Resolution vs aspect ratio when preparing footage for editing

I'm having some issues preparing some video to use in Hitfilm and one thing I don't understand is why the aspect ratio is separate from the resolution. I have a resolution of 1680*480 and the ratio of 1680 to 480 is 7:2, yet I can output this at a ratio of 16:9 (and get different results than when I manually change the aspect ratio to agree with the resolution).

• Moderator

Pixels don't have to be square and can have their own PAR, Pixel Aspect Ratio, distinct from the DAR, Display Aspect Ratio, which is the aspect ratio of the image as displayed. There is also the SAR, Storage Aspect Ratio, which is the ratio of the pixel dimensions, width to height.

SD systems like NTSC and PAL used non-square pixels when they got around to actually using pixels. Originally the video signal was defined as an array of vertical lines (raster scan) and a defined center point but edges or number of elements (pixels) per line? Yeah not so much. Neither one was ever defined in any analog SD system. The concept of pixels applied to video signals only came about as a result of the possibility of digital video. Rec 601 defined the number of pixels for SD video systems and specified they would be non-square pixels and the term Pixel Aspect Ratio was coined as a result. Rec 601 also led to a lot of headaches for programmers in the early days. The Lurker's Guide to Video still has a lot of great information on the background and history if you're interested.

Differing PAR's and DAR's are still used some today as a compression technique. HDV records 1440 x 1080 which is a 4:3 PAR and SAR but the DAR using square pixels is 16:9

Anyway when you manually change the aspect ratio on your clips the software is just trying its best to make sure the math between PAR, DAR and SAR works out to a right answer.

• Moderator

To build a bit on what @Aladdin4d writes, once (NTSC) video was "defined" in pixel dimensions the numbers ended up at 640x480 with a 4:3 DAR and a 1:1 PAR, OR at 720x480 with a 4:3 DAR and a 0.9 PAR (sometimes this was 720x486! Just to add confusion. Since analog video didn't have "pixels" the definitions got fuzzy).

Because computers think in 0/1 bits grouped into "words" (a grouping of bits into a single value. So 8-bit audio uses an 8-bit word, 16-bit audio uses a 16-bit word--8 bits is a "byte," still used as the base storage measure...) the 640x480 ended up chosen because both numbers are multiples of 16... At the point these standards were designed a 16-bit processor was the norm.

So... 640x480 resolution became the digital video standard because it was the closest pixel matrix to what analog video waa that conveniently divided by 16 for processing by a computer using 16-bit words... Digital video itself (including JPEG and MPEG standards) compresses data in 8x8 or 16x16 blocks--so, when you see really low quality digital images with blocky artifacts, those artifacts are either 8x8 or 16x16 blocks (I could be wrong, but I think some modern codecs might process 64x64 blocks).

When "wide-screen" got defined the 16:9 DAR came in because these ratios are the squares of 4:3. Easy to reprogram. This became 640x360. Again, multiples of 16.

This 640x360 display became digital video for COMPUTERS. This gives a 16:9 DAR ratio with a 1:1 PAR. NTSC SD 16:9 video for DVD had a 1.21:1 PAR and the DVD player would do the scaling.

This has been slightly oversimplified. For this discussion we don't need to delve further into the arcana of a mostly dead format.

640x360 is the basis for all VIDEO (16:9 DAR)  resolutions. 1280x720? (640 * 2) by (360 * 2).  1920 is 640 * 3. 2560 is 640 * 4. 3840 is 640 * 6, etc.

As Aladdin also notes, most cameras these days (anything HD or higher-res) will shoot at a 1:1 PAR. A rare example would be HDV cameras which tried to fit HD resolution into a storage format (Mini DV tape) designed for SD. HDV cameras record 1440x1080 with a 1.33:1 PAR.

Over 99% of the time most of this is just trivia. Pretty much any camera most Hitfilm users would use will record a "standard-format" resolution (an even multiple of 640x380) with a 1:1 PAR. Exceptions occur, but would generally only be found incameras from before 2010 or so. More common as an issue is screen capture. There's an active thread this week from a user who captured a game at 1024x768 (4:3 DAR) to output at 1920x1080 (16:9 DAR). Been trying to get him to understand the only way to make his "square" image fill a "rectangle" is to chop off the top and bottom or horizontally stretch the image. One throws away data, the other distorts the image.

This image sums up the difference between 16:9 and 4:3 DAR.

And this shows the result of stretching a 4:3 image to fill a 16:9 frame. (The right half is stretched)