Forum Replies Created

Page 33 of 98
  • Andrew Richards

    April 20, 2012 at 7:17 pm in reply to: One year later…

    [Chris Harlan] “Personally, I can’t remember a single post where-in anyone was predicting an immediate mass exodus from FCP 7. Frankly, I can’t imagine anyone that uses/used FCP7 in business making such a silly claim.”

    I can remember claims to that effect, though not necessarily coming from facility bosses like our esteemed Mr. Raudonis and Mr. Biscardi. Tom Daigon and Grege Burke spring to mind as particularly vociferous mass-exodus soothsayers in this forum who walked their walk into the arms of Adobe and Avid last summer. There were others too.

    Maybe the distinction is immediate mass exodus. But at the very least, there were plenty of exodus predictions to go around last summer.

    Best,
    Andy

  • Yeah, I thought that was Robert’s point. Is that new?

    Best,
    Andy

  • Andrew Richards

    April 19, 2012 at 2:58 am in reply to: The Fog Thickens

    [Juan Morales] “My prediction: within now and 1 year, Avid will be bought by Grant Petty/Blackmagic…

    Anyone thinks the same?!?”

    Buying Avid is buying a world of hurt. Remember, Avid is not just NLE software. Look at their product index. 535 products! Think of all the customers BMD would piss off if they bought Avid and killed 500+ products. And they’d have to kill most of them, because they’d drown if they didn’t.

    Best,
    Andy

  • Andrew Richards

    April 17, 2012 at 9:36 pm in reply to: Opitical Thunderbolt cables. On a Thinkpad.

    [Jeremy Garchow] “4 times the pixels do not equal four time the size in ProRes, there’s more math going on than that.”

    It’s an algorithm, not magic. I generated this graph with the numbers from Apple’s ProRes specs (because their table happened to paste perfectly into Numbers). Note the data rates are proportional to the number of pixels encoded. If double the pixels yields roughly double the data rate, it stands to reason that quadruple the pixels yields roughly quadruple the data rate.

    I’m done now too.

    Best,
    Andy

  • Andrew Richards

    April 17, 2012 at 8:12 pm in reply to: Opitical Thunderbolt cables. On a Thinkpad.

    [Jeremy Garchow] “Frame size does not matter. It is a data rate. The data rate will not go over 330 Mb/sec (with no alpha, and unless something changes) and it usually is under that since ProRes is not CBR, but rather VBR.”

    Not true. Frame size absolutely matters, and there is no cap on ProRes data rates (bitrate = data rate). Apple’s own ProRes documentation lists 2K ProRes 4444 topping out at 370 Mbps, for instance.

    [Jeremy Garchow] “720p60 = 1080i29.97. Same rate.”

    Yes, because it is the same volume of data. 1280×720 is roughly half the pixels of 1920×1080, and 720p60 plays at double the frame rate of 1080i29.97, hence roughly the same data rate. However, look what happens if frame size is the only variable:

    ProRes SD 720×486 @ 29.97 = 34 Mbps
    ProRes HD 1280×720 @ 29.97 = 73 Mbps
    ProRes HD 1920×1080 @ 29.97 = 147 Mbps
    ProRes 2K 2048×1152 @ 29.97 = 168 Mbps

    The data rate is proportional to the frame size.

    [Jeremy Garchow] “ProRes is VBR and scales”

    VBR permits elasticity in terms of the complexity of the image contained in a given frame, and since ProRes is an iFrame codec, there isn’t any efficiency gained by having similar frames adjacent like you get with long-GOP codecs. VBR does not mean the frame size is irrelevant to the data rate.

    [Jeremy Garchow] “4x times the frame size does not equal four time the file size when data rate stays the same.”

    But the data rate does not stay the same when you quadruple the frame size and keep the frame rate and quality setting the same. The data rate is relative. To quote Apple:

    “The actual data rate of Apple ProRes codecs depends on the dimensions, frame rate, image complexity, and quality setting you are using.

    The Apple ProRes format has a target data size for every frame, regardless of complexity, but allows frames to fall short of that target if they are simple (if they cannot benefit in quality from using more bits). Such a shortfall is not reclaimed for other frames; instead, it just produces a smaller overall file.”

