% X and Y into a D1 (720 X 486 @ .9) comp, or a NTSC Sq pixel Widescreen (864 X 486 @ square) comp in a D1 widescreen (720 X 486 @ 1.2) comp and you’ll see that not only do they fit exactly and that the renders are not distorted. You should never scale X and Y differently. If you follow the referred to Adobe doc exactly you’ll see that the X and Y scale values are exactly equal unless you use 720 X 534 square pixel example and the Fit command because there is no exact 720 wide square pixel equivalent for a 720 X 480 DV comp because DV isn’t 4:3 because 6 scan lines, 4 from the top and 2 from the bottom, are not encoded in DV and 720 X 534 is the closest whole pixel equivalent that’s .9 PAR

I don’t expect that this post will change your mind, but creating any square square pixel solid and then turning it into a perfect circle using a double click in the oval mask tool, then putting that solid in any par comp and rotating it will show you no wobble if you don’t mess with the X and Y scale values. If you try scale the circle to match the PAR of the comp your circle will wobble when rotated, which proves my point.

Members

[Sam Moulton]

oh yes you do. You never want them different unless you’re distorting intentionally. AE handles the PAR correction perfectly every time as long as the footage is interpreted correctly and the comp is set up correctly. If you want undistorted video the x and y scale values should NEVER the be different. drop a 648 X 720 square pixel image or solid in a 720 X 480 D1 comp at .9 par and it will fit perfectly every time and X and Y will be 100%