Bj Ahlen
Forum Replies Created
-
The link is definitely down.
Allume StufIt reads and writes .sit files really well, I have used it for many years.
-
Totally outstanding! Most interesting teaser I have seen in at least 10 years.
Great work, and thanks for showing us on the other half of the globe (the one that’s at the bottom of your world maps :O)!
-
Totally outstanding! Most interesting teaser I have seen in at least 10 years.
Great work, and thanks for showing us on the other half of the globe (the one that’s at the bottom of your world maps :O)!
-
Very clear, Andrew.
Viewing interlaced footage on a progressive display such as an LCD is rarely pretty if there’s motion involved. Old problem that’s getting much bigger as CRTs are discreetly disappearing out the back door.
Better deinterlacing algorithms make a big difference.
And for looking at pixel level blowups, it would be helpful to do some chroma smoothing, too.
If outputting to MPEG-2 for DVD or broadcast, chroma smoothing won’t have that impact on the 4:2:0 output though, because the HDV source is also 4:2:0.
-
I don’t see how 4:2:0 could cause interlacing artifacts. Perhaps you are thinking of encoding signal processing time, but this affects odd and even fields equally.
Interlace artifacts are caused by trying to capture images of objects in fast motion, with the odd lines being recorded 1/30th or 1/25th of a second apart from the even lines.
Even if you capture uncompressed, you still have the physical reality of being unable to match the odd and even lines.
You need a better deinterlacer than what’s in the AE box.
The Re:VisionFX Fieldskit AE plug-in (around $90) could do it, as could Magic Bullet Suite.
-
The 3GB switch is an OS issue, unrelated to Intel vs. AMD.
(BTW, Intel CPUs nowadays are AMD-compatible, meaning Intel had to add AMD’s added instructions, after Microsoft and several other major vendors told them they wouldn’t accept an incompatible Intel alternative).
For the deeper scoop on the 3GB switch, see Tam Tam on The oft-misunderstood 3GB switch.
The user has to enable the switch, and the application has to request the additional memory. Even then the memory the application receives may not be contiguous.
I’m also longing for the next step beyond the [in]famous “Why would anybody ever want more than 640K?XXXX64MB?XXXX256MB?XXXX2GB?”
-
Numerous comparisons have shown that AMD Opterons are way faster than Intel Xeons for anything related to video (Lucasfilm bought 1,000 Opteron workstations as soon as they got wind of them, I imagine their existing Intel boxes were discreetly dropped off at the loading dock for pickup).
AMD X2 dual core processors are also much faster compared to Intel D dual core processors. Note that AMD CPUs may even be more expensive than “corresponding” Intel CPUs, AMD has seen their performance premium as real value.
AMD has a much more advanced CPU architecture, and a much faster internal bus that doesn’t keep the CPUs waiting.
Intel has fired a whole bunch of executives because of this lost battle, but they are not expected to catch up until 2008-2009.
Whole brand name (Gateway, HP) PCs with AMD X2 4400+ for example are now available for about $1,000-$1,100. The only thing holding AMD back for the moment is manufacturing capacity, where Intel of course has a lead.
There are also truly great motherboards from Asus and some others if you want to build your own machine for less.
The only thing to watch out for it to get a motherboard with nVidia peripheral chips, rather than with the VIA chips that have given AMD an undeservedly bad name in the past.
The vast majority of Hollywood companies are buying AMD-based workstations now, they can’t waste time waiting for slower technology to do effects rendering.
-
CinemaCraft’s strength is in encoding progressive footage (because that’s what film transfers are).
Procoder’s strength is in encoding interlaced footage (because that was their bread and butter).
Multipass helps more at low bit rates (relative to the amount of motion).
-
Great, many thanks!
That gives me courage, as I have seen how sensitive this card is to the hardware configuration.
I seem to recall that BM recommended soft RAID instead of the hardware RAID for better performance (with the SATA controller configured for JBOD.)
Now my P4 3.2E is just about dead to the world when trying to do anything else during capture, it really takes over the machine.
Are you getting any left-over CPU cycles when capturing 10-bit via SDI with your AMD dual-core?
-
Aaaahhh, relief. Thanks for the feedback!
Just a few questions:
1. Are you using a PCI Express DL Extreme card, or the PCI version?
2. Are you using SATA RAID with the Asus board? I heard the SATA chipset could consume all the bandwidth. preventing the DL from getting enough CPU.
(I’m also using SDI 10-bit SD for now, but would like for the board to be good for at least basic HD (not 2K).)