Announcement

Collapse
No announcement yet.

What about Graphic Card Acceleration?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • What about Graphic Card Acceleration?

    Are the new editing software using graphics cards (AGP or PCI express) to help the final render?

    Which software (in theory, Pinnacle Studio 9 has an option inside)?
    Media Studio 7, Premiere Pro, Avid Xpress, Vegas... ?

    What they acdelerate: only preview, only final render, both, none...?

    Which brand is NOW or in the FUTURE better: NVIDIA or ATI?

    I heard few months ago that ATI, PINNACLE and INTEL join forces to create a Graphic Card to help video editing. Are they working in the project?

    Thanks a lot.


    P.D: Another question is about 64 bits software. Systems and S.O. are avaible now, but there is no editing software announced. Any news?

  • #2
    NO...the graphics card does not affect rendering speed in MSPro or most other editors absent a hardware editing card or software plugin that calls on it.

    Generally plugins like Canopus XPlode or Boris RED use the gfx card's OpenGL 3D functions to do fancy page curls, bounding boxes etc. Most other plugins do not.

    Even with fancy 3D effects it still doesn't take much in the way of gfx horsepower on the cards part. The RT.X100 uses an embedded G-550 to do its very fancy 3D effects in realtime.

    For most editors the main utility for display cards in editing is to provide an excellent 2D display (to cut eye strain), provide a high quality composite/S-video output (for previews and recording to SVHS/VHS) and multimonitor support.

    For this I find the Matrox Parhelia and P-750 to be top notch and they work particularly well with MSPro7.

    Dr. Mordrid
    Last edited by Dr Mordrid; 11 October 2004, 11:59.
    Dr. Mordrid
    ----------------------------
    An elephant is a mouse built to government specifications.

    I carry a gun because I can't throw a rock 1,250 fps

    Comment


    • #3
      I have found more information.



      And I think that everthing goes to 64 bits, if you want to edit in HD, of course.
      Another hardware-software change is inminent.


      "HDV at 720p30 requires all the CPU power that's currently available, and it's obvious that the moment we begin to edit 1080i, we will need a computer two times more powerful. Without that much power, smooth playback and realtime effect previews may not be possible. At NAB 2004, the president of a company highly regarded for its codec development and NLE systems told me that editing 1080i30 would likely require a dual Xeon PC. Is there another way?

      One potential solution is to use graphics processing unit (GPU) power rather than CPU power. Modern GPUs offer a hardware-based inverse discrete transform (iDCT) function that can offload much of the decoding computation from the CPU(s). This is accessed via the DxVA function in DirectX 9. Future NLEs might obtain the necessary computing power by using a GPU. (See “Home-Built HDV Editing,” p. 39.) Apple is following the same path by implementing Core Image/Video in the 10.4 Tiger version of OS X. At WWDC 2004, Steve Jobs explained, “…although CPU speed increases are slowing down, GPU speeds are increasing dramatically. Therefore, it makes sense to take advantage of that processing power to manipulate images and video.”

      Over time computer systems will become more powerful, and native HDV editing software will become more sophisticated. So eventually we will be editing HDV just as we now edit DV."

      Comment


      • #4
        Note that HDV will come in two flavors: 1280x720p and 1920x1080i. The first is JVC's standard and the second Sony's. Lots of bits in either one.

        Premiere Pro is going to a wavelet compression scheme similar to JPEG2000 for HDV editing (they're licensing Cineforms tech) in an attempt to lessen the system impact.

        While wavelet compression will be fast and easier on the CPU, peripheral busses and drive subsystem it will also cause the source files to suffer 2 transcodings; on capture and on export to MPEG-2, VC-1 or H.264 (the HDVD legal formats).

        If, on the other hand, the editing software were to work in the native HDV format for quality reasons then the user would indeed find themselves in need of a much heftier system.

        The benefit would be only one transcoding; on export.

        I'd say in that situation a PCI-X based system with a hardware RAID (probably SATA or SCSI to handle the data flow) would be the way to go.

