Flame vs Shake

[Pretorian]

I have never used Flame, but, as I heard, seems that the most diference is that Flame work real-time. So, as I heard, you don't need to render your work, as it is already rendered.

Again, I'm not sure on that, just have heard that. Does anyone know if this is true? Am I saying anything stupid?

I'm very curious on that too!



Thanks!

[Aneks]

Deke (beaker) and some others covered this topic very well a while back. I know its a cliche but at the end of the day it all comes down to the artist and using the right tool for the job:

These are two very different tools with very different user requirements. judging by your nick you know something about 3d. The comparison of shake to FFI is not like comparing max to maya or combustion to digital fusion. These tools were designed to do the same thing. Shake and flame were not. I can crap on all day about 3d cameras and editable bi-cubics but thats not the pont.

Keep in mind I have used both of the systems in a porfessional environment and think both of them are excellent.

FFI (flame,flint,inferno) are very heavily focused on client sessions, dedicated suites and the broadcast and TVC market.

shake is much more about teams of artist, backroom work and film visual effects.

It is not a fair comparison to say may and shake can do X and flame can do Y. In my experince the majority of elements composited in flame originated from maya also.

People don't buy flame so they can comp some shots they are working on. They buy a suite because it is an essential part of their post-production companies business model. They spend the HUGE ammount of money necessary ( in this region flame runs in excess of $500K and don't even ask about inferno !) because they have workflows built around a big finishing machine and because CLIENTS demmand and pay for it !

We are talking about a very specialised tool.

Shake is designed to be a compositing tool used by a team of artists in a post prodction facility (generally on film projects). You don't buy shake to do some motion graphics work. You buy shake to set up a vfx compositing department that can handle very complex integration tasks that need huge flexibility and lots of control. You buy shake because of its awesome integration of different formats and colour spaces and its increadibly open design and functionality. You buy shake because your ARTISTS/ or COMP Supervisor demands it and uses it everday !

Either system in combination with something like Maya can produce almost any effect you can think of. But the way in which you do this work, the kind of structure you have in place governs your choice.

[maxtomaya]

Awesome and right to the point.:thumbsup: Thanks Aneks. 500k!:eek: guess I won't see that in near future. Right now all of us here are obsessed with Shake and DF.


Thanks for the tip again:)

[Pretorian]

Aneks, is it true that Flame works in real-time? I mean that you don't need to render it?



Thanks!

[maxtomaya]

After digging in a little, I found out that it is the raw muscle power of flame. The power comes from the power of the hardware. It seems that the hardware itself is highly priced. It is Much more you pay than if you were to buy a high spec hardware. For example, in theory I could make my shake run much faster than the realtime Flame/Inferno for 500k. HOWEVER, unlike Flame, Shake CANNOT utilize more than certain processors (2 I think) even if I were to make such a system. Obviously, if Shake/DF were to take that restriction off, the difference is just that!

One of the guys at Apple told me this:
Shake was initialy made for Mac and is very old. During the development, the idea of implimenting a software architecture to support running on super computers (today we call those super servers:D) was out of budget that time. 8 years ago you would have to be millionair if Apple did that. However, today it is much cheaper. With 60K hardware, you can make things very much possible.

Like Aneks pointed out, the use is very different. In a production house where many people work in team, this power is emulated by different means. Thus we have more people to work on project and at the same time power to back it up intead of one artist on one machine whch would be a short time project which actually cut down the cost in the long run.

Thanks for all the info. It is amazing that all this time I never looked into the difference even though it was there for years. There is always something to learn from everything:)

[Aneks]

Hey,

matomaya : Thanks for your kind comments. A coupe of clarifications though. flame hardware is fast and can acess numerous procs. but for years it was the bandwidth throughput of the sgi graphics subsystems and the poenGL implementation which gave these machines an edge. Nowdays these are't such a factor. The graphics bandwith of GPU 's like those made by nvidia are now being used by sgi anyway.

A major factor in flames speed is its framestore and other dedicated hardware. Yes you could get shake to run fast with these extra bits for less than $500k......... nothing real did and it was called tremor. Unfortunately apple killed that along with windows support and a level playing field pricing approach.

Pretorian : Hell no. FFI garuntee real time 'playback' of certain resolutions in certain configurations. and can perform somefeatures in near realtime but real time compositing without rendering ..... forget about it ! Using burn and background rendering you can commit nodes to render in the background... kind of like 'clusters' in fusion and the net result is almost no rendering.

