Autodesk announced the release of Smoke on Mac 2010 today and it's a big move for Autodesk, as traditionally their system products have been bundled software/hardware packages. This change makes Smoke available to a much larger number of users and facilities around the world, hopefully strengthening the product and leading to even greater growth and development efforts in Montreal. In a multi-part series at fxguide, we'll have an in-depth look at the software and why it is good news for the industry.
If you want to try Smoke on Mac out for yourself, Autodesk is offering a downloadable 30 day trial of the software. And if you want to really learn how the software works so you can get a job, join our fxphd.com training site after January 1st where we'll have the software available over the VPN for the new term. Smoke artist Brian Higgins, who taught Smoke masterclasses in Asia for Autodesk, will be leading a project-based introductory course which builds upon the training materials you can find for free on the Autodesk Area site.
In part one of our series of articles on the new offering, we have an overview on the new software and take a look at the hardware requirements and price considerations of Smoke on Mac.
Autodesk's finishing product Smoke began life as Fire almost a decade ago. This was on the cusp of a transition to computer based editing systems from the traditional linear or non-linear disk based systems. In the past, the product as been sold as a combination of hardware and software -- if you wanted to own smoke you had to buy both together. Recently, there have been two versions of the software: Smoke and Smoke Advanced. The main difference between the offerings has been the inclusion of Batch, a procedural processing tree similar to the workflow in Nuke, Fusion, or Shake. Smoke on Linux's previous list price was around $45,000 and Smoke Advanced around $90,000.
With the Smoke on Mac release, Autodesk has freed the software from the hardware, at least on OS X Snow Leopard (the Linux offerings are still bundled and are not sold standalone). For $14,995 US (prices vary worldwide), one can purchase a software license to run Smoke on Apple hardware -- hardware that many facilities already own. This change is significant as it signals a change in thoughts up in Montreal regarding the systems products. A year ago, Autodesk announced their software product Flare which is effectively the batch subset of Flame. That software is still quite pricey, at about $45,000 per license -- and to purchase a license you have to own flame.
While that was a start in the right direction, the high price tag and $190,000 giant flame dongle was still far too high a price to be paid. Considering the current business climate as well as the continually increasing capabilities of software packages such as Nuke, Fusion, and even After Effects, this new "low price" wasn't ground breaking enough for the market.
Smoke on Mac breaks this ownership and hardware barrier and offers an attractive price point that allows a vastly larger market access to the software as well as the ability to pay off that investment. In a series of articles, we'll have an overview of what smoke is, breakdown pricing (and why it is a pricing "sweet spot"), discuss its benefits, and examine some of its shortcomings in comparison to other software only products.
Before going into smoke in more detail, a bit about my background. I spent over a decade at Avenue Edit in Chicago working on national tv commercials and also some film and broadcast work. I've worked with smoke since the early days on version 1.0 beta when it was born as Fire, but largely concentrated on working in flame. In tandem with the flame I also used a wide variety of other software. I always required a machine in the flame suite with Photoshop, Illustrator, After Effects, Shake, and other apps.
After leaving Avenue Edit and starting fxphd, a little over a year ago I began a collaboration with Hootenanny, a post boutique in Chicago started by my friends Jim Annerino and Liz Tate. I purchased a combination smoke advanced/flame system on Linux which Annerino uses to finish jobs. While my work history fits strongly in the discreet/Autodesk camp, for the last several years you'd be far more likely to find me working in Final Cut or After Effects, so I'm quite familiar with the strengths and weaknesses of the various applications in the creative field.
Smoke is an editing application with integrated compositing, grading and paint tools as well as VTR and file-based i/o. When many people talk about the software, they say its a great app for working fast with clients. The integrated toolset makes it much easier to work on a job as opposed to diving in and out of separate applications. Having all the tools in one application is also incredibly freeing when experimenting on a comp or graphic. Suppose you've got an edit where you need to work on a scene. You can quickly take a shot into the 3D compositing node Action to try an idea out, render, and drop it back in the timeline to view in context....it is quite freeing from a creative standpoint. And yes, one can certainly accomplish this workflow using Premiere or Final Cut Pro by going into Color or After Effects, but it really is much more effective to have those tools within the same app.
