Digital Processing Systems presents dpsReality, the SDDR of the future.
This FAQ is intended to further your knowledge about the dpsReality.
Q.
What is Reality?
A. Reality
is a Studio Digital Disk Recorder for animators, compositors and post
production artists.
Q.
Is Reality a hardware or a software solution?
A. Well, it is both. When we say Reality, we are really talking about
a complete solution, a combination of both hardware and software capabilities
that together were designed for the way an animator or compositor works.
Q.
How is Reality different from a Perception (PVR) or Hollywood boardset?
A. Reality is really a combination of the best of these two products.
It shares common features with each such as on board disk control and
video I/O, additionally Reality adds sophisticated features like uncompressed
video and alpha storage, uncompressed video or alpha playback, 4 track
audio, Ultrawide SCSI 3 disk control, and realtime VGA preview capabilities
to name a few.
Q.
What is VTFS and how is it different form VFS?
A. VTFS overcomes some shortcomings of the original VFS design and
gains features, speed and flexibility. The main difference from a user
point of view is that VTFS files must be prespecified in order to be used
on disk. This may sound like an extra step, but when you compare how the
two systems work, it becomes apparent how much better the VTFS is at handling
typical animation and compositing information.
VTFS
The Virtual Tape File Systems (VTFS) is the heart of the DPS Perception
DDR software technology. With the VTFS, users can save standard system
compatible files such as targa and bmp, directly to the Perception dedicated
disk subsystem. Those images are instantly and transparently converted
to video by the VTFS. Most of the other systems on the market also allow
transparent conversion from memory to video playable files through the
use of an AVI stream. In those systems, an application must either
fully support the Microsoft AVI model for loading and saving AVI files,
or a third party program must be used for conversion between sequences
of individual files, and the final output AVI file format. In VTFS, any
application that can write one of nine standard system file formats is
automatically compatible for both reading and writing directly to video
streams, through the VTFS. In the VTFS model, the file system represents
each video stream through a virtual model which allows each video stream
to appear as multiple file formats simultaneously. This does not require
more images to be stored, the actual alternate file formats (.tga, .bmp,
etc) are all represented ‘virtually’ by the file system.
VTFS files are organized on your hard disk in a specific hierarchy and it is easy for the user to find the particular file in question. VTFS files are located physically on the disk drives in much the same manner as your operating system would store your normal system documents. That is to say that they are stored on the disk in the largest area of free disk space. The problem with this method is that the result of this is disk fragmentation. The VTFS copes with this somewhat better than normal system drives because no system files are being stored on the VTFS drives and also because the nature of the files (video) is that they are relatively few, but very large files so there is less chance of fragmentation. Normally when a user wants to create a video clip from say a 3D animation package, the user only has to choose a base name for the clip (birdfly) and the package will automatically append the appropriate frame number to the end of the name (birdfly001.bmp). The VTFS will recognize the file as it is saved to the disk and automatically append it to the last frame in the sequence. Repeatedly saving files with numerically greater numbers (birdfly002, birdfly003, etc) will automatically fill out the video sequence. This was the way that the older VFS (virtual file system) coped with images sequences.
So far we have discussed features that existed in the original VFS architecture. Now on to VTFS. VTFS builds on the structure of the VFS, adding several new features to the file system. The most visible feature to the user is that you can pre-stripe areas of disk for specific projects, in much the same way that you would pre-stripe a video tape. For instance you can specify to the file system that birdfly.rvd will be 1000 frames in length, and that each frame will be stored at a specified compression ratio (which also means a fixed file size). Because the VTFS knows ahead of time the parameters the user has chosen, it is now possible to write frames to disk out of order. The VTFS will receive each frame and dynamically compress each one so that it fits within the user specified file size. Now that files can be written out of order, it opens the door for files to be written by multiple machines, over a network. A system based on AVI or Quicktime does not have this luxury. The system will not allow for these normal data files to be open multiple times for write access so each machine on the network would have to render its own AVI or Quicktime which must then be assembled later in another application.
A second feature of VTFS is that it can natively handle 4:2:2:4 video streams. These are normal YUV data streams plus an associated alpha channel. Because the hardware supports uncompressed video playback, both the video and alpha streams can be uncompressed, but only one of them may be played back in realtime (hardware limitation). The VTFS configuration application will also allow you to specify any combination of compressed and uncompressed for either channel (uncompressed video, compressed alpha for example).