Last year I was working on an SD project on a HD flame built on an Octane2 with 4GB ram and beleive me there was still plenty of time to kill while it rendered my batch scripts.
Where I am working now the inferno they have is increadibly fast and suped up. On this last job, which was PAL sd, once the compositor got to really heavy comps, 20+ layers and heaps of action nodes, there was still a need to walk away while it rendered.

I don't know who you spoke to at apple, but I remeber a time when a little company called nothing real were the ones making shake. From what I heard and read, and maybe beaker knows more about this, it was always the aim of the shake dev team to make a product which existed on desktop/workstation hardware without the need for big machines. This is what made shake different from systems like cineion and illusion...... to which it owes a massive debt in terms of design and paradigm.

[Pretorian]

Hey both of you, thanks for the answers! :thumbsup:

[beaker]

After digging in a little, I found out that it is the raw muscle power of flame. The power comes from the power of the hardware. It seems that the hardware itself is highly priced. It is Much more you pay than if you were to buy a high spec hardware. For example, in theory I could make my shake run much faster than the realtime Flame/Inferno for 500k. HOWEVER, unlike Flame, Shake CANNOT utilize more than certain processors (2 I think) even if I were to make such a system. Obviously, if Shake/DF were to take that restriction off, the difference is just that!Ahh, maxtomaya, your making me want to tear my hair out, bad/wrong information everywhere.

Yes SGI hardware is high priced, but it doesn't really make it that much more expensive then any other turnkey system on windows. Avid DS, flint and smoke which all run on pc hardware cost over 100k-300k each. Shake can use more than 2 processors, there is no limitation of this anywhere.
One of the guys at Apple told me this:
Shake was initialy made for Mac and is very old. During the development, the idea of implimenting a software architecture to support running on super computers (today we call those super servers:D) was out of budget that time. 8 years ago you would have to be millionair if Apple did that. However, today it is much cheaper. With 60K hardware, you can make things very much possible.Whoever you talked to has no idea what they are talking about. Maybe they worked at the Apple Store or something. Shake did not even come to mac till 2 years ago. Shake is only 6 years old. Also it runs on the same exact machines flame runs on.

[solun]

I can tell the interface of flame is not good enough, i always get lost in flame. Shake is much better.

[beaker]

I can tell the interface of flame is not good enough, i always get lost in flame. Shake is much better.Yea, great post, simply because you don't know flame very well, say that shake's interface is better.

Please don't post such a one sided comment unless you have full experience in both packages. It makes you look silly.

[maxtomaya]

You are right beaker, it was the guy in the Apple store here in New York City. I wouldn't know much about Shake's past apart from what people tell me about. My bad. Sorry of the 8 years part in the post.

However, I do say this; the hardware itself is not worth that much if they were to lookfor different means. 500k for a real time rendering? May be a great deal until end of 2002, but nothing more than a single sided control based on software now. I have seen realtime rending of millions of hair strand dynamics done in realtime and that is in 3D with all calculations done. That also without official OpenGL support where certain things were done in software mode. NOTE that I am only comparing the actual processing power required to calculate the effect. I am not concerning with the data trasfer of the media which is a different issue.

Also, beaker, 32 bit OS systems cannot utilize more than certain processors unless you can make a cluster but cluster is not one system. This is because of the os architecture MS and some Unix derivatives created. Proprietory OS such as Solaris, AIX et.. are exceptions. There is also a Memory limit of 4Gig and this is why discreet decided to go with systems that doesn't have all these limits. You have to consider the fact that 64 bit PC (AMD came in a year ago and Itanium was there but still new) is new considering the fact that MIPS architecture has 64-bit working as early as 1999 at full range. And only Unix was running on it.

I hear that the Flame's UI is very different from Combustion and people who have used it never want to get away from the system. I would love to see it in action once (working on it would be fine too:))

[solun]

maxtomaya , great reply.

beaker, I'm, sorry for one sided comment. I am silly and

I just wonder why i have to exit the batch for output clips.

I just wonder why i have to have to seach the nodes the the batch, there are no nevigation windows and shotcut.

I just wonder why there are at least three places to set the image size in flame.

I just wonder why i have to seach the channel in channel editor when i make animation in flame. when I use the filter in the channel editor, it hanged.

I just wonder why i have to back to desktop to use wraper and paint.


I just wonder why I have to add soucre to move the matte against the front image.

I just wonder .......................................
.
.
.
.
.
.
.

[solun]

Anyhow, there are some unique great tools in flame such as 3Dkeyer and tracking tools.

