[Jason Roberts]

Jason and Aaron, thank you both so much for the further thoughts – I truly appreciate it. At this point, it may become a moot point – I’m on the verge of walking off this mess – I think I’m the only one doing actual pre-production work. The AD still isn’t doing anything, and the DP has got the director hot on shooting anamorphic again, and now are wondering if they should use a 5D Mark III with the custom firmware and an Anamorphot adapter. I have explained that the EF lenses will not take the Anamorphot adapter, that you need lenses with smaller than a 50mm thread, and that you can kiss shooting at night without lights geared for shooting in the night goodbye, and that the 1.33 Anamorphot doesn’t accomplish quite as much of the anamorphic “look” as the 2x, but that would require the GH4 in 4:3 mode, which really won’t work because of a maximum aperture of 4.0 . . . and the DP was looking at me like I had multiple heads. I proposed sticking with the a7s and doing the matting in post. Is is a cheat? Yes. I also pointed out that Roger Deakins never shoots anamorphic and basically uses only three lenses (32, 40 & 50 – that’s by his own admission on his discussion board) shooting on standard 35mm film, and so all his films are matted in post (and most would consider him the best cinematographer working today (Yeah, you could make arguments for several others, but Deakins is the man as far as I’m concerned – should have won 5 Oscars just for the look of “No Country for Old Men” alone). And the DP – I kid you not – actually said, “Who’s Roger Deakins?” I almost cried.

I just looked at the post above and realized it’s a rant. If I’m ranting this much and we are still three weeks out from rehearsals, I’m feeling like maybe walking would be the best. I have no investment in any of this except time and aggravation.

[Jason Roberts]

Aaron – again, thank you so much for the thoughts! I have been going back and forth between the 5960 and the new i7-6660 (Skylake), the reason being that Haswell is now a two year old architecture, and looking at Anadtech Bench, there is a huge advantage in rendering times even with the 2 cores and four threads – performance in heavily threaded workloads shows about a 10% advantage, and it seems the cost difference between 2011 and 1151 (processor and motherboard wise) might not justify the 10%. And the pricing on LGA 2017 is just . . . disgusting, so waiting for a Skylake Hexacore may not be worth it. I do think a AIO liquid cooler and overclocking the hell out of whatever processor I buy will be in order, however.

As for the array, I’m wondering if a RAID 10 might be the way to go to get both the read and write gains while having the fault tolerance (along with some random external 8TB as a backup) – it seems most of the good hardware RAID cards support 10, and if I’m buying 8 drives the 4x increase in transfer speeds will be nice just for the sake of the footage dump.

So let’s assume I build ye old Premiere Pro machine. Can anyone speak to the Ninja Assassin’s metadata tagging capabilities? It would be nice if they were nice and robust since that would cut down on time spent in Bridge or Prelude for metadata. Jus a thought, as always – I think too much 🙂

[Jason Roberts]

Aaron, thank you much for your input. I find myself in the unenviable position of having to play peacemaker on so many things because the assistant director isn’t doing her job, the producer isn’t doing her job, and the director – well, let’s just say he thinks he’s the cock of the walk. So I get to be combination peacemaker and bad guy. You should see the email string in which I explained that shooting on anamorphic lenses (original plan was a GH4 and Lomos – yes, Lomos, which wouldn’t work in most of our locations at all, the DP thinks he’s a combination Robert Surtees and Gordon Willis) was going to put us way over budget – it’s hilarious in a schadenfreude kind of way.

Fortunately, the DP and the equipment come separate, so I’m not worried about that. The editor – well, he just hates FCPX and wants to find an alternative. Said I would ask what others were doing, so I ask. I wish we could afford a trash can and array – sadly, it got out that I’d built a Hackintosh or two in my time, and now they think I can build a Hackintosh for something like this is we need to, despite my protests that Hackintoshes aren’t stable enough for work like this.

And yes, I would rather rent a RED and rent the editing station for this project, but they won’t hear of it – they want to own the editing station and rent the shooting equipment. But hey, the producer didn’t even form an LLC for this, it’s all going on his credit card – how old school of him. Suffice to say, I’m pretty damn close to walking myself.

[Jason Roberts]

Alex –

Thanks, I was thinking that downscaling the HDV would be the way I’d have to go, sadly.

