[Chris Kenny]

[Walter Soyka] “Are you really arguing that since you can’t attain unreasonably fast speeds that no system offers, you shouldn’t bother attaining the reasonably fast ones that other competing systems can?”

What I’m arguing is that for any set of specifications, in this business you can come up with a use case that requires even more. Apple has never targeted the very highest reaches of this market. To understand what Apple will likely do, one should look at common industry practice, not evaluate whether the proposed system will meet any obscure high-end requirement someone wants to name.

[Walter Soyka] “Maybe most of your projects don’t need that — but nearly all of mine do. Sending multi-channel 16-bit 3D renders to compositing is an everyday workflow in my corner of the industry.”

And you’re doing this with external media how? By moving 8+ spindle SAS RAID enclosures around? And you do so much of it that outputting a little more slowly would be a major problem? Again, I’m not saying nobody does this stuff, but — and this isn’t meant as an attack on you, because I have no idea what your particular situation is — there’s a tendency in this industry to play this sort of game where, instead of looking at what’s actually, realistically required to perform certain tasks, and making a reasonable cost/benefit analysis, people say things that amount to “Oh, well, I’d use X equipment to do Y task, and if you’re advocating doing Y with something less than X, you’re not as serious as I am”.

The truth is, Moore’s Law has brought us to the point where there are seriously diminishing returns on high-end specialty computing gear, and the higher you get the more they diminish.

[Walter Soyka] “Further, you are not actually saving any space, noise, or cost by skipping PCIe slots when you are then adding in outboard Thunderbolt video I/O and storage. You’re just moving it around, and sacrificing speed for flexibility.”

That’s potentially true of a fully-loaded system, but slots and drive bays take up space whether they’re full or not, and take up the same space regardless of the size of what’s in them. An external UltraStudio Mini Monitor adds a lot less volume to a machine than a slot that can potentially accommodate a full-length PCIe card does.

Everything keeps getting faster. At some point, it gets so fast that it actually makes quite a lot of sense to start sacrificing some speed for additional flexibility. In the consumer market, it’s exactly this phenomenon that has lead so many people to laptops and now to tablets. In our market, “What slot am I going to install a RedRocket card in?” will someday seem as pointless as “What slot am I going to install this SoundBlaster card in?” does in the modern consumer computing world.

—
Digital Workflow/Colorist, Nice Dissolve.

You should follow me on Twitter here. Or read our blog.

[Chris Kenny]

[Frank Gothmann] “Oh, those pesky screens. You know you can simply switch them off in the Bios if you don’t care about the fact they give you useful info, right?”

If there are such options in this case, they’re not obvious. I’ve looked.

[Frank Gothmann] “Well, simply assign fixed drive letters to your external media. All paths are stable then.”

We’ve just been mapping paths, which also works, but it’s an extra 16 clicks or something every time we hook up a new drive for the first time, which is usually several times a day.

[Frank Gothmann] “In the same way as you’d have to install Perian or other tools to get OSX to pick up certain media for previews, same applies to Windows if you want previews for certain Quicktime movies in explorer. Calculated folder sizes: again, just install some freeware. Classic Shell works just fine and it’s free.”

I’m aware that some of this can be fixed with third-party software (though my searches for something to solve the folder size calculation issue turned up mostly complaints about the fact that what seems like it used to be the standard solution didn’t work with Windows 7), but installing half a dozen third-party system enhancements of a production system doesn’t seem like a brilliant idea.

[Frank Gothmann] “Let’s see, motherboard comes with ONE driver disc and is a one click install for chipset, Lan, audio, etc. Basically everything that’s on the motherboard. Then there is the graphics card. And then…. hmmm. That’s it I guess. Install time: 10 minutes.”

When you say “one click install”, at least in the case of this ASUS board, what that means is you click one thing that launches about a dozen different installers, takes easily 20 minutes, doesn’t always work perfectly, and if you’re not careful will install some demoware you didn’t want. I know Windows fans tend to be oblivious to such things, but the whole user experience is fairly atrocious.

[Frank Gothmann] “Yep, full system image restore over network… 15 minutes.”

Yes, if you keep images available like that, which requires a fair bit of research to figure out how to set up, and probably isn’t worth the time for specialty machines where you’ve only got one with that configuration. It’s also not clear to me that it’s legal absent a corporate volume licensing agreement.

[Frank Gothmann] “Or get Episode, or Promedia Carbon, or the free FFmpeg.”