[Aneks]

I get the feeling that you are experiencing the typical flame blues. You gotta look at where ffi came from and when they were built.

There are serious design limitations in the ffi architecture. If you really want to start a heated discussion you should go ober to fxguide throw some of your questions to that group. There are some total evangelists on that list.

I used to get very frustrated in flame but a lot of things you are describing. Source nodes, psuedo node based systems. You cant imagine my supprise when I first got onto inferno v4.x and found out it was not resolution independant !!

You can't ask a compositing system to be all things to all people. I have used shake enough to get through my days without having to use UNDO. But tell a digital fusion person this and they will stare at you like you're insane ! But your right for the money one pays for a discreet system sometimes the limtations are really hard to bear.

After a while I gave up looking for the perfect peice of software and just started trying to get as good as I can with the apps I know.

[solun]

Aneks, I agree with you.... totally

[Liamwhitehouse]

Maxtomaya,
I've been using Flame and Maya exclusively for about 5 years So I know your interest in this strange and wonderful software.

Much like yourself I was intregued by this "high end" software which costs more than a nice house.

PC's and MAC computers are primarily designed for the Consumer, SGI systems are not. They are specifically geared towards Broadcast artists and budgets, like sony cameras and filming equiptment.

Flame is the same, it is totally no frills, Broadcast Production software, There are no Dancing paperclips found here, it is as serious as it gets. In Brisbane Cutting Edge rents out their Flame Operators on system for $ 440 p/hour last I heard, for TV and Film work. So learning it thoroughly can get you into the Bling Bling.

As far as realtime is concerned, SGI octanes have awesome Open GL graphics boards which just show the final Composite image on the screen. If you press play though it might jitter at 1 frame per second, so you render the timeline to the framestore to see it play smoooth.

As for the Framestore it's simply a SCSII hard drive plugged into the back of limitless size. 20mb per second exchange rate so you can watch uncompressed frames at 24 per sec.

The interface is set up beautifully like a Robust tool. swipe bars allow quick access to info and options and the full screen is filled with Image at most times. Overall it makes software running on PC and MAC look like it was made for Uni Kids.


SHOULD YOU LEARN FLAME, most Definately Yes. It is the best Piece of software I have used bar none. Plus it runs on the OS (UNIX) that never dies on you.

You can learn flame with a minimum of an OCTANE MXE sgi system and it is taught at AFTRS in Sydney and Discreet locations worldwide. After you've looked at Flame check out Smoke as well.

[JoshKirk]

"Flame" is a much cooler name than "shake" if that counts for anything.
How different is AVID DS and flame? I got to work on an Avid DS once and I noticed how fast it was compared to desktop compositors I have used, mainly in the tracking area. Everything else was about what I was used to speed wise.

[thomaspecht]

liam:

20 mb per second for sure is not enough to play back uncompressed pal/ntsc.
a flame normally comes with one or two dual-port fibre-channel cards which push quite some more than those 20 mb/s ;)

i don't get what's so special about octane's v12 board. it's really well supported within the software, that's probably what makes the difference - other than that, accumulation buffer maybe? but i didn't notice it being used extensively inside flame anyway.

the interface is something to argue over but i guess as a full-time flame operator you're so married to your app that it will always be the most comfortable to you. even if it's gui design is from the stoneage, sometimes confusing, often limiting. important knobs well being hidden behind a swipe bar, etc etc.

now: since irix/mips roadmap seems to end in 2006 and that makes the tezro the last of the line of irix workstations, what will be the new platform for ffi? is linux/ibm intellistation up to it yet? or will they go prism and stay proprietary?

[Aneks]

Rumors abound about Linux Flame to follow on from the Linux flints now available. Certain folks have been talking about Apple's new commitment to Dual Core dual-procs and serious open GL cards (Quadro fx 4500), as a sign that a OsX port of discreet is now a certainty. Who knows ? A lot of people saw the screen shots of lustre running on a mac. Could this be so different !

Discreet are keeping any such development plans very secret so I guess that the only people who really know the anwers are bound by NDA's.

I read on one compositing web board that certain folks were testing SGI Prisms for playback and 'other' video related funtions and that they smoked anything out there. For that money they should.

I personally would love to see flame on the g6 for less than $200k !

[Lorecanth]

Just thought I'd chime in on our shake compositing box here. Well first off we have it hooked to a 15 TB xsan, dual 2 Gb fibre channel for an aggregate bandwidth of around 300 MB/s. The mac is a dual 2.5 with 8 GB of RAM. 2 Render nodes are also hooked into the xsan.