Now, as for some of the ProRes setting when converting everything to the same codec:

1. Converting the HDV to ProRes SD – Using 720 x 480 as the size, what’s the proper pixel ratio to use, since the HDV is anamorphic? Options are such things as Default for Size, Square, NTSC / DV 601, NTSC / DV 601 (16:9), etc. Same question for converting the anamorphic DV to ProRes. When looking at the clips in Final Cut, only the DV clips have a check in the Anamorphic field, whereas the HDV does not (I assume this is normal).

2. Interlacing: the HDV is upper dominance, the DV is lower dominance. Should I use ProRes for Progressive as the setting for all? Or use ProRes for Interlaced and change the DV to upper dominance, and use ProRes for Interlaced on the HDV and keep the upper dominance?

Can’t you tell I am SO happy to be doing all this? And thanks to any and all for all help in advance.

[Jason Roberts]

I know that the GSD needs to be written into the code. One would think, or hope, that when the coding was done on FCS 3, the developers at Apple would say to themselves, “You know, with Snow Leopard coming out around the same time we release this, maybe we should write it into the code”. From what I’ve read so far, that isn’t the case. So it looks like I”ll be waiting for FCS 3.5 (or whatever they decide to term it) to get the GPU acceleration benefits.

And as for PCI Express 2.0 – so far as I know, it’s the same physical slot with no major changes in I/O, just more bandwith: a 2.0 graphics card should work and work well as long as it’s an x16 slot, it just won’t be at full possible bandwith. That’s how it’s worked on the PC side of things since the switchover, anyway.

[Jason Roberts]

Every time I’m ready to pull the trigger on some kind of purchase, it seems like I find one more bit of info . . .

I was reading through the Caldigit RAID manual, and saw the instructions on how to install your system drive in the 2nd optical bay so you could use 4 internal drives for the RAID. I’m assuming the bandwith I could get from using the 4 drives internally will be roughly as good as using them externally in the ProAvio 4MS box?

[Jason Roberts]

Now that I’ve finished making sure I understand three words, shoving specs up my @$$ and doing some final shopping & research, it looks like I’ll be doing with the ProAvio 4MS box and the CalDigit RAID card. None of the Areca or Atto cards seem to be able to beat the price/performance ratio of the Caldigit RAID.

I prefer turning to experienced users as opposed to salesman, figuring I can get a better ground-level feeling for what’s out there and what’s true in real life as opposed to paper-specs. I know posts on this subject are somewhat common, and searching through here found the most recent info being a year old. I asked because I wanted to see if anything major had changed since that time that I couldn’t dig up in my reading. I apologize for taxing the patience of more experienced users than myself with these questions and not making clear what I had / had not already read.

[Jason Roberts]

David – thank you for the reply. I’ve been doing considerable reading on hardware vs. software RAID, naturally, and am aware of the differences beyond price. I apologize for not indicating I was already aware of that issue. I was thinking of the horror stories of cards like RocketRAID 3522 or 2522, both hardware RAID cards, when referring to cheaper cards.

My concern is this: I spent the last year editing DV, HDV, AVCHD, and ProRes on internal, non-RAIDED SATA drives. The biggest pain in the butt was render times (esp. with Color) but if I’m not mistaken, that’s a product of CPU bottleneck on my configuration, isn’t it? I am hoping that GPU acceleration in Snow Leopard and FCS 3 will address some of that (taking into account I’ll need to replace my X1900 with something current, of course).

OWC advertises the Mercury Pro Qx2 as a hardware RAID solution over eSATA, with a 237MB transfer rate on the 6TB array (using the exact drives I’ve inherited) in QuickBench. Granted, OWC doesn’t tell you what RAID card they were using or what the drive config was (I would assume RAID 0).

The ProAvio lists the 4ML’s bandwith as 217MBS over mini-SAS in RAID 0, CalDigit claims up to 330 mbs on 4-drive RAID 0 arrays with the CalDigit RAID.

Can you see where I am concerned? I understand not to “cheap out” – what I am concerned about is value, as I’m sure we all are. If the CalDigit RAID makes the most sense given what I’m going to set up and edit, if it will really make that much of a difference in my workflow and time, then I will spend the money on it. But if it’s going to be “overkill” for my source material and not address some of my main issues (rendering time), then why shouldn’t I use the Mercury Pro?

[Jason Roberts]

Kent – I assume you’re referring to the “Drobo Pro” unit? It looks interesting, though this idea of BeyondRaid is a little odd. But it seems to have gotten good reviews. Just wish the bare enclosure was a bit cheaper than $1200 . . . oh well, you get what you pay for.

Shane -I have also seen the inconsistency in Time Machine / FCP conflicts. I hate it when problems arise without consistency or clear cause . . .