Still requires you to output and transcode, generally more than doubling output time and, of course, requiring you to have a huge amount of disk space available to output a feature-length project.

—
Digital Workflow/Colorist, Nice Dissolve.

You should follow me on Twitter here. Or read our blog.

[Chris Kenny]

[Bill Davis] “I might be totally wrong about this. But I think Reel ID that seems to trivial to handle at first – in the context of a database system – isn’t really so trivial at all. Unless you work with only ONE kind of file generation where ALL your reel names are very consistent.”

I think this is throwing the baby out with the bathwater. Yes, reel IDs can sometimes conflict if they’re generated by different processes, or if they’re generated by file-based cameras that have their reel counters reset, or whatever. But usually they don’t conflict, and they often end up being a very import piece of metadata in a world where file names can be changed to easily, and where offline/online workflows are still quite common.

As I said at the end of my last post, every single frame should really get a globally unique identifier — anything short of that is a compromise, including the use of reel IDs. But in practice, reel IDs work pretty well most of the time.

In the video world, footage is fundamentally independent of files. The same footage can exist in multiple files in different formats or of different durations, as a consequence of transcoding, trimming, etc. Given that timecode wraps around every 24 hours, there has to be some additional identifier that can be used to figure out what footage in this set of offline files over here corresponds to what footage in the other set of online files over there. Reel names are the standard approach to this, and Apple seems to be backing off of supporting them without having presented a comprehensive alternative.

—
Digital Workflow/Colorist, Nice Dissolve.

You should follow me on Twitter here. Or read our blog.

[Chris Kenny]

[Walter Soyka] “An uncompressed 1920×1080 16bpc RGBA frame requires 15.8 MB; at 29.97 fps, that’s 474 MB/s for a single stream. Given that compositing by definition is never single-stream, I’m sure you can imagine that there are is a significant class of users that do benefit from faster storage or more than 6 spindles in a RAID set.”

Look, you can play the multiplication game all day long. What if I need five streams of 4K at 60 frames a second? Hey, that’s 22 GB/s a second, good luck finding any sort of RAID controller for that.

Most projects don’t need that. They’re not even processed or finished in 16bpc RGBA, and if they are, it’s not the end of the world if you have to render the handful of shots that actually require compositing (i.e. multiple streams) — certainly most users won’t spend thousands of extra dollars on storage to avoid it. Our experience is that indie features are overwhelmingly either 10 bit DPX film scans, or Red/Epic/Alexa footage. 700 MB/s is enough for multiple streams of any of those. I’m not saying nobody needs anything more, I’m just saying that at that point you’re talking about a tiny fraction of the already tiny fraction of computer users who do pro video.

Apple has made some fairly powerful Macs over the years, but fundamentally the Mac is a personal computer. With the relentless march of Moore’s Law, full-sized towers are someday going to be relegated to the kind of niche that refrigerator-sized computers currently occupy (or maybe the niche that was once occupied by those mini-fridge sized “desk-side” SGI machines). Apple has never been interested in niches quite that obscure, so at some point before that happens, they’re going to stop making a full-sized tower.

[Walter Soyka] “Why should the Mac Pro exclude any workstation user by design?”

Because doing so will allow Apple to make it smaller, quieter, less expensive. In other words, a better system for the vast majority of pro customers who don’t (with the existence of Thunderbolt) need quite that much internal expansion. I know I’d find a significantly smaller Mac Pro very convenient for use on location, for instance.

—
Digital Workflow/Colorist, Nice Dissolve.

You should follow me on Twitter here. Or read our blog.

[Chris Kenny]

[Bill Davis] “How many digits should it have. Numeric only? Alphanumeric? Do you allow non-standard keyboard characters and/or wildcards?”

It should allow anything, up to at least 255 characters.

[Bill Davis] “Many here seem to be really interested in seeing a REEL ID field – and it seems to me that what they want is to be able to do the same thing they’re doing currently – so what exactly is that?”

Right now we’re working mostly with Red/Alexa projects, and moving them from NLE software to Resolve. Resolve is heavily reliant on reels, and can read them out of QuickTime files. With projects from FCP 7, we manually assign reels to any elements like VFX where they’re not already present. Often these are descriptive and/or based on file names, so if you import an EDL/XML and the file doesn’t re-link, you can tell from the reel name what file it is.