    Best,
    Andy

  • Andrew Richards

    April 17, 2012 at 6:56 pm in reply to: Opitical Thunderbolt cables. On a Thinkpad.

    [Herb Sevush] “1000 MBs Read; 831 MBs Write”

    Thunderbolt has the bandwidth to support that (duplex 1250 MB/sec).

    [Herb Sevush] “And even at these speeds I get the green preview line for my 5 camera multicam timelines, using ProRes 720P60.”

    With your 5 stream multi cam ProRes 720p60, you are only pulling an aggregate 100 MB/sec (750Mbps aggregate), or less than 10% of the bandwidth capacity of a single Thunderbolt pipe. You’d have plenty of room left for video I/O. Plus, that 10Gb is bidirectional, so read bandwidth from your RAID does not stomp on your video out bandwidth on the same port.

    Best,
    Andy

  • Andrew Richards

    April 17, 2012 at 6:44 pm in reply to: Opitical Thunderbolt cables. On a Thinkpad.

    [Jeremy Garchow] “Who would 4k ProRes be any different from 1080 ProRes? “

    It’s over 3 times the frame size, for one. Like pretty much any codec, the bitrate on ProRes is relative to the quality (4×4, HQ, etc), the frame size, and the frame rate. The bitrates are not absolute.

    [Jeremy Garchow] “LT is 100 Mb/sec, ProRes is 145Mb/sec, HQ is 220Mb/sec and 444 is 330 Mb/sec at the very top level, those are bits, not Bytes. It’s a VBR codec so usually the rates are lower than that.”

    Embarrassing mistake on my part, yes, bits not bytes. But those are the bitrates for 1080 @ 29.97. Everything else being equal, 4K would be more than double the bitrate of 1080 at the same quality level and frame rate. That works out to roughly 500 Mbps or 62.5 MB/sec. So not that bad after all, even for several streams.

    I stand by my earlier point though, that Thunderbolt is not a step up for anything that already has PCIe slots. It is a huge step up for machines that never had them. Still, you’re not going to be able to hang as much outside a computer with Thunderbolt as you can hang inside a computer with PCIe slots. I’m glad Thunderbolt is opening up so much potential in iMacs and MacBooks, but I would hate for it to be the excuse for killing off Macs with PCIe slots (whatever they be called).

    Best,
    Andy

  • Andrew Richards

    April 17, 2012 at 5:19 pm in reply to: Opitical Thunderbolt cables. On a Thinkpad.

    [Jeremy Garchow] “I guess if you’re working in uncompressed 4k.”

    Even Pro Res 4K would be in the 500MB/sec range depending on the flavor.

    [Jeremy Garchow] “My fibre connection is 8Gb now. Seems to be all good.”

    And that 8Gb FC link is only doing one thing. If you are cramming 500-800MB/sec storage I/O and 200-250MB/sec video I/O into a 1.2 GB/sec Thunderbolt pipe, it could get crowded. If you are doing one thing at a time with a Thunderbolt bus, it is a lot of bandwidth. If you are splitting it up among a few bandwidth-hungry devices in a daisy chain, it isn’t.

    Best,
    Andy

  • Andrew Richards

    April 17, 2012 at 5:09 pm in reply to: Opitical Thunderbolt cables. On a Thinkpad.

    [Herb Sevush] “High speed raid. Can’t get enough speed. Multicam is a beast on bandwidth.”

    How fast is your RAID today?

    Best,
    Andy

  • Andrew Richards

    April 17, 2012 at 5:00 pm in reply to: Opitical Thunderbolt cables. On a Thinkpad.

    [John Kaley] “I looked at that workstation too, and had to image the jet-engine roar the fans must make when that thing’s at full-tilt boogie.”

    Sure, at full tilt, but then so is a Mac Pro at full tilt. The idea is several large diameter fans turning slowly can cool very quietly. Presumably, the ProMax rig is monitoring its internal temps and has a fan controller only turning the them as much as it needs to to keep cool.

    Best,
    Andy

Page 33 of 98

We use anonymous cookies to give you the best experience we can.
Our Privacy policy | GDPR Policy