So several things that beat flame there. The framestore is a local collection of hard drives. You want a SAN like xsan? Prepare to shell out for a Stone Share $$$. Its silly the premiums that people pay for non totally realtime hardware. Inferno I can understand. But if you're going to make a client wait 3 minutes to see a 30 second render no matter what, you can get the same level of impressiosity (a term coined here at 3V ) from a highend mac system.

I've sat down at Flames, this system is faster, and not nearly as limited. It hasn't failed yet in wowing clients and it gets the job done.

[Kai01W]

Just thought I'd chime in on our shake compositing box here. Well first off we have it hooked to a 15 TB xsan, dual 2 Gb fibre channel for an aggregate bandwidth of around 300 MB/s. The mac is a dual 2.5 with 8 GB of RAM. 2 Render nodes are also hooked into the xsan.

So several things that beat flame there. The framestore is a local collection of hard drives. You want a SAN like xsan? Prepare to shell out for a Stone Share $$$. Its silly the premiums that people pay for non totally realtime hardware. Inferno I can understand. But if you're going to make a client wait 3 minutes to see a 30 second render no matter what, you can get the same level of impressiosity (a term coined here at 3V ) from a highend mac system.

I've sat down at Flames, this system is faster, and not nearly as limited. It hasn't failed yet in wowing clients and it gets the job done.

Our flame does ~330MB/s. Stones are not that much more expensive (if at all) than an xsan config that you just described (lacking the san featuress though) Shake on a mac still feels rather slow. The overall workflow is different, and in my humble opion much better in flame. But that might be a matter of taste.
What flames did you work on? Most people these days seam to measure their FFI experience on old hardware. You cannot compare an dual octane1 to a 4 CPU tezro...

-k

[Aneks]

Is it really worth having this sort of discussion. Re : shake/flame performance ?

In this country broadcast clients will not pay the sort of 'suite' premiums to sit infront of a shake box. period. The agencies and directors know all the brand names and love the feeling of knowing that they can sit on the comfy couch, play on the opps' PSP and have the cute girl from catering bring them coffee. They also love the feeling of paying $700-$900 an hour to have a 'name' inferno opp working on their job. Beleive me the post house loves finishing jobs here for that reason too ! Shake is hampered by a lack of I/O and limited audio tools. For this kind of work shake isn't an option.

I do vfx compositing on HD and film. I use shake and I sit in VFX with the other shakers and the 3d team. In this environment FFI is hampered by its lack of support for many formats and true resolution and bit depth flexibilty. I don't care how fast playback on your "Quad 1ghz 16GB of Ram Tezro v9.5 of Flame" is if I can't deal with zDepth passes and lots of other things we need to do here. I could not do client based sessions in this environment, nor would I want too. For this kind of work FFI just isn't a viable option.

Performance test and workflow aside, these are different tools. Anyone doing production knows this.

I have done, 'front of house compositing' on a mac, with a broadcast monitor and a RAID before. It was fine. But it wasn't a high pressure inferno session. These are different beasts. I can't turn around to my client and say "sorry I'll need to render this and re-import the file before I can paint that because the paint tools in this version of shake are buggy !" Nor will I expect a client to ask me to wait while I ensure all my colour corrections are concatenting properly before adding my film grain node. FFI can't do that.

There should be no Flame vs Shake discussions. Its like comparing apples and oranges !

[Lorecanth]

lol well again I'm going to to disagree. I think shake/fcp/motion vs flame/smoke is a legitimate discussion. shake/fcp/motion vs inferno is definatly definatly not. I've had clients walk because we didn't have an inferno, but then I've turned around and rented out our shake/fcp suite for 700 an hour without an operator.

It's all about the impressiosity, and while I admire the skills that are needed to successfully "handle" an in client session. I honestly believe flame/inferno operators don't give themselves enough credit. The hard part in such a session is just dealing with the client and making them feel like they've got the absolute best, while still using the tools to create something brilliant.

As far as what flames I did work on I got a couple evenings way back, a few years ago, what struck me even the was that the major issue was disk bandwidth.

In any case hey if you can make a viable business model running autodesks gear wonderful. I know NY and LA definatly don't have nearly the price pressure on high end content that vegas does. So whatever works.

[beaker]

Our flame does ~330MB/s. Stones are not that much more expensive (if at all) than an xsan config that you just described (lacking the san featuress though) Yes the are much more expensive. Xsan with a Xraid is 1/4th the price just because of pure component cost. Stones are usually Fiber Channel or scsi raids and Xraid are SATA drives. 1/4th the cost right there. Then the stone "tax" on top of it.