This doesn't mean that all the tools are "best in class". For some of the tools you'll more than likely find another one out there that is specialized does a better job at a specific single task. But the whole is definitely greater than the sum of the parts -- smoke does a great job of getting a majority of your work done quickly and efficiently. For those times when Smoke's toolset doesn't have the capability to do something, you can always export a clip, work in an external app such as After Effects or Nuke, and then bring that result back into Smoke for integration into the master edit. In fact, many facilities work this way -- they have numerous Flame, Nuke, or After Effects artists feeding the smoke finishing hub.
Smoke handles a wide variety of media formats, making it an ideal finishing station to put all the elements together at the end of a job. Just throw out some formats and smoke most likely provides some type of support: from QuickTime, P2 MXF, AVC-Intra, Photoshop, 3DS, FBX, image sequences, AAF, MXF, and EDL. While it handles these formats and more, it's also important to note that there are some significant limitations to the support, which we'll be covering later in our articles.
Even though Smoke on Mac is a software offering, there are some recommendations you should follow in order to make sure the software runs to its full potential:
- 8 core Mac Pro 2008 or later
- Snow Leopard OSX 10.6.2
- 8GB RAM
- Wacom Intuos Tablet
- NVIDIA® Quadro® FX 4800 or FX5600
- AJA Kona 3 Card
- Single display, capable of a resolution of at least 1920 x 1200 or 2560 x 1600
- Fast storage
Let's talk about a couple of items on the list and how they relate to Smoke.
Snow Leopard and Mac
Smoke on Linux has been a 64-bit application for several years now. This is due to the fact that is eats up RAM quickly when loading high resolution uncompressed imagery into memory. Speaking first hand having seen the transition from 32 to 64-bit Linux, the software became much more stable once more RAM could be accessed. On the OS X side, Autodesk's timing worked out great with the release of the 64-bit Snow Leopard. Autodesk was in on the early days of the 10.6 beta, as it was absolutely critical to porting the app from Linux to Mac. It wouldn't have made much sense to port the application any sooner without a full 64-bit kernel.
What about performance on the Mac platform? We'll be publishing some specific benchmarks in the future, as we've done tests in the past on various flame hardware systems. However, these were batch setups and without batch the tests need to be reworked a bit before publishing our results. The media storage subsystem also makes a big difference in performance, so it's tough to compare Apples to Apples -- or HP to Apples. That being sad, in our initial tests, Smoke on a high end MacPro is not as fast as the currently shipping high end Autodesk HP Linux workstations. That's not incredibly surprising as the HP hardware screams, but more details to come later.
A QuadroFX card is currently required on the Mac platform -- matching the family of cards which are used on the Linux platform. Why only NVIDIA at this point? Smoke (and Flame) make substantial use of OpenGL in the graphics card to render effects and provide fast interaction with imagery while working in the app. This requires data not only being read to the card for display, but also back out of the card to save the imagery to disk. When Autodesk first starting working with the cards in their systems products, they had to work very closely with NVIDIA in order to get them to work. This is because for most purposes graphics card's data paths are one way...you need to get imagery to display on a monitor at high frame rates -- but the flow ends there. When Autodesk started using them to render imagery, there was a bottleneck getting the data back off the card and the drivers needed to be improved. To make a long story short, Autodesk was quite helpful in working towards getting improved drivers from NVIDIA on Linux as well as on OS X.
While a QuadroFX card is an official requirement, Autodesk developer Sheila Santos admitted in our our fxguidetv episode about the initial announcement that some of the work they did in porting from Linux to OS X could actually lead to supporting more cards and more vendors. Not that it is their official goal to support other graphics cards, but that the possibility is there. I think it's a good idea to start with a limited set of "blessed" configurations as a first step due to the technical challenges of porting the software, but to fully gain traction I think they'll have to do more testing with other cards moving forward.
That being said, the software does actually run on other graphics cards including -- yes, it's true -- current MacBook Pro 17" systems. I've installed Smoke on Mac on systems with GeForce cards as well as the MacBook Pro 17" and it does run, certainly well enough if you want to install the free trial on a Mac system. To be clear, this isn't listed in the requirements or recommendations, so If you're gonna put together a system without these, buyer beware. Especially if you're investing $15,000. But I'm sure over time we'll hear from the web universe what works and what doesn't. Smoke certinaly isn't optimized for the dual CPU and low RAM capacity of a MBP 17".