A third feature of the VTFS is that it now uses the core image loading/saving and processing library from Digital Fusion. This library runs as a service for the filesystem. Because of this there is a guaranteed consistency in image translation (color format conversion) both into and out of the file system, and through Digital Fusion or any other application that uses this standard. Also because it is using the Digital Fusion engine, it is faster and more robust than other third party image libraries.
The fourth major feature of the VTFS is its support for full duplex rendering/playback. This means that it is possible for a user to be playing back video in realtime (even uncompressed) while reading and writing to the file system. Even remotely over a network. This allows Reality to play a pivotal role in a workgroup environment as both a common media pool and an instant playback device, something no other board can do.
Q.
What software applications does Reality work with?
A. Reality will work with any application that can read or write to
any of the popular file formats supported by the file system (VTFS). Those
formats include:
- SGI 24 and 32 bit
- TGA 24 and 32 bit
- BMP 24 and 32 bit
- PIC (Softimage) 24 and 32 bit
- TIF 24 and 32 bit
- IFF 24 bit
- VPB (Quantel) 24 bit
- RAS (Sun raster) 24 and 32 bit
- RLA 24 bit and 32 bit (the file system will discard extraneous channels)
Q.
How do I use Reality’s alpha channel capabilities from within my favorite
3D application?
A. Any program that can read or write one of the standard file formats
supported by the VTFS will work with Reality. Simply save or load files
as you would from a normal system disk and Reality will handle the rest.
Q.
Is Reality supported by Digital Fusion?
A. Yes, Reality is natively supported by Digital Fusion 2.5, Digital
Fusion POST, and Digital Fusion DFX.
Q.
Will Reality work with Adobe Premiere?
A. At this time, no.
Q.
Can I import video clips from my Perception (PVR) or Perception RT?
A. Yes. Reality includes a utility program which can convert Perception
(PVD) or Perception RT (RVD) files to the native Reality (DPS) format.
Q.
Can I import video clips from my Video Action RT?
A. Yes, Reality can both import and export to Video Action RT.
Q.
What does the Reality solution come with?
A. Reality ships with the main RDR3 board, software, breakout cable,
and manuals.
Q.
Are there any options available for Reality?
A. Yes, there is an SDI I/O option which includes single SDI input
and dual SDI outputs. This will also upgrade the Reality board to full
AES/EBU audio support. This upgrade module can be additionally upgraded
to add real-time DV I/O.
Q.
Is there a breakout box available for Reality?
A. Yes. A professional rackmountable breakout box is available which
consolidates all of the dpsReality audio and video I/O connections. The
breakout box also provides digital audio and video connections for the
optional SDI and DV I/O cards.
Q.
How does Reality compare to Perception in terms of system requirements?
A. Reality is a 3.3v universal PCI bus design which allows it to work
in many PC platforms such as the SGI Visual workstation 320. It will also
work in any 5v (standard) PCI equipped workstation.
Q.
What types of effects are possible with Reality?
A. Reality has a basic mix system on board and has the ability to
do real time overlay of graphics, logos, or titles.
Q.
How many PCI slots does Reality need?
A. One PCI slot for the main board, a second slot if the optional
SDI I/O board is used.
Q.
Does the Reality have a built in disk controller? And how does it compare
to the Perception and Perception RT?
A. Yes it does, the disk controller on Reality is both faster and
more flexible than the ones we have had previously on the Perception and
Perception RT. It is capable of striping together as many as 15 drives
into one single volume. The Perception had no real drive striping capabilities
and the Perception RT can only stripe two drives at a time. The Reality
disk controller is able to transfer the full 40MB/s of the UW SCSI2 specification.
Q.
dpsReality supports uncompressed video recording and playback, how many
drives does it require to achieve the necessary bandwidth?
A. Uncompressed video playback is possible with a single, very fast
hard disk, but for practical reasons, we recommend two 10,000 RPM drives
such as the Seagate Cheetah or IBM Ultrastar series.
Q.
I understand that the Reality supports 4 track audio. How many physical
I/Os are there?
A. Yes, internally the Reality supports up to 4 tracks of audio for
mixing purposes with 4 parametric EQ’s (4 band). There are two (stereo)
channels of audio available on all audio inputs and outputs.