[Kai01W]

Yes the are much more expensive. Xsan with a Xraid is 1/4th the price just because of pure component cost. Stones are usually Fiber Channel or scsi raids and Xraid are SATA drives. 1/4th the cost right there. Then the stone "tax" on top of it.

Depends. If you want to have really high bandwiths you have to go with lots of disc, which make the xraid rather expensive cause they are bigger than the discs in a stone. So for your money you'll get much more GB in a xraid but not bandwith, if that is critical.
Plus you have to pay for the xsan, you need a controller mac/pc, etc.


-k

[beaker]

Well yea, the xraid only gets 200 MB/s, but you can pair up two, xraids(not sure what the speed increase on that is, probaby not 200%). Apple's fiber channel card is only $500.

[joconnell]

For starters there's nothing that flame can do that another compositor can't - in terms of the final output flame doesnt have anything mega or wow that makes the difference. What it does well is operate smoothly in front of a client as everyone mentioned - the way most people work in flame at this stage is actually really close to shake (from a compositing perspective) using the node system called batch. The workflow is no where near real time in most cases - you'll generally tweak aspects of a comp one by one (Ie in the keying operator / colour correction operator / sparks & filters operator) til the client is happy with it and move on to the next aspect - if you want you can pre process the result of each operator and move on to the next stage but most people I know leave everything until the end and process the entire comp chain in one go - in this regard it's no quicker than an after effects box or a shake box pulling all its footage from a raid.

In terms of cost, flame was the first turn key system to offer a lot of nice things like tracking, good keying, masking etc and this was back in the day when everything in post was bloody expensive - people invested in really nice rooms for the systems so clients could sit on an operator all day an tweak stuff and charged them wads o cash. The operators generally were invested in heavilly too with discreet training or the such and commanded a lot of money - effectively the whole thing was a huge investment and as a few people mentioned there was a certain kudos about doing a job in flame or being an operator. Discreet have held this up to a certain extent by always keeping the software quite forward thinking in terms of features but it's only the investment and reputation keeping the price there.

In terms of career I reckon it's a good move to learn flame. It's a tough one to get into as people generally dont want the down time for someone to learn it - flame ops have to repay their wages and machine costs so their suites are normally fairly tightly booked. What's good though is the profile it gets - you're sitting there with the client and director, they see you doing the work. With a shake or after effects compositor they dont always so you mightnt get the credit you deserve. as a flame op you'll also have to put up with more nonsense from clients and if they're the type of people that will walk out of a post house cos they dont have an inferno then you'll earn your keep dealing with primadonna cunts like that...

From a comping point they've nothing better than a big shake (unless you like the workflow) from a career point they put you in the spotlight a bit more.

[roto baggins]

i work at hydraulx and we use flame/inferno to do all of our comps. it's a great tool. we also have several shake boxes 3 on linux and 1 on a g5.

i use both, but i mainly finish everything in ffi. plus in flame/inferno you can build 3d object and map textures to it. it's a better package then shake. don't get me wrong, i like shake and combustion.

i think combustion is only good for clean up and roto.

just my 2 cents

[Alanbell]

I've used combustion for literally hundreds of effects shots at 2K and for my money it's good for much more than just clean up and roto. I'll take ten artists sitting infront of 10 combustion workstations over a single FFI system. (about the same cost)

But that's just my opinion.

Regards
Alan Bell

[Aneks]

i think combustion is only good for clean up and roto.

Until recently I had a pretty dismal view of combustion, but was offered a job as lead on a feature all in Combutsion 4 on Windows. Have to say that it does have its bugs but I was AMAZED at how well it performed. There was a lot of wire and rig-removal work, as well as some tough 3d matte shots and it performed very respectably.

Having just returned to normality (ie Shake now version 4) I am well disapointed with the bugs and shortcommings I am experiencing, as well as the less than complete 3d multiplane feature.

Doesn't pay to be too biased in this marketplace. If I was would have missed out on a great job opportunity !

[Alanbell]

I am seriously taking a look at DF5 as it seems to contain the best of both shake and C4.

I agree it pays to have an open mind.

Alan Bell

[beaker]

I am seriously taking a look at DF5 as it seems to contain the best of both shake and C4.As long as you don't have to do any scripting. As soon as scipting comes into play then Shake kicks them all in the jimmy since C* doesn't have any and DF's is still pretty sad.

[Aneks]

true enough Deke. But fusion 5 has much better scripting and all.