One critical point is that the display needs to support at least at 1920x1200 so that the entire UI can be displayed. The other point is that if you are buying a system that you need to have running at the highest level -- go with the recommended hardware. Autodesk tests the software on systems that meet the suggestions, so if you're running on a different configuration there is no guarantee it will run.
AJA Kona 3
The AJA Kona 3 card is used for both videotape input and output as well as feeding a broadcast monitor which displays a realtime output from the edit timeline and effects modules. This hardware setup differs slightly from the configuration on the Linux platform, where Quadro SDI daughter cards are used to feed the broadcast monitor. The reason for this is that there is no Quadro FX daughter card available on the Mac platform, so Autodesk uses the AJA card for this purpose.
This does have an implication on functionality of the application in comparison to the Linux product: no realtime deliverables are available on Smoke on Mac. What are realtime deliverables? The QuadroFX card is used to do realtime format conversions -- say from HD to SD and/or from 24fps to interlaced 60i -- and you can record the SDI daughter card output directly to a VTR. This is a significant difference, but we actually don't use this functionality at Hootenanny and I know many others who don't use this functionality either.
Smoke on Mac - including the 30 day free demo - can actually run without an AJA Kona 3 card, but you won't be able input or output to tape or have a broadcast monitor output. The Smoke on Mac installer can, in fact, disable the use of an AJA board automatically during install if one is not installed.
Finally, you'll want fast storage and a lot of it to work with smoke. Smoke always internally processes footage as full resolution uncompressed RGB imagery, so at HD and higher resolution the storage requirements can quickly add up. How you use Smoke and how it manages media can also determine the amount of storage you need, so here's a bit of background into how the app works with files and imagery.
Smoke and Flame can work with imagery in one of two ways, via importing into a managed framestore or via soft import. For the first method, you define a folder as a "framestore" on your RAID array. Think of it as what it is -- simply folder where images are kept -- with the difference being that the application, not the OS, handles moving media in and out of the folder. This is why the framestore is considered "managed"...it is managed by the application. In fact, it's quite important that you don't delete any imagery within this folder from the Finder.
I've read some posts in forums online which talk about forcing you to buy into strict hardware, including proprietary disk arrays and using a proprietary stone file system. This has not been the case for several years now, so here's a bit of a primer to some of the terminology related to media storage. First, some terms (from Autodesk documentation):
- Stone and Wire: The software package that encompasses local management of media and the transferring of media between Visual Effects and Finishing workstation
- Stone Direct Disk Array: A direct-attached storage (DAS) from Autodesk. Fibre channel links and the 4-Gb adapters of the XR-series disk arrays provide high performance and throughput.
- Stone Filesystem (Stone FS): A proprietary filesystem that is configured on a Stone Direct array to store media managed by Stone and Wire.
- Standard Filesystem (Standard FS): Any UNIX/POSIX-compliant filesystem, such as XFS and CXFS, recognized and mounted by the host Linux® operating system. You can configure a standard filesystem on a direct attached storage (DAS) array, a SAN volume, or a NAS device. Visual Effects and Finishing applications can manage media located on a standard filesystem in a manner identical to a Stone filesystem.
So those are some terms...what does this mean as an end user? The key is the fairly recently introduced Standard Filesystem. In the past, you had to use a Stone array (sold by Autodesk) and Stone FS to manage your media and especially in the early days of Flame and Smoke those arrays were ridiculously overpriced.
Today, even on Linux systems, you can now use your own storage for the media used in smoke. On the OS X side, you have to use the Standard Filesystem. This means the array you might have for Final Cut can also serve as the managed media folder for Smoke. It's as simple as pointing to a folder when you install Smoke.
In fact, if you look inside the managed media folder with the default configuration, you'll actually see dpx files that you can open and view. You can't easily track what clip is what, but the point is that Smoke is actually saving the files internally as DPX files. In the deeper configuration files, you can actually select between saving your files as DPX, TIF, JPG, CIN, SGI®, or RAW (16-bit float files are always saved as RAW). I personally haven't tried JPG or the other formats, but I suppose you could work with compressed media all the time. One key difference is that file system fragmentation can become an issue with Standard FS and you may consider using defrag utilities to correct this (if you're willing to take the chance on corrupting clips should something go wrong during the process).
