 |
|||||||
 |
|||||||
 |
T436 - Fall 2012 - Week 13 Agenda
Announcements/Reality check Here's what still remains to be done THIS WEEK:
Extra Credit: I could use two people who are very strong editors to assemble the final show(s). This would mean you'll get an incomplete until the show is finished and likely amount to an increase by one letter grade. HDTV Production Codecs There a literally hundreds of variants of codecs, frame rates and pixel dimensions available to you as a video producer. Some codecs are optimized for acqusition and delivery. Others are optimized for editing. A wise video producer will understand the strengths and limitations of popular codecs and pick the right tool for the job. He or she will also understand that bitrate is a limiting factor. Recording/Acquistion - Videographers usually want an inexpensive, compact camera that can record for long periods of time in the highest quality possible. This results in an abundance of H.264, HDV, and other interframe recordings typically shot with DSLRs, and cameras such as Sony's NX5 or Panasonic's XF300. Recording quality and bitrate are limiting factors.This usually results in a lot of compression and loss of color space. All things being equal, camcorders that employ larger bitrates can achieve higher quality recordings. In order to achieve higher end recordings, one needs to record at a higher bitrate. This usually comes in the form of either a more expensive camcorder or an outboard recording device such as AJA's Kona KI. Intraframe vs Interframe HDV a popular codec, encodes video as MPEG-2 with 4:2:0 color sampling. MPEG-2 uses interframe compression, which compresses both spatially and temporally. In the particular iteration of MPEG-2 used to record HDV, the video data is broken down into Groups of Pictures (GOPs). Intraframe codecs such as DV treat each frame individually and thus only compress spatially. Because MPEG-2 can compress over time as well as space it is capable of delivering a high-quality image in a smaller amount of bandwidth than an intraframe codec can deliver. A great deal of MPEG-2’s efficiency is due to the fact it compresses the video into groups of pictures (GOPs) and not simply individual frames. In MPEG-2 compression, images are divided into macroblocks, which are typically areas of 16 x 16 pixels. GOPs are created with three types of pictures: I, P, and B frames. I frames are intracoded frames, which are sometimes referred to as index frames. P are predicted frames and B are bidirectional frames. A GOP starts with an I frame. In MPEG-2 compression, P frames are compared to the previous I or P frame. If there is a difference, a proper vector is determined to move the macroblock. If there is no change (if there is no movement within the shot), the bit rate can be reduced significantly. B frames, or bidirectional frames work similarly, but reference previous and future frames. Intraframe compression treats every frame individually, compressing one after the next. These types of compressors (such as Apple ProRes or Avid DNxHD) facilitate editing because each frame is independent of the others and can be accessed at any point in the stream. Since interframe codecs (such as HDV & H.264) break the video stream into chunks known as GOPs, ease of editing is reduced in favor of maximizing compression. So while MPEG-2 is perhaps ideal for transmission, its multi-frame GOP structure is not optimized for editing. It is possible to edit MPEG-2 without recompression as long as the edit points resides on a GOP boundary. HD Production Formats HDV – Canon, Sony and JVC offer lower cost HDV cameras that record at a maximum resolution of 1440 x 1080. HDV uses a form of MPEG-2 compression that can be recorded onto miniDV cassettes. In 1080i mode, HDV can record a 25 Mbps signal. In 720p mode it records at 19 Mbps. Because MPEG-2 employs Groups of Pictures (GOPs) instead of discreet frames, HDV data ideally should be up-converted into a different format for editing. Because the data rate is relatively low, HDV content can easily be transferred over a FireWire (IEEE-1394) connection. DVCPRO HD – Also known as D12, DVCPRO HD was developed by Panasonic and has versions that record on magnetic tape as well as memory cards. The 100Mbps data rate is still low enough to be transferred over a FireWire connection from a VTR into an editing system. DVCPRO HD is restricted to a maximum resolution of 1280 x 1080 pixels. Apple ProRes 422 - Introduced in 2007. It uses 422 color sampling and has two target bitrates: 145 Mbps (normal quality) and 220 Mbps (high quality). ProRes also uses VBR (variable bit rate) recording. Avid DNxHD - Avid's response to ProRes. Variable bit rates in 220 (10 or 8 bit), 145 or 36 (8 bit) XDCAM HD - Sony’s tapeless format records onto Blu-Ray optical discs using several possible frame rates and codecs. It can record HD content using MPEG-2 encoding at 35 Mbps or DVCAM at 25 Mbps. Its HD resolution is restricted to 1440 x 1080 pixels. D-5 HD – Developed by Panasonic in 1991, the D-5 format has been updated to HD. It records at a 235 Mbps data rate and can handle 720 and 1080 content at most possible frame rates. HDCAM - Sony’s format records onto 1/2” videocassettes at a number of possible frame rates. It uses a 140 Mbps data rate and supports up to 4 channels of audio. It too is restricted to a maximum resolution of 1440 x 1080 pixels. HDCAM SR – Sony’s higher end version of HDCAM shares some of the same features but can write data rates up to 880 Mbps with up to 12 audio channels.
Editing While linear, tape-to-tape based editing is still viable (and sometimes best suited for the job), most editors work with computer-based, non-linear editing systems. With dozens of vendors making HD-capable editing systems, there are many codecs available to choose from. Some codecs require proprietary hardware to use, while others are hardware independent. In addition to the standard bit depths of 8 and 10, there are also higher end, 16-bit codecs available from companies like Pinnacle and Digital Anarchy. While HDTV is routinely compressed using MPEG-2 for transmission and delivery, uncompressed or mildly compressed data is preferred for editing. Since it’s often necessary for editors to composite many layers of content together in order to create special effects, it’s important to keep the signal as pristine as possible. This is why editors will often upconvert footage to a codec with better bit depth and higher resolution. 10-bit files contain more information than 8-bit files but also require more storage. Projects with demanding chromakeying or color compositing needs will be better served by codes with higher color sampling ratios (4:2:2 over 4:1:1, etc.). Similarly, compressed footage requires less bandwidth at the tradeoff of some quality loss. Lastly, choosing a lossless codec that operates in a 4:4:4 resolution will offer the highest quality but at the expense of requiring the greatest amount of storage. Editing with personal computers in the production lab HDV or MPEG-2 can be edited in its compressed form, but it's extremely processor-intensive. At 25 MBps it uses the same amount of bandwidth as DV, so users can easily use standard internal and external firewire drives for storage. Because of it's heavily compressed data structure, it's difficult to process video effects, such as color correction, and the entire sequence must be rendered or conformed before outputting to an HDV recorder. This can only be done via firewire interface. There are no HDV recorders at the time of this writing (11/2006) that accept HD component or HD-SDI- all use a firewire interface as the sole means for input. DVCProHD aka DVCPro100 is an intraframe format, which in essence uses 4 streams of DV data. At 100 MBps, it's data can still fit on fast internal drives as well as external firewire drives. Some users have reportedly gotten up to 3-4 streams of DVCProHD from a single firewire800 external drive. Booth 2 & 3 in the production lab are outfitted with AJA's Kona LHe cards that allow for HDV and DVCProHD acceleration and monitoring. While they have HD-SDI outputs, these are only utilized for monitoring.. We have no decks capable of real-time HD recording. Reviewing Shorts As mentioned earlier, we'll be reviewing the finished shorts the week after Thanksgiving. Everyone will be asked to write down and discuss elements that were working or not working. This will go to your participation score. Remember that you will be asked to include feedback from at least 8 different people. ----------------
Production Meeting
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|