Canon 70D HD video

[Paul Robbins]

[John Rofrano] “Having said all this I don’t understand why you need it. Your source is your acquisition format and your project is your delivery format and no one delivers 80Mbps video so I see no need for this high a bit rate.”

That is what I really wanted to know. Looking at the video side by side (original and the one rendered at 25 Mbps) they look the same quality. I am new at video editing but have experience with using Photoshop and when you reduce a file size you loose quality, that is what I was comparing it to.

[John Rofrano]

Ah, OK. Now I understand your frame of reference. Video editing is not quite like photo editing. You always deliver a smaller format than you shoot.

To use your photo example, let say you shot RAW photos. The files would be huge and full of detail, but you can’t give grandma a RAW file from your camera and expect her to be able to view it on her computer. So you save it as a JPEG with high enough quality so that it captures the essence of the picture. Video is the same way. Your acquisition format may use a codec that no on else has on their computer unless they own your camera. So you save it in a common MP4 format that everyone can view.

As long as it looks good to you, you should not care how small the file is because come codecs are more efficient than others and that smaller file may have all of the details of your larger file. (sort of like saving a TIF image vs a JPEG image, the TIF will usually be larger and the JPEG smaller because the JPEG uses more efficient compression). A 16Mbps AVCHD file will have the same quality as a 25Mbps MPEG-2 file because AVCHD is a more efficient codec than MPEG-2 and produces a mush smaller file at the same quality.

So don’t get hung up on file sizes. Use your eyes to judge quality.

~jr

http://www.johnrofrano.com
http://www.vasst.com

[Dave Haynie]

[Paul Robbins] “Ok but back to my original question. The original file is 600mb for one minute of video. The highest quality I can do in rendering is 110mb. That is roughly a 1/6 drop in quality. Is there any way either in Vegas or some other program that I can get the same size and quality out of the file that I have put together as the original.”

Ok… you have a 600MB file that’s 60 seconds in length, which translates to 10MB/s or 80Mb/s… video rates are usually measured in megabits per second. So you’re shooting in “All-I” mode, as Canon calls it… everyone else on the planet shooting video would call this AVC-Intra. It’s basically like HDV or Motion JPEG, only using the AVC I-Frame. In short, you’re encoding every single frame separately, none of the “MPEG Magic” stuff that really shrinks video down.

This will make editing much faster (unless you have a slow HDD) and increase quality, particularly for fast motion, obviously at the expense of storage on the SD card and the need for a Class 10 or UHS-I SD card, rather than the Class 6 needed for older Canon video. That’s ok, you want a really fast card for burst shooting photos, anyway.

So when you encode to AVC for Blu-ray or the Internet or anything, you’re absolutely not losing 1/6th of your quality (or even 5/6th, if you did the math correctly). What you’re doing is changing the nature of your encoding.

In full fledged MPEG, all sorts, you start with an I-Frame, which as mentioned, is the entire content of you 1-minute of video. I means “Independent”.. it’s like a JPEG photo, more or less (it was originally EXACTLY like a JPEG photo, back in the days of Motion JPEG, but DV, HDV/MPEG-2, and AVC/MPEG-4 have each added their own refinements to the still photo part). You’ll typically see only one I-Frame for every 15 video frames.

So then, what are the other frames. The next kind is the “P” frame, or predictive frame. Crazy sophisticated algorithms take your I-Frame and compare that next frame of video to it, and compute “motion vectors”, which essentially turn that first frame into the second one. Motion vectors are tiny. And they’re not perfect, so the encoder computers the difference, then encodes that… a full frame with practically no information in it. The P-Frame encoding plus motion vectors are stored in a tiny bit of space compared to that second frame. And then the third frame is compared to the second frame, and so on .. until it’s time to store another I-Frame. The result is, rather than getting 5:1 or 8:1 or so compression from the Intra Frame only video, you get much, much greater compression: as much as 50:1… all depends on the bitrates set.

Your 70D can record in “IPB” AVC as well, not just I-Only. The “B” stands for “bidirectional”; it works the same as a P-Frame, except that it can be encoded based on predictions from either the former or next frame.. or both. So they take even less space to store.

So basically, going to this more sophisticated encoding scheme, you aren’t necessarily losing any quality. When you are, particularly at higher bitrates like those for Blu-ray (16Mb/s or 25Mb/s are typical Vegas presets), you’re getting into some digital distortions — aka encoding artifacts, particularly in areas of high motion. But on many subjets, you won’t see any difference in quality. And that’s the key — your eyes work. If it looks bad, it’s bad. If it looks good, it’s good. But all commercially distributed AVC and MPEG video: DVD, Blu-ray, YouTube, Satellite, Cable, etc. it’s all IPB encoded. What the AVC-Intra mode gets you is a clean copy without certain classes of MPEG artifacts (eg, those due to predictive frames) even possible. That makes for a cleaner edit, etc.. though usually, with the higher bitrates used in DSLRs and pro camcorders, that’s not much of an issue anyway.

And there’s no I-Frame only distribution format. You could cook up a video pre-set (at least in Vegas Pro) for I-Frame only rendering, and maybe even successfully upload it to YouTube or Vimeo. But they’re re-code it to IPB anyway… for one, no one’s happy streaming 80Mb/s video 😉

-Dave