Now let's take a look at the two main ways of getting media into Smoke. With "normal" import, you select an image sequence or media file on the OS X file system and import it into a Clip Library. Smoke reads the media off the OS X filesystem and then creates a full resolution uncompressed RGB clip on your framestore. This method is useful for safety and speed reasons. For instance, if your image sequence is located on a network drive -- importing the media to your local framestore allows you to play back the clip in real time. Processing of effects is also simply faster as you'll be reading the file off the fast local RAID. Also, if the original media file is deleted from the networked or OS X file system, it is not a problem because you are actually using the full resolution clip from the framestore. The drawback is that media is duplicated and with uncompressed RGB files, disk space can quickly be filled.
With soft import, you also select an image sequence or media file on the OS X file system and import it into a Clip Library. In this case, however, only a link is created in the library without copying or converting the media -- it effectively is a symbolic link to the file on your file system. The benefit in working this way is that media is not duplicated by Smoke and if the media is located on your fast local RAID, you'll get real time performance for many file types. As discussed above, Smoke is already using image sequences in the managed media folder, so performance is not an issue.
However, you also need to be aware that if you move or delete the original media file you won't have access to it any more. Working with soft import is an ideal way of moving between applications such as Final Cut or After Effects. When you start a Final Cut project, just make sure your media is located on the RAID drive. When you export an XML from Final Cut and import into Smoke, you can simply reference this same media without duplication.
While this workflow is great, there are some limitations. QuickTime media clips work, but not all codecs are supported by Smoke for realtime performance. Optimizations have been done for Pro Res, Uncompressed 8-bit and 10 bit, DVCPRO, and AVC-Intra clips, so working with these flavors works quite well through HD resolution. Others don't work as well and you'll get frame stuttering or dropped frames.
With either of the import options, media metadata which you use in the app is tracked via what are called Libraries. At a basic level, you work a Library the same way you would work with a project bin and folders in Final Cut Pro, Premiere, or After Effects -- you see and load clips from the library into modules and the Smoke EditDesk. The difference is that when you delete an imported clip, you are actually also deleting the media. When deleting a soft-imported clip, you are merely deleting a pointer to the clip.
In addition to the lack of real-time deliverables, Smoke on Mac has a few other shortcomings compared with the Smoke on Linux offering. The following Linux features are not supported on the Mac version:
- Tape archiving and VTR archiving (file archiving is only method to archive)
- Broadcast monitor does not work when inputting media from VTR
- There is no export to QuickTime DNxHD
- No spell check in the Text Module
For the full, official listing of system requirements, visit the Autodesk requirements page.
Now that we've taken a look at the hardware and some of the reasons behind the hardware, let's talk pricing. If you look at the price of building a system from scratch, the software + hardware price of Smoke on Mac is comparable to Linux Smoke system. In several posts on our site and elsewhere on the web, readers have tried to make the point that because the combined price is the same, that that the new offering isn't all that groundbreaking. This is a fallacy. If you're building a system from scratch, yes, it's true that there isn't a significant price savings. However, Smoke on Mac software provides an incredibly affordable point of entry for boutiques and facilities who have already invested in hardware for a high end Final Cut or Premiere workstation. Even if you're not a hugely busy facility, as you'll see below, it's a fairly simple business proposition to pay off the software.
If you wanted to build a system from scratch, here are some ballpark numbers which would provide the foundation of a solid Smoke on Mac system:
- Smoke on Mac: $14,995
- MacPro: $5,000
- Display: $2,500
- QuadroFX 5600: $1,800
- AJA Kona3: $2,400
- LaCie 12big Rack Fibre Drives (4x4Gb): $5,999
- LaCie 12big RAID Controller: $2,800
- Atto Celerity FC-42ES Fibre card: $1000
This ends up costing around $36,500 to build a system from Scratch, compared with about $15,000 if you have already made an investment in required hardware. On top of this, you'll want to add $2,000 a year for software updates and support. The support I've gotten over the years from discreet and Autodesk has been fantastic. It is a level of support that I haven't gotten from other applications, which makes sense as it does add to the cost of the system. While down time has been minimal, when the system has had issues the Autodesk support folks have been top notch in getting the system up and running. Not shipping a spot and delivering it on time for a client would have a far greater cost than support -- and that's a chance I'm not personally willing to make.
