This is part of Sony's push to cater for not just HD, but higher resolution capture, including 4K and above. Of course, "resolution alone does not make better pictures. There are other imaging parameters that we need to consider, such as higher dynamic range, wider colour space, multiple view points (such as 3D), and higher frame rates," said Yasuhiko Mikami, its Senior Manager for Product Planning (pictured above with the F3).
"By improving the technology and our products in these five directions, we can start addressing new applications that exist outside of the traditional broadcast business, such as 4K cinema and ODS (other digital stuff), UDTV, 3D, and HDR." Ultimately, he hopes that Sony can come up with "a group of new products that pretty much does everything."
The new SR Memory is the first generation of this, and will allow real-time recording and playback, file-based post production, and offer scalable compression levels for different types of application (one of the first of which he believes will be for the Olympics in 2012). It uses the MPEG-4 Simple Studio Profile, which covers everything from standard definition to 4K x 2K 4:4:4 at 10- or 12-bit 50/60p.
To enable SR file based operation, it is providing the native SR compressed file and wrapping it in MXF. "SR is based on MPEG-4 SStP. A completely open codec," he explained. For high-end post work, this can be encapsulated in DPX files. However, this would be unnecessary for most broadcast work, so it is also adding a new level: SR Lite, which will record at 220Mbps. The other levels are 440Mbps SQ, for acquisition and post, and 880Mbps for 1080 50/60p and visual effects.
On File
SR Lite means "you can shrink the file size to half with minimum sacrifice in image quality. We firmly believe it is important to keep your camera original material at the highest possible quality. However when you're finished with grading, compositing, editing, you can shrink the file to make it pass through the broadcast food chain more easily. We also plan to create SR Lite files from regular HDCAM tapes, so that images acquired on your ten-year old F900 will still fit into the modern post production environment."
Sony has been working with the EBU to offer a production codec for HD production, and tests have shown that "SR Lite seems very promising, even after multiple generations."
Avid's Media Composer, Apple Final Cut Pro and FilmLight will all be supporting SR, and Mikami, who was speaking to a dealer meeting in London, showed plans for an SR Data Transfer Unit that will allow fast ingest of cards or tapes to servers or NLEs. It will support NFS and CIFS disk formats to mount a shared directory or folder on NLEs or servers, and can ingest using Gigabit Ethernet or, optionally, 10GbE. It will also clone SR Memory to tape at two times normal speed and will be controllable via a web browser.
On Deck
"The Memory Deck is very flexible in terms of I/O. It can combine four cameras onto a single card and use another card as backup, or one 2D or 3D camera onto four cards, if required," he explained. Theoretically it could have 12 inputs and record 12 channels onto a single card.
It can also do an "extremely fast" copy between a card and the built-in solid-state drives. It can also be used for slo-mo recording, and Sony is talking to third-party slo-mo controller companies about this, although it won't work at very high frame rates, but that is an interface problem rather than an internal problem. "You can't feed a 1,000 frame per second signal into the recording yet, but it will have a 4K video card in future which will support up to 240fps HD."
SR memory has a guaranteed transfer rate of 5Gbps, more than fast enough to enable multichannel HD and high framerate recording, and cards will be offered in various capacities, from 256GB to 1TB.
Sony chose 5Gbps as its design goal because if you record 4K 60p, full RGB, 10-bit material, the uncompressed data rate is about 20Gbps. "Then apply 4:1 good quality compression, like SR, and that gives 5Gbps."
An SxS card has a maximum transfer rate of 800Mbps, but this can fall to 200Mbps due to fragmenting and poor memory management. However, in SR Memory "a very clever memory chip controller mechanism is built in, so you will never skip a frame during recording or playback." It ensures that no single memory sector gets over used (making it more likely to fail), and has "incredible error correction."
Even if, in the worst case, some memory chips in the card fail, the system can recorder the data. This means that users don't have to stack up multiple cards with a RAID controller to make the system secure.
On Camera
When used with an optionally-upgraded PMW-F3 Super 35mm camcorder, it can record 10-bit S-Log RGB files, with look-up tables (which are baked in to the MPEG-2 4:2:0 copies recorded to the SxS card, for offline use). "It's a very self-contained, integrated solution," said Mikami. However, it does need a dedicated power supply. The unit could be used with any camera with dual-link SDI, but it will only handle variable speed recording on the F3, thanks to a dedicated connection – the F3 apparently won't do variable frame rates with third-party recorders.
It offers RGB 4:4:4 sampling without interpolation (as in graphics above), at 1920x1080, with no deBayering needed (and therefore no interpolation as in typical Bayer pattern sensors - graphic below), as its Super 35mm sensor has an equal number of RGB pixels. It also has 12 +stops latitude at ISO800, S-Log and S-Gamut.
The F35 (below) and F23 will also be usable with the portable recorder, using a different connection that allows it to dock directly with the camera (the same as the existing SR tape recorder). It also adds the ability to do DPX recording and to work at up to 50 or 60fps (respectively) in 4:4:4 at full HD.
By David Fox
No comments:
Post a Comment