        Because of the massive amout of graphics data involved in editing HDV a PCI-X display card would also be handy.

        Another alternative to make HDV manageable would be for someone to bring back a feature MSPro had for ages but dropped several years ago: proxy editing.

        With proxy mode you could work with a low rez versions of your sources and only doing a high quality rendering of the changes when the project was exported.

        I can remember people screaming to the rafters when proxy mode was dropped, and with good reason. It was a great feature.

        Dr. Mordrid
        Last edited by Dr Mordrid; 20 October 2004, 07:53.
        Dr. Mordrid
        ----------------------------
        An elephant is a mouse built to government specifications.

        I carry a gun because I can't throw a rock 1,250 fps

        Comment


        • #5
          SDI / HDSDI on consumer equipment would be nice.
          ______________________________
          Nothing is impossible, some things are just unlikely.

          Comment


          • #6
            I don't understand very well why they are changing the non intraframe compression.

            The best of DV is that every frame is individual. All of the new "domestic" compressions are intraframe. Bad, bad, bad...

            I have done few numbers:
            DV PAL 720x576 = 414,720 Pixels x 25 frames = 3,5 Megabytes/sg (with audio NON COMPRESSED)
            (Hipotetical) HD-DV PAL 1280x720 = 921,600 Pixels x 25 frames = 7.5 Megabytes/sg (also with audio NON COMPRESSED; of course NON INTERLACED).

            Well, here are my questions?
            - 7.5 Megabytes/Sg (or 10 Megas), NOW, at this time, IS A VERY BIG TRANSFER RATIO????
            - What's the problem? The tape? If probably in a pair of years videocams will work with Blue Ray Discs or HD or Solid Memory... with double or triple capacity versus DV TAPE (15 Gigas aprox). And THEY CAN'T MAKE High Quality Tapes with the same or little fast speed to reach 30 Gigas????
            Why are THEY looking for DV Tape compatibility if the dissaventages exceed aventages???????
            - Is the conexion with the computer the problem??? Well, if it's true, it's only with TAPES. Discs, HD or Memory have their way. And don't forget that Firewire 400 can transfer much more than 3,5 megas. And there is Firewire 800, and USB 480, and SATA 150 (why not?) or SATA 200, etc.

            I think that HD videoediting is a revolution. Years ago, five, you can't change your computer easily. Now, you can afford 64 bits computing, a pair of 200 gigas HD and 1 Giga RAM or more without lot of pain.

            Sincerely, I don't understand what are THEY DOING?

            DV was a big improvement, in 1995. Ten years after... ONE STEP BACK????

            Comment


            • #7
              Originally posted by Fluff
              SDI / HDSDI on consumer equipment would be nice.
              Blackmagic Design creates the world’s highest quality products for the feature film, post and broadcast industries including URSA cameras, DaVinci Resolve and ATEM switchers.


              Despite my nickname causing confusion, I am not female ...

              ASRock Fatal1ty X79 Professional
              Intel Core i7-3930K@4.3GHz
              be quiet! Dark Rock Pro 2
              4x 8GB G.Skill TridentX PC3-19200U@CR1
              2x MSI N670GTX PE OC (SLI)
              OCZ Vertex 4 256GB
              4x2TB Seagate Barracuda Green 5900.3 (2x4TB RAID0)
              Super Flower Golden Green Modular 800W
              Nanoxia Deep Silence 1
              LG BH10LS38
              LG DM2752D 27" 3D

              Comment


              • #8
                Actually DV was a step BACK in quality vs. analog capture/editing.

                The first problem is that it's more higly compressed: 5:1 vs. 3:1 or less for most MJPegs. This could only be achieved by the following;

                Analog video has a 4:2:2 colorspace, the same as broadcast, and composites very nicely. This provides a resolution of 360 samples wide horizontally with each taking up 2 pixels.

