-
Why shouldn’t odd numbers be used for video dimensions?
Hello.
This isn’t really specific to After Effects, but since that’s what I’m working in, I’m hoping someone can help me with this answer. A general rule I’ve always followed when creating video-based work is to always use even numbers for the dimensions of my work. It’s something I was told a long time ago and never really thought about the “why”. But, I now have a client requesting an odd-sized dimensions for a final video piece and I’m hoping someone could explain to me (so I can explain to them) why this shouldn’t be done, and that if we just modify the dimensions by 1px so the dimensions are even-numbers, we’ll all be happier.
The project’s final use will be for the video to be inserted into a webpage, so I’ll be supplying them with an h.264 file that they’re going to use for producing an FLV. And I know from past experience that odd-numbers and h.264 just don’t work, so I’d like to explain to them the “why” so we can make thing right before we get to deep into the job.
So, any information would be much appreciated!
Thanks,
k. -
6 Replies
-
Dave’s right.
There is another reason that this dimensional rule of thumb exists. When NLE’s and computer-based video developed, the codecs were optimized for even numbers.
Before it was feasible to deliver video on the web I produced interactive CD-ROMs that had video content.
Most of these video’s were reduced to 320×240 (anyone remember CinePak? Ugh, Sorrenson?) I have a vague memory that QuickTime would compress and decompress playback better when each dimension was divisible by 4.
-
Thanks guys!
Outside of a picture quality issue and codecs based on pre-web delivery, is there a technical reason why this shouldn’t be done? I mean, most current rendering/encoding/transcoding applications don’t even allow this (either rounding up or down the crop), so I assume it’s for a reason outside of these applications just “looking out for me” in terms of quality?
For example, I typically render out a lossless, high quality master from AE and then use Adobe Media Encoder for various delivery outputs (typically mp4 or flv). If I’m using AME CS4, it allows me to create an odd-numbered outputs. But, using AME CS5, it doesn’t (it rounds to an even-number). Again, I assume there’s a technical reason why it doesn’t allow this anymore (since it’s obviously “possible” since CS4 could do it)?
The main reason I ask is, while on this project, I’m able to deal with the client directly to suggest going with even-numbered dimensions, in the past, I’ve been dealing with sources (websites) that have odd-numbered dimensions that can’t be changed that I have to supply content for — in these cases, I’ve had to go back to AME CS4 for producing my odd-numbered mp4 or flv files.
Anyway, just trying to educated myself (so I can help educate my clients) and make everyone happy.
Thanks again,
k. -
Many codecs break an image down for compression into a smaller chunks called macroblocks, rather than processing the entire image as a whole. Macroblocks are fixed in size (often 4×4, 8×8, or 16×16 pixels), so the raster size of the compressed video cannot be arbitrary and must be evenly divisible by the size of its macroblock.
Walter Soyka
Principal & Designer at Keen Live
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
RenderBreak Blog – What I’m thinking when my workstation’s thinking
Creative Cow Forum Host: Live & Stage Events -
Thanks Walter!
Given that explanation then, is it enough to just use even numbers, or should the macroblock size of a particular codec be taken into account. For example, say I was asked to create a video at 100px x 100px and a particular codec’s macroblock size was 16 — that would result in a division of 6.25. Assuming the result should be whole numbers, would this be an issue?
Again, I imagine this situation would be a rare case, but again, just trying to educate myself as much as I can on this topic (as I’ve been dealing more and more with odd-size requests for video work as of late).
Thanks again,
Kristin. -
For codecs which use macroblocks, the frame size must be evenly divisible by the macroblock — you cannot have a fraction of a macroblock. AME is probably adjusting the frame size to fit over the macroblock structure for you.
Walter Soyka
Principal & Designer at Keen Live
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
RenderBreak Blog – What I’m thinking when my workstation’s thinking
Creative Cow Forum Host: Live & Stage Events -
OK, thanks again!
k.
Reply to this Discussion! Login or Sign Up