High quality video with small file sizes.

January 20, 2009 1:21 PM

Compressing quicktime movies for the internets.

I'm in a new position where I have to prep a lot of video content for the web. I'm looking to maximize quality and resolution while keeping file size as practical as possible (my target audience does not include high-end users with blazing connections). Looking at the trailers that Apple throws up makes me feel my work has something left to be desired... are there any tutorials available on how to prep content for the web using Quicktime, Final Cut, or Compressor? I feel like I have a rudimentary knowledge of how this works but I am left wondering if there are any cool tricks or advanced tips.

posted by phaedon to computers & internet (11 comments total)

I don't mess with Quicktime, generally speaking, however I'll save you hours of trouble with one tip: if you wish to stream the video, the Quicktime file generally must be hinted (their term) before it hits the server.

The last time I staggered around this issue for quite some time before discovering that this particular step was something Apple felt was best left for another document.
posted by adipocere at 1:54 PM on January 20

Are you using H.264? I'm pretty sure that's how all of Apple's stuff is encoded.
posted by designbot at 2:14 PM on January 20

What designbot said. It changes all the time as new codecs are deveoped and patents are awarded, but h.265 is the latest in a long line.
posted by rhizome at 2:29 PM on January 20

er, h.264
posted by rhizome at 2:30 PM on January 20

In Compressor, use h.264, but use the one specifically for video podcasting. It makes an .m4v file that default plays in iTunes, but works in most players. I used regular h.264 codecs and they were all interlacey, but the video podcasting one works wonders. Upload to a site like blip.tv and you're set.
posted by yellowbinder at 6:18 PM on January 20

In case you're having trouble finding the one I'm talking about, here's how to find it in Compressor: Apple/Other workflows/Podcasting/h.264 for Video Podcasting.
posted by yellowbinder at 6:23 PM on January 20

> I'm looking to maximize quality and resolution while keeping file size as practical as possible

You don't mention rendering time. My colleague tells me that H.264 is the best format, but the slowest to encode.
posted by AmbroseChapel at 8:26 PM on January 20

All of this advice is good, and should be taken, but bear in mind that compressing is an art form, not a science. Most likely it will just take tons and tons of experimentation for you to get what you're looking for. I've done my fair share of compressing and posting to the internet and I'm still in awe of whoever gets paid to put movie trailers on apple.com.

Just don't be discouraged if you spend a couple of dozen hours on that first compression to get a nice balance between quality and size. Once you establish a method the others will go much faster.

It's not like there's a master document out there that everyone knows about and you don't. Google "quicktime compression tips", do some reading, and then put aside some time to experiment. Good luck!
posted by Bobby Bittman at 9:06 PM on January 20

A couple of tips:

• Make the resolution of your output piece (width and height) evenly divisible by 4. Most codecs work in on a 4 x 4 pixel space, or multiples thereof. Making the video 401 pixels wide, for example, forces the codec (and playback) to work extra hard.
• For QuickTime, and full motion video, H264 is generally the way to go. Sorenson is also good, albeit (and arguably) no longer best-of-breed. Treat compression of web video as you would (or should) JPEG for the web - start from the greatest amount of compression (lowest "quality") and work up. Don't become attached to the words on the slider. You will find, as noted, that the greatest amount of compression results in the longest processing times and the smallest files. If in doubt, compress a 20-second segment of footage (something with a great deal of motion and / or high color contrast would be ideal) and produce a test mov, rather than trying to compress the whole thing in one go. Check the quality on the test, then apply the same settings to the whole video.
• You can also set H264 to try to optimise the movie to a certain bandwidth. I haven't found the need to do this, but if you have orders from on high to, say, "make this playable on the boss's computer at home", determine what Internet provider your boss uses, the service s/he receives, the usual bandwidth rate after 5pm, and go from there. (Also helpful is any research of your expected audience that the marketing department might have).
• Work from the best quality footage you have - the bigger (higher resolution) the better. Don't forget to compress audio as well - it's doubtful that the audience has the need, or even the ability, to play back 48-bit 48kHz stereo audio, to use an extreme example.
• Keep in mind there will never be one perfect "recipe" for optimising compression. Even if you're handed video with the same theme week after week (five minute business presentations, wrestling videos, whichever), each will require careful tweaking to get the very best optimization results / lowest file size. Depending on the amount of work you have, you will likely settle on a "one size fits all" solution that you will use over and over, that achieves good compression for most videos.
• Again, depending on your workload, automating or scripting your production workflow may be a really good idea.
• Using keyframes may be a good idea, if the video is fairly static or predictable. (Talking head footage is a perfect example). Again, there is a balance - too few keyframes causes the codec to work more as the image "drifts" from the last keyframe; too many and the codec is essentially re-initialising itself every other frame.
• Don't use the Animation codec just because you have a 3D rendering in the video! (Sorry, but that one drives me crazy).
• Like most things, doing this well is an art, and there is a degree of learned instinct that takes place after a while. Experiment. Compare. Try again.

Hope this helps!
posted by Bora Horza Gobuchul at 9:30 PM on January 20 [3 favorites has favorites]

Another thought that occurred to me while I was cycling home, phaedon:

It's very likely that, especially for the big movies, that the guy or girl working on compressing the film trailer has been handed a full 4K resolution file, probably losslessly compressed. This doesn't mean that you can't achieve great results from SD footage that has already been stepped on, but stellar results are harder to achieve. GIGO - Garbage In, Garbage Out - is still very much the rule.
posted by Bora Horza Gobuchul at 9:57 PM on January 20

I'm not sure if you'll be exporting out of editing software. If you are, make sure you deinterlace any interlaced footage on the timeline. Do not deinterlace on export ever. This is a common mistake.

Also, if you're editing pay careful attention to the aspect ratio of your source footage. Make sure your sequence settings match your source footage. Export using square pixels, not rectangular (D1 .9) pixels for the web.

Here's another mistake many people make when working with DV footage:
Don't export to DV if your video is for the web. DV compression is muddy and ugly. H.264 is the only codec you should be working with to ensure quality and (mostly) cross-platform friendliness.

Get QuickTime Pro if you don't already have it. Uninstall Perian if you're exporting h.264 video. There's a bug in Perian that causes QuickTime to wash out your colors.

I spent the last two years troubleshooting video exports for a large podcasting company. This is what I've learned in two years. Hopefully this helps!

Some simple starter settings:
mp4 or m4v with h.264 encoding
Automatic keyframes
750 Kbps video
128 Kbps audio
Do not deinterlace on export
posted by plasticbugs at 10:11 PM on January 20 [2 favorites has favorites]