Now, one could make an argument that in file-based workflows this shouldn’t be necessary — file name + timecode is sufficient to uniquely identify any frame in the project. However, this makes me very nervous. Most of the projects we work on are edited from offline footage, and though it’s a really bad idea, clients sometimes rename those offline video clips on disk. This means if they don’t have embedded reel names that the NLE can read and export, there is absolutely no way to automatically relink them to online media. It’s not hard to imagine this occurring with FCP X projects as it’s more widely used.

So the issue here for me isn’t that there’s no good workflow if FCP X doesn’t read reels out of QuickTime files, it’s that there are more bad workflows, and it won’t be obvious to our clients that they’re bad until it’s too late. This could all be resolved by FCP X simply reading reel metadata into the ‘reel’ field whenever it’s present in a source media file that FCP X can natively import.

Of course in a really ideal world, timecode would be expanded to include day/month/year, plus a randomly generated universally unique frame ID just in case a reset clock would otherwise create duplicated timecode. Plus every file-based camera should also write a universally unique camera ID. It’s silly that in 2012 we still don’t have a reliably way to uniquely identify frames on a global basis.

—
Digital Workflow/Colorist, Nice Dissolve.

You should follow me on Twitter here. Or read our blog.

[Chris Kenny]

[Bernard Newnham] “In what way was Windows annoying?”

There are just a ton of annoying little things with Windows:

– It takes longer to re-install, if you have to.

– If you build your own box you have to deal with installing lots of drivers.

– It takes forever to boot if you have a motherboard with lots of on-board features or many PCIe cards installed (since some of this stuff insists on showing screens at boot that take a few seconds to go away).

– There are updates. Always updates. So many updates. It seems like if you walk away for five minutes, there are new updates that the system wants you to reboot to install.

– Explorer can’t show calculated folder sizes in a column, preview QuickTime movies, or do a bunch of other things one takes for granted on a Mac. QuickTime in general sucks on Windows, to the point where just opening a QuickTime movie, even on an extremely high-end system, takes much, much longer than you’d think it possibly could.

– You can’t, of course, jump back and forth between FCP and Resolve on the same system.

– Virtually every drive we get as a post house is Mac formatted. We use a utility called MacDrive in Windows, and it’s pretty good, but sometimes if drives are incorrectly unmounted it won’t mount them until they’ve been mounted by a real Mac.

– Windows uses drive letters rather than labels in paths, and drive letters can change every time drives are remounted. This means nothing has a stable path (unless you manually assign a mount point, which takes way too many clicks). Media in projects will therefore need to be re-linked all the time, etc.

– We have a bunch of in-house custom workflow scripts. Windows lacks the standard Unix-style command line utilities and uses rather than / as a path delimiter, requiring me to maintain two versions of these scripts for Mac/Windows.

– While the Windows version of QuickTime decodes ProRes, it can’t encode it, so if we need ProRes outputs we either need to render to uncompressed QT and convert, or simply move the whole project to a Mac to render.

Honestly, when building this system I figured “Meh, we’re using it to run one app (Resolve), how much can Windows really get in the way?” In practice it turns out it gets in the way quite a bit.

As far as going the Hackintosh route:

– Finding drivers that work with everything is pretty much trial and error, searching through Hackintosh-related forums and downloading different things. Expect kernel panics along the way.

– You can’t safely install system updates — they might break compatibility or not work with the crazy hacked up drivers you’re using.

– I encountered odd performance problems I’ve never seen on real Macs — jerky mouse movement, etc.

– Getting two GPUs working on a Hackintosh is non-trivial, to the point where I just gave up.

Put it this way — had we just bought a Mac Pro, our most recent grading system would have been up and running in two hours. Between my mostly unproductive Hackintosh experiments, screwing with Windows, and swapping various cards between PCIe slots, getting a usable grading environment on this system took more like two weeks, and months later there are still a lot of unresolved (and mostly unresolvable) annoyances.

—
Digital Workflow/Colorist, Nice Dissolve.

You should follow me on Twitter here. Or read our blog.

[Chris Kenny]

[Walter Soyka] “GPU, Red Rocket, RAID card — oops, three cards, two slots. Two GPUs, Red Rocket — oops, three cards, two slots. Two GPUs, RAID card — oops, three cards, two slots. Three GPUS — oops, three cards, two slots.

Next-generation Thunderbolt could help there on the storage side (if the controllers and peripherals would be ready in time), as it wouldn’t be bandwidth-limited like current Thunderbolt/4x PCIe RAID controllers.”