Q.
What sampling format is used for the video data that is stored on disk?
A. 32 bit 4:2:2:4 through the file system and 24 bit 4:2:2 for the
recorded video.
Q.
What format of compression does Reality use?
A. Motion JPEG (MJPEG) or uncompressed YUV
Q.
What compression Ratio does Reality use?
A. Reality is capable of utilizing any compression ratio as well as
handling uncompressed video. Data rates of greater than 20MB/s are possible
with frame sizes more than 690KB/frame.
Q.
What is bit rate control, what is Feed Forward Entropy Calculation, why
is it better?
A. Bit rate control is the mechanism that allows video to be recorded
at a pre-specified target data rate. For instance if you tell the software
to record video at 5 MB/s, the bit rate control mechanism modulates the
MJPEG compression engine, telling it how much compression is required
to maintain the 5 MB/s data rate.
A. Feed Forward Entropy Calculation is a technology developed by DPS to better control the dynamic compression of motion video. It has a large impact not only on the quality of the recorded video, but also on the systems ability to handle sudden changes in video content. The competition uses a bit rate control mechanism based on previous field data for bit rate control that is built into the popular Zoran MJPEG compression chipset. This mechanism is based on previous field data. Feed Forward Entropy Calculation is better than the Previous Field method used in the Zoran chip because the system cannot be ‘surprised’ by any sudden changes in video content, and it gives us the same dynamic control over video compression that the other systems have.
Q.
What does x : 1 compression mean in MB/sec?
A. When calculating video compression, the baseline number is the
frame size, times the number of samples per pixel. In the case of NTSC
video it is 720 horizontal samples * 486 lines * 2 samples per pixel.
That is a total of 699840 bytes per frame. Multiplied by 30 (frames per
second) the total video data rate is 20995200 bytes per second, or about
21 MB/s. 3:1 compression ratio is equal to about 7 MB/s as a data rate.
Q.
What video I/O does Reality come with?
A. Reality ships standard with Component (Y/R-Y/B-Y), Svideo, and
composite. SDI and DV are available as an upgrade.
Q.
What audio I/O does Reality come with?
A. Standard stereo I/O includes: Balanced and unbalanced (RCA) analog.
Digital formats such as S/PDIF (RCA and BNC, BNC connectors conform to
AES on coax spec BNC), and AES/EBU digital are available with the SDI
option board.
Q.
Does audio stream with the SDI video?
A. Yes, 4 tracks of digital audio can be embedded.
Q.
Does Reality work over networks?
A. Reality supports network import and export. This may be accomplished
in several ways, data may be imported or exported directly to/from another
Reality system, or file formats can be imported and exported directly
to/from the VTFS in any supported file format. Reality also supports network
rendering from multiple sources to the Reality disk array.
Q.
Will it work over a (fibre channel) network?
A. Reality supports fibre channel or any other system level network
device for import and export of data.
Q.
Does Reality support Genlock?
A. Yes
Q.
Can I add external Hardware such as a modem, printer, art tablet to a
Reality system?
A. Yes, adding standard off the shelf third party hardware or software
is possible.
Q.
How much storage space do I need for a one hour program?
A. Storage space is dynamic and will be effected by the compression
ratio of the video data. For a rule of thumb, a broadcast quality video
stream of about 5:1 compression will easily yield over 1 hour of storage
on a 18GB drive.
Q.
Will it keep audio sync?
A. A/V sync is guaranteed.
Q.
Will it do 16 x 9 and/or HDTV? Which HDTV format?
A. There are several different ways of approaching 16:9 production.
One way is to shoot letterbox to the correct aspect ratio. Another way
is with an anamorphic squeeze which will squeeze 16:9 data onto a 4:3
frame (everyone will be tall and thin). Reality can work with either one
of these formats. If you are using anamorphic data, put your video monitor
in 16:9 mode to view the image in the proper aspect ratio. Reality is
not compatible with the ATSC HDTV specifications for progressive scan
video.
Q.
How do I back up / restore my work?
A. Reality will support backup/restore through the file system, and through
project management (future version).
Q.
Can I use a media server for shared storage with my Reality system?
A. Reality is designed as a distributed media pool system, not a common
media pool. You can share data between two or more Reality systems, or
third party media pools can be used for offline storage of Reality media.
