Here's a (hopefully) lucid explanation of why one 720x480 image looks fine, but another one doesn't. Images created on the computer are created with an aspect ratio of 1.0 (otherwise known as "square pixels"). They take up the same amount of space both horizontally and vertically. A typical computer image is 640x480. However, images on NTSC video are not "square", meaning that a pixel in video is actually taller than it is wide. The aspect ratio is about 0.89; PAL is similarly non-square, but different from NTSC. Because the pixels are narrower, you can fit 720 of them in the same physical screen space as a computer's 640 pixels. So what happens when we use an image created for the computer on a video screen? Assuming you created a 720x480 image which looks fine on the computer, that image will take up less horizontal space on a video screen, while the vertical space stays the same. Thus, you get tall and thin images. In order to get a computer image to look "right", you need to scale one dimension or the other to get the proper aspect ratio. The easiest way to do this is to take an image which is a multiple of 640x480 (I often use 1280x960), and resize it (in photoshop, turn off the "constrain proportions" box) to 720x480. The image will look a little too wide on the computer screen, and everything will appear short and fat. But put that onto video, and it looks right again. Now, if you start with a still frame from your DVcam, like Don does, it's already in video's aspect ratio. So you don't need to do anything in the computer to rescale it if you're planning on putting it back on video. However, if you want to keep the image on the computer (to put on the web, or to print out), you should rescale the image to 640x480. This will fix the aspect ratio for the computer. I'd recommend staying away from scaling twice (coming from video, scaling to 640x480, then back to 720x480). That will look pretty awful because you're throwing away information, then trying to recreate it again. So, to summarize, just remember where the image originated, and where the final destination will be. If it goes from computer to video, scale a multiple of 640x480 to 720x480. If it goes from video to the computer, scale 720x480 to 640x480. If it goes from video (even via the computer) back to video, don't scale. -Dave P.S. If you are compositing multiple images together, make sure the aspect ratio of each image is the same when you do the composite! __________________________________________________________________ More from John Asbacher on 9/10/00 This is also something I stole from someone before the big crash. DV is always rectangular pixels. The D1 pixel ratio and for NTSC that's 720 X 486. DV throws away the top 3 lines and the bottom 3 by increasing the horizontal blanking. If you want absolutely no distortion you need to render to 720 X 486 and then crop to 720X480 using D-1 pixel ratio. 720 X 480 is probably close enough, but if you want to exactly match some stuff for compositing then you need to use the 720 X 486 D1 pixel aspect ratio. Some 3D apps let you pick DV or D1 or set your own pixel ratio. If you can't do that, can't render to rectangular pixels, you need to render to 640 X 480 square pixels and then scale the image in your editing application. The same goes for scanning stills. Some say they prefer to scan at 720 X 540 and let the image app correct from there but it is my opinion that you get a better match with less interpretation if you scan and 640 X 480 and scale that. The 480 pixel height is exactly the same as the 480 scan lines in interlaced video and I think the math just works better this way. Try this for a test. Exactly center a circle on a 640 X 480 image and on a 720 X 480 and a 720 X 486 image. Import each into premere and then use the motion controls to spin the image. You'll see the wobble in all but the 640 X 480. It's all quite simple really. TV - NTSC and PAL - is exactly in a 4 X 3 aspect ratio. All pixel ratios that are not exactly 4 X 3 are rectangular pixels. If you don't want to distort your image then you need to change your square pixel computer artwork to rectangular pixels before it is turned into video. The only catch is that DV is the only format that throws away some of the scan lines. In all full frame video 1 pixel row horizantally is one scanline. The same thing holds exactly for 16 X 9. It's all 16 X 9 and if your pixel ratio isn't 16 X 9 then the pixels are not square. Simple as that. Just to clarify: the aspect of the pixels is not 4:3, that's the ratio of the screen. According to Adam Wilt, the NTSC pixel aspect ratio is about 10:11; slightly taller than it is wide. He recomends using 720 x 540 screens when rectangular pixels are not available as a save option. I like the idea of scaling down rather than scaling up a 640 x 480 screen. Its better to lose information than to create it out of nothing. Though I can see your point about the 'direction' of scaling. And remember, you can't import a non-DV avi into RexEdit. You'd probably want to do it as a series of stills Its not that hard a process to create an avi from stills in RexEdit (Raptor Edit?). Set still length to 1 frame, select all (cntl-a) in the file requestor (do NOT select one frame first--it will screw up the order), auto-arrange clips (cntl-g), and save timeilne as AVI. I'd be leary of doing it in Premiere or MSP, I really don't think either program has a good grasp on non-rectangular pixels... For a wealth of information about the DV format visit the link below. (I couldn't find the info on the aspect ratio problem--he answered my question directly.) You can always check for distortion by rendering a circle or sphere in the exact center of the frame and then using the Motion controls in Premiere to spin the circle. If you've got the pixel aspect ratio correct the circle will spin without wobbling. If you're off you'll see it instantly. You have to match pixel aspect ratios before or after importing the file. There is no way around this. There was a response about using the stills captured from a DV