                NTSC DV uses a 4:1:1 colorspace, making it tough to composite because its color channels only have a horizontal resolution of 180 samples with each spread over a 4 pixel wide space.

                This problem can be seen easily if you do chromakeys or overlays that have much detail.

                DV may look better than analog to the untrained eye, but that's only because a higher proportion of DV's data is from the lumance channel and the human eye "likes" luma over chroma. Looks are one thing; compositing is another.

                These days to get really high quality composites with DV you either have to upsample it to 4:2:2 (or better yet 4:4:4) in software or get a hardware board like the RT.X100 or Storm2 that do this in hardware.

                As far as editing MPEG-2 goes; I agree that if you edit in "normal" MPEG-2 as we know it (Main Level/Main Profile at 4:2:0 and a 15-18 GOP) you get marginal results BUT there are editing MPEG-2 profiles that use High Levels and Profiles and just a 6 GOP (or smaller). Some MPEG editing profiles even use a 4:2:2 colorspace.

                There are even such profiles for MPEG-4, and they are very nice indeed. It only has a bad rap because the profile most of us are used ot is SP (Simple Profile), which is intended just for the internet. MPEG-4's high level profiles are also a thing to behold.

                Dr. Modrid
                Last edited by Dr Mordrid; 20 October 2004, 11:32.
                Dr. Mordrid
                ----------------------------
                An elephant is a mouse built to government specifications.

                I carry a gun because I can't throw a rock 1,250 fps

                Comment


                • #9
                  A few things here:

                  Mikele - You're forgetting that each pixels needs 24 bits (3 bytes) to represent it and your conversion math is off. Raw full-frame PAL requires about 30MB/sec.

                  414720X3bytes/pixelx25frames/secx1kB/1000Bx1Mb/1000kB=31.104MB

                  As for why miniDV format tape? Who knows, perhaps the tape manufacturers wanted to keep the miniDV revenue and were fearful a format change would be a move to hard drives in camcorders. For now we're stuck with this new MPEG-2 TS (transport stream) format. Don't worry though, it's a BIG step up in visual quality from DV

                  Some other info:

                  HDV formats are 1280x720 (720p) and 1440x1080 (1080i).

                  Presently, only Pinnacle's Liquid Edition 6 makes substantial use of the GPU for 2D and 3D rendering. They're website claims 1000's of GPU enabled effects. Most any current OpenGL card should have the processing power to perform such calculations.

                  Finally, I agree with Doc regarding Proxy mode. It's a great way to sidestep all of the processing overhead problems associated with these new HD formats.

                  Reducing the resolution of the original content by a factor of 4, ie 1280x720 (720p) resampled to 640x360, or 1080i to 720x520 could create interframe compressed proxy files with resolution comparable to present day DV, which most current day computer can handle quite nicely, include real-time editing.

                  In fact, there could be various options for encoding the proxy files to different compressions depending on the preference of the user.

                  Although Cineform claims that their codec provides superior results than actually working with the raw HDV stream since they convert to 10bit color samples (from 8), and they claim these samples "hold up" better though editing math, it's hard to believe that a transcoding can be GOOD for any video signal.

                  Finally, one big advantage for us NTSC users is that HDV uses 4:2:0 sampling, just like the final MPEG-2, WMV, H.264 format so additional color information won't be discarded during final render due to 4:1:1 to 4:2:0 colorspace conversion.

                  Now if HD-DVD or Blu-Ray would just drop out... here we go with another "+" and "-" type mess.

                  - Mark
                  - Mark

                  Core 2 Duo E6400 o/c 3.2GHz - Asus P5B Deluxe - 2048MB Corsair Twinx 6400C4 - ATI AIW X1900 - Seagate 7200.10 SATA 320GB primary - Western Digital SE16 SATA 320GB secondary - Samsung SATA Lightscribe DVD/CDRW- Midiland 4100 Speakers - Presonus Firepod - Dell FP2001 20" LCD - Windows XP Home

                  Comment

                  Working...
                  X