For playback, media must be copied locally to the Reality media drives.
Q.
Can I print to tape in insert or assemble?
A. Yes, and Reality also supports batch digitizing.
Q.
Does Reality have support for importing animation files or clips?
A. Yes, through the file system and through Digital Fusion
Q.
What is the impact on rendering speed in my 3D or compositing application
when saving to the Reality VTFS?
A. Reality will have little to no impact on rendering or compositing
speed. This is partially do to the design of the board which can utilize
DMA in conjunction with hardware color space converters (RGB <->
YUV) to rapidly process and store image data. Saving data is usually a
small part of the actual rendering or compositing process, usually an
application will spend more time manipulating image data than it will
saving the data.
Q.
How does Reality compare to MPEG-2 based editing systems?
A. MPEG-2 and JPEG are based on the same basic principles. They both
use the DCT transform together with some sort of compression to reduce,
sort, and pack the image into a smaller file size. Since this compression
is accomplished on a frame by frame (or field by field) basis, it is known
as interframe compression and the resulting images can be referred to
as ‘I’ frames. MPEG differs from JPEG in that it can additionally store
only the differences between frames. MPEG calculates what has changed
between frames, and stores only the changed data (intraframe compression).
If this data that is stored only contains information for video moving
forward in time, the data that is stored is called a ‘P’ frame (predictive).
If the data that is stored contains information for both forward video
play, and reverse data, then it is called a ‘B’ frame (bi-directional).
While bi-directional or ‘B’ frames are a bit self explanatory, the ‘P’
frame name is a bit misleading. ‘P’ frames do not contain predictive data,
they in fact contain absolute data that has been derived as the difference
between one frame and the next. The ‘predictive’ part comes as the playback
engine must reconstruct the full frame given two ‘I’ frames and all of
the ‘P’ frames in between.
Both JPEG and MPEG-2 systems are capable of working in a JPEG like ‘I’ Frame only format. This loses all advantages of MPEG-2, but it does allow it to be frame accurate. However, this is very rare these days. Most MPEG-2 systems are based on using both I and P frames with some systems varying the number of ‘P’ frames so that they can claim more accurate editing, but still not as accurate as JPEG. Some systems store data in an IPPP format. This grouping is referred to as a GOP structure or Group Of Pictures structure. In order to recreate any given ‘P’ Frame in a video clip, the system must decode the entire GOP structure.
Because the MPEG-2 system has to rebuild the requested frames, and the following P frames, the result will not be the same as the original. It will be an approximation of the original. The more expensive the MPEG encoder, the better the approximation will be. There will also be some delay while the system decodes the appropriate frame(s). Another factor to keep in mind is that most low cost MPEG-2 encoders are 4:2:0 sampling. MPEG-2 has many benefits such as lower cost of storage (data rate is much lower), but for the professional editing environment, those benefits are far outweighed by the degradation in quality that MPEG-2 encoding will incur.
Reality specifications (subject to change)
Video Inputs:
- Composite (BNC)........................1V p-p, 75 Ohm
- S-Video (4-Pin Mini-DIN) - Y Signal (1V p-p, 75 Ohm), C Signal (286mV p-p)
- Component Betacam (BNC's): - Y Signal (1V p-p, 75 Ohm), R-Y, B-Y (Pr,Pb) - 700mV p-p
Note(s):
1) Video inputs levels correspond to 75% color bar signal, with 100% white
reference bar.
Digital Video Inputs (Available with Digital I/O board):
- Serial SDI (SMPTE 259M) (BNC)........800mv, 75 Ohm, Auto EQ to 300m.
- DV...................................IEEE-1394.
Note(s).
1) Embedded AES audio conforming to SMPTE-272M is supported on the SDI
input. Audio may be synchronous or asynchronous with respect to both the
SDI video, and system audio sample clock.
Audio Inputs (2 channel/stereo pair):
- Balanced (XLR-F).....................+4 dBU Nominal, +20 dBU Maximum
- Unbalanced Line (RCA).............-10 dBV Nominal, +6 dBV Maximum
- Unbalanced AUX (RCA)............-10 dBV Nominal, +6 dBV Maximum
- LTC Input (RCA)........................0.1-5.0V p-p.
