Apple iMovie 8.0.5 update debuts new iFrame video format
A support article describing the new format, which was just added to iMovie 09 in the 8.0.5 update, says the new iFrame Video format "is designed by Apple to speed up importing and editing by keeping the content in its native recorded format while editing. Based on industry standard technologies such as H.264 and AAC audio, iFrame produces small file sizes and simplifies the process of working with Video recorded with your camera."
Support for the new format was announced by Sanyo, which has added iFrame recording to two of its new camcorders introduced today, the HD2000A and FH1A.
The new iFrame format captures standard H.264 video at 960x540, a quarter the resolution of full 1080 HD. The new cameras from Sanyo default to record in the iFrame format, but can also be set to record in full 1080 HD.
Finding a format
Digital camcorders began recording in MJPEG (Motion JPEG, a series of still photo captures) before moving to the better compression of the popular DV format. While DV recording allowed for high quality capture, it wasn't optimally designed for direct editing in QuickTime; it uses non-square pixels and is oriented toward TV resolutions and aspect ratios.
JVC improved upon the consumer DV format with its HDV format (also supported by Canon Sony and Sharp) using MPEG-2 video similar to a DVD, although HDV uses a transport stream rather than a program stream (like DVD), as it is optimized for delivery rather than storage. The recording format is also optimized for playback rather than editing. Importing HDV into iMovie using an intermediate codec makes editing more efficient, but also requires more disk space.
A variety of other competing digital formats have appeared on the high end, including Panasonic's DVCPRO HD (based on DV encoding) to Sony's DVCAM (also based on DV) and XDCAM EX (using MPEG-2).
Panasonic and Sony paired up to create the AVCHD format, which is based on modern MPEG-4 H.264 video. However, AVCHD still multiplexes its audio and video into an MPEG transport stream rather than recording it as a standard MPEG-4 file. In order to edit the AVCHD video captured by camcorders, iMovie still has to import and transcode it into the Apple Intermediate Codec, which requires time and consumes lots of disk space. Final Cut Pro similarly imports AVCHD video into AppleProRes.
By floating the new iFrame format using the same standard MPEG-4 H.264 video, Apple hopes to simplify the import process for consumers, making it easier and faster to ingest camcorder video for editing.
The name of the new format appears to reference both Apple's consumer product line and MPEG's I-frames, or intraframes, which act as keyframes in the video recording. Between full I-frames, MPEG compression uses P-Frames or predictive frames, which only present what has changed since the last I-frame, as well as B-frames, or bidirectional predictive frames. These present part of a picture like a P-frame, but reference changes relative to a future frame. In other words, B-frames come in advance of an I or P-frame that fills in the missing details.
56 Comments
Wow. Can't wait to try it.
Hmmm
I expect an iFrame Pro in a couple of years that bumps up to 1080 resolution.
The difference between MPEG4/QuickTime and AVCHD isn't in media multiplexing. It's how the table of contents is stored.
AVCHD is in an MPEG2 streaming format comprised of blocks of media, with each block having its own short description. Cameras can record this format without needing to buffer anything more than the current block of media. Random-access playback requires a scan through the file to build a table of contents. It does not need to be imported before editing as Apple would like you to think. Importing is performed because QuickTime does not support certain H.264 extensions in AVCHD.
The MPEG4/QuickTime file format has one contiguous table of contents. That's great for random access playback but awful for everything else. To produce an MPEG4, a camera would have to write raw data and metadata to two files during recording. When the stop button is pushed, the camera would have to copy the two components to a single file, re-index the metadata into table of contents, then delete the two component files. It's a long process that would make the camera unresponsive and prone to data loss.
These are the kinds of stories I love about Apple on AI: No room for know-nothing blowhards that dominate general-interest threads (sometimes I fall into that category too) to peddle their vacuous BS.
I'd be curious to hear about whether/how iFrame lives up to its promise, from movie pros.
What? Are you kidding? I don't record anything below full 1080p HD.