ALL-L vs. IPB encoding, real world advantage?

Rich Parry wrote on 12/11/2012, 4:15 PM

The Canon 5D Mark III allows the user to select video recording with ALL-I or IPB frames. I understand (more or less), I, P, and B frames and GOP. ALL-I frame recording means you have more data which also means the files are larger (3x approx.).

The Canon manual states encoding with ALL-I frames is “more suited to editing”. That comment is a little nebulous. I wonder if there is a significant advantage to ALL-I recording and what it means in the real world.

In case it matters, I am NOT a professional. I shoot exclusively outdoor landscapes, National Parks, rivers, mountains, clouds, etc. for my own person use. I don’t shoot indoors and never shoot people.

Is there an advantage to me shooting with ALL-I frames?

Thanks,
Rich in San Diego

Comments

john_dennis wrote on 12/11/2012, 5:32 PM

Since a full frame of video is saved in an i frame, the video editor has less work to decode the video on the timeline. Preview should be easier than with a long group of pictures. Since there are more bits, the video could be more accurate.

Serena wrote on 12/11/2012, 6:17 PM

Presumably your scenes contain a lot of detail (leaves, waving grasses, flowing water, etc) and it is in such detail that video compression is most obvious. Therefore I would shoot all I-frames to minimise compression artefacts, and in this controlled environment longer files are of no consequence.

riredale wrote on 12/11/2012, 8:12 PM

But that's the thing. Are there artifacts?

HDV is basically MPEG2 with all the attendant I-P-B complexity. But I thing reviewers were astonished at the lack of artifacts when the first Sony HDV cameras were introduced.

I suppose, however, that if you zoom in enough you will always find artifacts, so for the most robust and flexible source imaging it would make sense to shoot only I frames.

musicvid10 wrote on 12/11/2012, 8:47 PM

At 1080 resolution, artifacting from interframe compression is less of an issue than it was say, in DVD production.

Rather than the "wormy" kind of artifacts common to MPEG-2 and JPEG compression, I would shoot some high-detail, high-motion footage and compare the two. I would also try some deliberately underexposed shots and look for blocking artifacts. These are more of a concern than worms with AVCHD-like formats.

I suspect that you will find IPB interframe footage quite palatable, but be sure to compare timeline handling with your intraframe (all I) tests as well. It may be worth the huge tradeoff in size, maybe not.

Rich Parry wrote on 12/13/2012, 7:55 PM

Thanks to all that replied. Because I am fairly new to video, I don't trust my eyes to know what artifacts look like. I suspect that if I knew I might say "how did I ever miss those artifacts".

I wish I had a guru to say, "you idiot, they are are". From that point on I would probably wonder how I ever missed them.

Thanks again,
Rich

riredale wrote on 12/13/2012, 8:20 PM

The easiest way for me to demonstrate MPEG2 artifacts is to look at video of an ocean with a busy surface. MPEG2 artifacts show up as a gritty or granular appearance of the water.

Another manifestation is to see video fading up from black. A low encoding rate gives pretty obvious banding of the near-black portion.

Jpeg artifacts are easy. Look at a photo, even one which looks great, and just keep zooming in. At some point you'll see blockiness, which are the 8x8 pixel regions showing up.

As for AVCHD, I have no idea. Motion smear?

john_dennis wrote on 12/13/2012, 9:46 PM

"Because I am fairly new to video, I don't trust my eyes to know what artifacts look like."

Watch the Olympics swimming competitions over-the-air in 1080-60i at ~14-15 mbps. You'll get the hang of artifacts soon enough.

Or you could watch the 2 mbps multicast channel where they show old TV shows.

musicvid10 wrote on 12/13/2012, 9:52 PM

Or try and fit the whole Superbowl with ads and halftime on a standard DVD.
;?)