Everyone knows I love my shake and thats what my job description lists...... but its giving me serious irritation at the mo' . paint in v4 is so craptacular. apple support did give us a good response but still no real fix. Multiplane is very incompltete comapred to combustion,fusion ae and nuke. Cache node is useless ! Roto is better but nowhere as good as combustion.

What amazes me is that despite all these problems I still like is soo much !?!?

[jbradley]

lol well again I'm going to to disagree. I think shake/fcp/motion vs flame/smoke is a legitimate discussion. shake/fcp/motion vs inferno is definatly definatly not. I've had clients walk because we didn't have an inferno, but then I've turned around and rented out our shake/fcp suite for 700 an hour without an operator.

Couple thoughts here. First, there's zero difference between flame and inferno aside from the hardware - the software is identical. inferno has a higher throughput and can handle film resolution playback in realtime (neither package is a realtime compositing solution, unless doing simple operations). Take into account the SGI hardware and disk arrays make inferno 550k, whereas flame is 250k ... only because of hardware.

Trying playing back uncompressed 10 bit 4:4:4 footage in shake realtime through hardware (not flipbook or PFClip, which are RAM based systems that require caching).

I'm starting to love shake myself as well and am considering that it may be a nice addition to have a shake seat next to a flame seat. There isn't any comparison between the products though, because there is no setup you can get on any machine aside from an SGI or the latest Linux boxes running FFI that can handle the IO required for film and broadcast work. A Tezro box has 1.6 GB throughput internally for graphics - there's nothing else out there short of a supercomputer that can do this (definitely no PC or Mac). This is also why the Avid Nitris DS systems have their own hardware boxes, to really be able to handle that type of processing in an edit suite. So, you pay hundreds of thousands for this capability that one cannot get through a system like shake - which is a software solution, not a hardware/software solution.

Shake is becoming quite the powerhouse for VFX and film work but it will be a very long time - if ever - before it truly is in the same category as software/hardware solutions like FFI. There are an incredible amount of hardware capabilities that can be used on a flame system that shake just can't handle (such as specular shading with normal interpolation in 3d).

Still, there are many things that make shake an incredible addition to any VFX pipeline.

my $0.02.

[Aneks]

there is no setup you can get on any machine aside from an SGI or the latest Linux boxes running FFI that can handle the IO required for film and broadcast work. A Tezro box has 1.6 GB throughput internally for graphics - there's nothing else out there short of a supercomputer that can do this (definitely no PC or Mac)

Not to be nit picking, but this is just perpetuating myths.

The tezro figues are just meaningless statistics. Listen to the podcast on fxguide, the guy from dicreet admits that the performance on the new dual core linux flames is a 'nice improvement from the tezro' Not equal... faster. Therefore why would this same hardware run so much slower with other people software !!? Look at the ammount of 2k grading systems that are based on 'destop' platforms ! The exclusive days of SGI and FFI are long since over ! Sure post houses love to bill their clients for sessions but as for NEEDING one of these machines ......

As for the 'requiremnets of broadcast'. HD IO and playback is fine on mac with a disk array and a good IO card. I have used it to final TVC's in SD (PAL therefore high demmand than NTSC) many times.

You mention feature film, plenty of large production companies perform fine without large discreet boxes. Where I work now we happen to have two infernos, one smoke and one flame. But everyone here knows if its a film job (2k Log) then only the desktop tools are gonna be used !

[beaker]

A Tezro box has 1.6 GB throughput internally for graphics - there's nothing else out there short of a supercomputer that can do this (definitely no PC or Mac). This is also why the Avid Nitris DS systems have their own hardware boxes, to really be able to handle that type of processing in an edit suite.As Aneks said, this is bogus info someone fed you. Flame and Inferno are both already running on linux. Also an Avid Nitris DS is not their own hardware, it is a IBM or HP windows box with a DVS HD card in it (same family of card that Tremor, Cyborg Flint, Flame, Inferno, etc... all have been using). PCI express is equally as fast as the Sgi bus these days.

http://www.dvs.de/

Anyways, Cyborg was doing much of what flame/inferno was doing on windows/linux 3 year ago, up untill Autodesk/Discreet bought them and burried it in a 6 foot hole never to be seen again.
There are an incredible amount of hardware capabilities that can be used on a flame system that shake just can't handle (such as specular shading with normal interpolation in 3d).Shake can do this with recent plugins and older macros, also DF5 and Nuke already does this. Hardware rendering 3d in a compositing environment is not that big of a deal these days.

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.