Digital Audio Inputs (Available with Digital I/O board, 2 channel/stereo pair): Professional Format:
- AES/EBU (EIAJ CP-340, XLR).......5V Balanced, 110 Ohm.
- AES/EBU (SMPTE-267M, BNC).....1V Unbalanced, 75 Ohm.
Digital Audio Inputs (Available with Digital I/O board, 2 channel/stereo pair) Consumer Format:
- S/PDIF (RCA).....................500mv Unbalanced.
Note(s):
1) Digital audio data may also be derived from either of the SDI or DV
input data streams.
Video Outputs: Primary (Program):
- Composite (BNC)..................1V p-p, 75 Ohm
- S-Video (4-Pin Mini-DIN): Y Signal (1V p-p, 75 Ohm), C Signal (286mV p-p)
Video Outputs: Secondary (Preview/Alpha channel):
- Composite (BNC)....................1V p-p, 75 Ohm
- Component Betacam (BNC's): Y Signal (1V p-p, 75 Ohm), R-Y, B-Y (Pr,Pb) - 700mV p-p
Time Code Video (VITC):
- Composite w/VITC (BNC).......1V p-p, 75 Ohm
Note(s):
1) Video inputs levels correspond to 75% color bar signal, with 100% white
reference bar.
2) The secondary video output(s) are software selectable between composite
and component formats.
Digital Video Outputs (Available with Digital I/O board, 2 channel/stereo pair):
- Serial SDI (SMPTE 259M) (BNC x 2)....800mv, 75 Ohm.
- DV...................................IEEE-1394.
Note(s).
1) Embedded AES audio conforming to SMPTE-272M may be inserted into the
SDI output stream.
Audio Outputs (2 channel/stereo pair):
- Balanced (XLR-M).....................+4 dBU Nominal, +20 dBU Maximum
- Unbalanced Line (RCA)............-10 dBV Nominal, +6 dBV Maximum
- LTC Output (RCA).....................1V p-p.
Digital Audio Outputs (Available with Digital I/O board): Professional Format:
- AES/EBU (EIAJ CP-340, XLR).......5V Balanced, 110 Ohm.
- AES/EBU (SMPTE-267M, BNC)....1V Unbalanced, 75 Ohm.
Digital Audio Outputs (Available with Digital I/O board): Consumer Format:
- S/PDIF (RCA).....................500mv Unbalanced.
Note(s):
1) Digital audio data may also be inserted into both of the SDI and DV
output data streams.
Genlock Inputs:
- Composite (BNC)......................1V p-p, 75 Ohm
Timing Adjustment Range:
- Horizontal.......................> +/- 2usec
- Subcarrier.......................360 Degrees
Note(s):
1) The genlock video input is shared with the main composite video input.
In normal operation, the video output is genlocked to the video signal
applied the the composite video input. Video output timing may also be
derived from an on board crystal reference oscillator when no reference
video signal is available.
Hard Drive Interface:
- Internal/External 68 pin............Ultra Wide SCSI-2.
Note(s):
1) Compatible with Ultra Wide SCSI-2, Wide SCSI-2, and Fast SCSI-2 type
devices.
2) Compatible LVD devices operating in single ended (SE) mode.
Native Raster Size:
- 525 Line (NTSC).....................720 x 486.
- 625 Line (PAL)........................720 x 576.
Video Performance:
- Frequency Response..................0 - 5.0 MHz (+/- 0.25 dB), 5.0 - 5.5 Mhz (-1.5 dB)
- K-Factor (2T)......................................< 1%
- Differential Phase...............................< 1 Deg.
- Differential Gain.................................< 1%
Signal/Noise:
- Playback..................................> 68 dB
- Capture + Playback...............> 60 dB
Audio Performance:
- A/D.........................................18 Bit oversampled Sigma-Delta.
- DAC.......................................18 Bit oversampled Sigma-Delta.
- Sample Rates........................11/22.05/32/44.1/48 KHz.
- Frequency Response.............20 Hz to 20 KHz, +/- 0.5 dB.
- Dynamic Range......................88 dB (A weighted).
- Total Harmonic Distortion.......0.005% Max
Power Consumption:
- +5V.................................1.5/3.0 Amp
- +12V................................400 mA
- -12V................................80 mA
- Total...............................13/20 Watt
Note(s):
1) Higher values include DVE module.