Camcorder. They are kind of weird. A publishing app or photoshop will distort these images. You have to turn them back to Square pixels to remove distortion. Just change the image from 720X480 to 640 X 480 and you'll remove the distortion. Check this out with a still of a circle also. You'll see that to print or use these captured stills without distortion you need to correct them. You can fix this before you start by selecting 640 X 480 in the capture stills section of your canopus capture software. ______________________________________________________________________ And John continues....... After Effects Square Pixel Issues Working With Non-Square Pixels (D1, DV) in After Effects D1 or "digital video" (also referred to as CCIR-601or ITU-R 601) is an international television standard for component video. It has a 4:3 frame aspect ratio, a .9:1 pixel aspect ratio and either a 720x486 screen resolution (D1 NTSC) or 720x576 (D1 PAL). Another video format with a non-square pixel aspect ratio, DV has a resolution of 720x480 pixels and uses D1 pixel aspect ratio. Because D1 is an international standard, you can use it with video formats for both NTSC (National Television Standard Committee, the broadcast standard in the U. S. and Japan) and PAL (Phase Alternating Line, the broadcast standard in Europe). The D1 video standard is commonly used with professional desktop digital video systems, whereas NTSC format is commonly used with consumer-grade desktop digital video systems. D1 comprises rectangular (i.e., non-square) pixels, instead of the square pixels for most Mac OS- and Windows-based systems. Some video components that use non-square pixels include: Media 100 and AVID editing systems; Digital Disk Recorders (DDR), which store uncompressed D1 frames; and most DV capture cards. DV video format uses D1's rectangular-shaped pixel aspect ratio but has a slightly smaller frame resolution; 720x480. In After Effects you will need to choose one of the presets that specify either DV or D1 from the Frame Size pop-up menu in the Composition Settings dialog box. When working with non-square pixel source footage -- particularly when mixing non-square pixel source footage with square pixel source footage -- you must take care to interpret the pixel aspect ratio of footage (square vs. non-square), separate the fields of D1 and DV video source files, and set-up the compositions correctly to prevent distortion and artifacting. NOTE: The Adobe After Effects 4.0 User Guide, p.107, and Online Help incorrectly state that After Effects automatically separates fields for D1 and DV video source files. Fields must be manually separated for all interlaced source files by choosing File > Interpret Footage > Main and specifying the field order. For more information on separating fields before editing, see the After Effects 4.0 User Guide, p. 107, or the After Effects 3.0 User Guide, p. 97. Choosing the Correct Pixel Aspect Ratio In Adobe After Effects, when you select a non-square pixel aspect ratio Composition setting (e.g., NTSC D1, 720x486), After Effects automatically sets the pixel aspect ratio of the composition to D1/DV. As a rule, After Effects interprets all imported footage files with a frame size of 720x486 or 720 x 480 as D1/DV, whether the file has square or non-square pixels. Therefore, you must change the pixel aspect ratio interpretation for all 720x486 square pixel footage, (e.g., images generated in Adobe Photoshop or other desktop applications). To change the pixel aspect ratio interpretation of a footage file, select the file in the Project window and then choose File > Interpret Footage > Main. To quickly check what the pixel aspect ratio interpretation is for the file, select the file in the Project window. In the thumbnail view at the top of the Project window, After Effects includes "D1/DV NTSC" or "D1/DV PAL" after the frame size for footage that has been interpreted as non-square. For example (pixel aspect ratio information is in bold text): 720x486, D1/DV NTSC D 0:00:05:02, 29.97 fps Millions of Colors 44.100KHz / 16 bit / Stereo As long as the pixel aspect ratio for each source footage file is set correctly, you can combine footage that has different pixel aspect ratios and generate output that displays correctly. Viewing Your Work Ideally, you should view your work on a D1 monitor. Because horizontal pixel spacing on a computer monitor is wider than that of D1 format, images in D1 format appear wide or horizontally stretched. If you don't have a D1 monitor, you can still view D1 compositions so they are correctly proportioned by using one of the following methods: - In After Effects, view the composition using a video monitor driven by a video card that supports D1 resolution (e.g., Intelligent Resources' Video Explorer or TrueVision's Targa 2000) and zoom to full screen by pressing Control+Shift+\ (Windows) or Command+Shift+\ (Mac OS). - In After Effects, create a viewing composition that has a 648x486 square pixel aspect ratio, and then drag your 720x486 D1 composition to the center of this viewing composition. After Effects automatically resizes your D1 composition to fit the viewing composition and correctly interprets the pixels to provide a preview of your final D1 output. This does not, however, show all the pixels in the D1 composition. If you then edit the 720x486 composition, any changes also affect the square-pixel composition, making rendering slow and more memory intensive. When you render, be sure that you render the 720x486 composition, not the viewing composition. - Using your monitor's size controls, adjust it so that a 111x100 pixel solid block looks square. This size reduces the width proportionately by 10% to compensate for D1 spacing, however, your desktop and square pixel footage will look distorted. Creating Square Pixel Footage for Non-square Output Because most applications create images that consist of square pixels, most of the animations and still images in your project will have a square pixel aspect ratio -- not a non-square, D1 pixel aspect ratio. However, some