The last item I'd consider a necessity for a Smoke on Mac system would be a base set of plugins -- and the stalwarts on the Autodesk systems would be GenArts Sapphire or Tinder from The Foundry. While there is currently no support for sparks on the Mac platform (due to the lack of a Sparks API from Autodesk), we expect it to be available in the future. Since no product has been announced, I'll budget $4,000 for the plugins for Smoke on Mac.
Here is a look at the rough pricing (US Market), along with numbers of how much one would have to spend to pay off your investment over 3 years (a standard period to expense equipment):
There will be those who complain that the price is still too high -- or feel that it should and will eventually come down in price to be closer to Final Cut. I don't buy into that. From the numbers above, it is quite easy to see that the software can be paid off with minimal effort for many markets. This is a targeted application and not for everyone. If you can't pay off the purchase price with the work you're doing, then quite simply...the application isn't for you.
As a current Smoke on Linux owner, I want a healthy and vibrant software development for Smoke moving into the future. Instead of being upset that I paid a significantly larger amount of money for Smoke about a year ago, I'm actually excited to see the potential user base grow. That being said, had I bought a Smoke system last month, I might feel a bit differently.
I'll also suggest that the market to which applications such as Smoke, Flame, Scratch, and even Shake and Nuke are targeted is quite limited in the grand scheme. Once you sell copies of software to your market, there's not many more artists or facilities to which you can pitch the app. I think this was one of the issues back in the Shake days when it was developed by Nothing Real. At some point, they had fairly saturated the market with Shake and needed to come up with another version of it to expand the market -- so they came up with Tremor. Externally, this had the appearance of distracting them from their core market, so instead of adding features that enhanced compositing they added features that didn't have as great a value. This may not have been the case, but it is still a perception. Once the product got to Apple, it became fairly obvious they weren't selling many copies as development languished and we saw very few updates.
What about pricing the software closer to Final Cut Studio? The argument being that his would increase the market segment Smoke could target and the cut in price would be more than made up by increased sales. I think that's also dangerous to the future of a product. As you expand the target of your software into more and more areas, development becomes less focused because your market isn't as clearly targeted. Toxik seemed to be a victim of this thinking. Initially, the product was targeted as a hard core compositing replacement for Shake. Over time, however, wider market features such as broadcast output were added instead of concentrating on a very focused segment. The software became a bit of jack of all trades and master of none. Focus on your core market and make a great app.
Finally, if pricing software at a low price is such a great idea, what is happening to Final Cut Studio? From any vantage point (except possibly a true Apple fanboy), it is clear that development of the product has stagnated over the last several years. Is this because Apple doesn't have enough income from the suite to support three vastly different application development environments? I can't help but wonder if Apple charged $2,000 or more for each of the suite biggies (Final Cut, Color, and Motion) that we would see much more solid development on the applications.
Contrast the Final Cut Studio development with a couple of other higher-priced app's development over the last couple of years. Avid seems to have hit their stride recently, offering several big improvements to their software including direct support of filesystem media, mutli-resolution, and stereoscopic editing. Smoke's development over the last several years has also reached a quick pace. Each year there has a been a major update with new features as well as minimal mid-year upgrades. From a stability standpoint, I also find Smoke to be way more stable than Final Cut. In fact, it's not even close with the recent release of Final Cut Pro riddled with bugs. This is certainly not to say that Smoke doesn't have bugs -- far from it. I just find them to be far less serious and unpredictable than the ones found in the studio.
Does smoke replace Final Cut Pro? No. After Effects? No. Nuke? No. The software serves a more specialized market -- high end finishing where creative tasks can be done quickly and efficiently. There are requirements for this market that require more specialization than they do for other markets, so it is best to focus on those things in the software. Opening the software up to too many vertical markets ends up pulling the development in too many directions instead of concentrating on what the software does best.
The bottom line? If the software is too expensive at this point in time, maybe it isn't for you.
In the next part of our Smoke on Mac series, we'll walk through the creative tools of the software and examine where they excel, where the promise doesn't fully live up to the reality, and why it's still a fantastic tool for the job.
We've been a free service since 1999 and now rely on the generous contributions of readers like you. If you'd like to help support our work, please join the hundreds of others and become an fxinsider member.