Thunderbolt seems to be able to achieve real-world speeds of about 700 MB/s for external storage, and for all we know that’s limited by the storage controller in the R6 and it could go higher — Thunderbolt’s theoretical maximum bandwidth is a full 1250 MB/s and its design indicates it should have fairly low overhead. Vanishingly few users, even in pro video, actually need more than 700 MB/s.

Resolve (probably the most demanding widely used app) as of version 9 has much-improved single-GPU performance (to the point where there are now single-GPU systems in the official configuration guide), and of course OS X now supports much faster graphics cards than it did a couple of years ago. Filling internal slots with a GTX 580 and a RedRocket and using Thunderbolt for external RAID and video I/O would produce a Resolve system capable of handling the vast majority of projects with ease.

[Walter Soyka] “A 2013 Mac Pro with 1 or 2 PCIe slots and any number of first-generation Thunderbolt would be a disappointment, though, because there would be bottlenecks built into the system.”

The Mac Pro is already a system built for maybe 5% of Mac users. If you want to then talk about people for whom two PCIe slots + four Thunderbolt ports would be insufficient, I suspect you’re now talking about 5% of 5%. I believe Apple is interested in its pro customers, but I think their focus is on what we could perhaps call ‘mainstream pros’, not the most obscure upper reaches of the market.

—
Digital Workflow/Colorist, Nice Dissolve.

You should follow me on Twitter here. Or read our blog.

[Chris Kenny]

[Walter Soyka]
But that said — has anyone here pick their NLE in 2012 based on price?”

I’m not sure “here” is the right place to ask. The primary strategic impact of price is probably to sway which NLE new users choose.

It may also somewhat influence how many seats of each NLE multi-seat facilities that use multiple NLEs install. If something is cheap, you tend to just throw it in every suite to gain flexibility; if it’s expensive you’re a bit more selective.

—
Digital Workflow/Colorist, Nice Dissolve.

You should follow me on Twitter here. Or read our blog.

[Chris Kenny]

[Craig Seeman] “The lack of additional internal PCIe slots and HDD bays will be controversial.”

I think you’re right that we’ll see a redesigned enclosure with less internal expansion. If it’s really four Thunderbolt ports and two and PCIe slots, it’ll be fine. It’ll annoy some people, but it’ll fundamentally do everything a “pro” Mac needs to do.

If they go with only one PCIe slot and two Thunderbolt ports, however, that might be a little dicey. And then there’s the nightmare scenario, which is one half length slot, like the old G4 cube. But I’m pretty sure Apple’s smart enough to recognize that a major #1 attraction of the Mac Pro is the fact that you can stick a giant graphics card in it.

We ended up building a Windows machine for Resolve as a consequence of the lack of a Mac Pro update this year, but Windows is pretty damn annoying (as is working out the issues surrounding a Hackintosh), so we’ll immediately jump on anything Apple ships that we can plausibly use.

—
Digital Workflow/Colorist, Nice Dissolve.

You should follow me on Twitter here. Or read our blog.

[Chris Kenny]

[Jeremy Garchow] “I disagree. It’s obvious that there’s not one NLE/Camera/Company/etc that can and will standardize metadata in the video world. So why should we keep trying to squeeze everything in to limited parameters? Why not just accept them all? “

You keep arguing as if my position is that FCP X should refuse to read camera-specific metadata or otherwise support camera-specific formats. That’s not my position at all. Rather, my position is that in addition to supporting camera-specific formats, it should also be able to read standard metadata out of standard container formats — or at least out of QuickTime files, given that this is Apple’s own file format.

[Jeremy Garchow] “Why not make better use of the systems that are currently in place instead of trying to reinvent the wheel when passing data to all other systems?”

One of the systems currently in place — and already used by camera systems Alexa — that that reel metadata can be embedded in a standard location in QuickTime files.

[Jeremy Garchow] “FCPX allows user assignable reels, which are crucial to his organization system and archive. “

User-assgnable reels are useful, but they’re more useful if they can be shared across apps — which they mostly can’t be with FCP X, because unlike FCP 7, when you change reel data in FCP X, it doesn’t change it in the file, meaning that another app reading that file (like, say, Resolve) has no way of associating the reel name in the FCPXML file with that media file.

—
Digital Workflow/Colorist, Nice Dissolve.

You should follow me on Twitter here. Or read our blog.