applications (e.g., Illuminaire) can create D1 output. Solids created in After Effects (Layer > New Solid) have the same pixel aspect ratio as their enclosing composition. Masks, as well as most effects, honor the pixel aspect ratio of the layer's source file. If you use square pixel source footage with a dimension of 720x486 pixels, you will need to either crop the sides or scale your image to make the footage fit perfectly in the frame of your NTSC D1 720x486 composition. To avoid cropping or stretching square pixel footage, create your square pixel source footage to mix with D1 at 720x540, then use the Shrink to Fit command (see the Shrink to Fit Command section below) in After Effects after adding it to the composition. If your square pixel footage does not need to fit the composition's frame exactly and the resolution doesn't simulate that of D1/DV, use the file as is; After Effects will interpret it correctly as square. When creating square pixel still image files to mix with your DV footage, use a frame size of 720x534 and use the Shrink to Fit command (see the Shrink to Fit Command section below) to scale appropriately. A second approach is to create still images at 640x480 resolution and shrink them to fit the composition. This second technique, however, will result in a slight loss of resolution in bitmap-based images from scaling them up (i.e., greater than 100%). Shrink to Fit Command To use the Shrink to Fit command, select the layer in the Composition window and press Control+Alt+F (Windows), press Command+Option+F (Mac OS). NOTE: Check that Preserve Frame Aspect Ratio is selected in the Scale dialog so the layer is not distorted during render. You access the Scale dialog by selecting the layer and choosing Layer > Transform > Scale. Rules to remember: - Before adding DDR footage files, files from D1 systems, DV footage, or video digitized in D1 format to a composition, always select the file, choose File > Interpret Footage > Main, and then choose either NTSC D1 or PAL D1 from the Pixel Aspect Ratio pop-up menu. - When creating graphics on a Mac OS or Windows system, use a frame size of 720x540 for D1 and 720x534 for DV. If graphics were created at 720x486 or 720x480, change the interpretation from D1/DV to square. - If you want to combine footage that has different aspect ratios, choose the correct pixel aspect ratio for each footage file from the Interpret Footage dialog box. Workflow Options for Non-square Output To render non-square output using square pixel source footage, use either of the following techniques. The first set of steps are easier to use and outline the way most people prefer to work. Motion graphic designers who work almost exclusively with square pixel footage that is output to non-square media may prefer the method outlined in "Work in a Square Pixel Composition," as it provides a more accurate display of layer placement. Work in a Non-Square Composition When you set up a composition and files, layers may look slightly skewed as you work on the project, so it may be difficult to gauge subtle positioning relationships between layers. Each layer's transformations and effects will be correctly calculated to account for its own pixel aspect ratio within the non-square composition. After you render the composition, all layers will appear correctly proportioned when viewed on a D1 monitor. For example: 1. Set the Composition Frame Size to one of the NTSC D1 720x486 settings in the Frame Size pop-up menu. 2. Create square source footage at 720x540. Add D1/DV footage as needed. 3. Shrink to fit the square source files by pressing Control+Alt+F (Windows), or Command+Option+F (Mac OS) during or after adding the footage to the composition. Work in a Square Pixel Composition When you set up your composition and files this way, all layers are correctly proportioned on-screen as you edit the project and in the final D1 output. This technique is primarily used for motion graphics work using square pixel source when exact positioning is necessary. For example: 1. Create a new composition and set the Composition Frame Size to D1/DV Square Pix. The pixel aspect ratio will be set to square pixels. This composition will contain all the source footage (layers, nested and precomposed comps). 2. Use the 720x540 frame size for all source footage you create. Add D1/DV footage as needed. NOTE: If you use non-square source footage, shrink the footage to fit the composition. To use the Shrink to Fit command, select the layer in the Composition window and press Control+Alt+F (Windows), or Command+Option+F (Macintosh). The Shrink to Fit command snaps the edges of the layer to the edges of the composition, and then scales the layer to 111%. 3. Prior to rendering, create a new composition, setting the Composition Frame Size to NTSC D1 720x486. Drag the 720x540 composition you created in Step 1 into this final composition. 4. Select the nested 720x540 composition either in the Composition window or the Time Layout window, and then press Control+Alt+F (Windows), or Command+Option+F (Mac) to shrink it to fit. 5. With the nested composition still selected, click Collapse Transformations/Continuously Rasterize in the Switches panel of the Time Layout window. NOTE: As After Effects renders the composition, frames will look stretched on-screen, but the output will be correctly proportioned on a D1 monitor. Working with PAL While the examples above detail NTSC work, the same rules apply to D1/DV PAL work, a non-square pixel format that uses a frame resolution of 720x576. The exception is that you should create square pixel graphics at 768x576 so that they fit the frame exactly. Use the same Shrink to Fit command when adding the square pixel source footage to your D1/DV PAL composition, and select Preserve Frame Aspect Ratio in the Scale dialog box for the layer. To adapt the "Work in a Square Composition" workflow to PAL, use the PAL D1/DV Square Pix, 768x576 composition to position layers, then nest the composition in a new composition set to PAL D1/DV, 720x576 for final output. Email: asbacher@telus.net