PDA

View Full Version : your idea about in-house cg softwares


hprogrammer
12-11-2006, 04:35 AM
Hello,
i read about popular cg companies such as pixar,blue sky,dreamworks,squire enix,etc.
i noticed that these companies use their own in-house cg softwares such as sgi workstation,
renderman,Marionette,fizt,etc.
what is their purpose?
why they don't use powerful packages such as maya,softimage|xsi,lightwave,etc.

Per-Anders
12-11-2006, 08:07 AM
um well a couple of things, firstly SGI workstation is not software, it's as it says a workstation computer by SGI (Silicon Graphics), Renderman is just a standard so I suspect you mean PRman (Photoralistic renderman), which PIXAR make and sell, anyone can buy...

anyhow ignoring that stuff for now, they use proprietary software because it fits their pipeline and allows them control, it also allows them to do certain things that generic off the shelf software is not designed to do efficiently and quickly. on top of that much of the custom tools development that studios go in for is to fill in the gaps in the generic applications toolsets where either a tool doesn't exist, or a tool exists but doesn't work in the way required or even if a tool is substandard (and many other similar reasons).

those "powerful" off the shelf packages are geared towards generalisation, and are excellent in their way, however they lack the specifics that make up a good pipeline, or would make certain effects possible in a timely manner. that's why with the more powerful of these applications their scriptability and public API/SDK's are so very substantial and deep, to allow their full customisation and integration into existing pipelines, and the plugging of the gaps.

gga
12-11-2006, 11:25 PM
Hello,
i read about popular cg companies such as pixar,blue sky,dreamworks,squire enix,etc.
i noticed that these companies use their own in-house cg softwares such as sgi workstation,
renderman,Marionette,fizt,etc.
what is their purpose?
why they don't use powerful packages such as maya,softimage|xsi,lightwave,etc.

a) Most of them also DO use maya, xsi, lightwave, etc. for some tasks or to complement their own software (and also keep some talented artists happy). However, companies that strongly believe in propietary software will often not advertise this fact very much.
b) One reason for propietary software is to allow companies to do things that other companies (hopefully) just cannot do, cannot do with the same level of quality or cannot do so in the same time frame without overworking their artists.
c) A more common reason for propietary software is to also keep costs down. Consider that a full license of maya costs U$D8,000 and that a big company may end up having 50+ artists and more than 100+ machines in their render farm. That easily adds up to a million dollars or more just in software... all of which has to be renewed every time a new version comes out (each year, usually).
d) Finally, propietary software also gives you some level of control. Say, for example, that maya decides to incorporate mental ray as their primary rendering solution, but your whole facility is based around prman. Obviously, you are not too happy about that. If you control the software itself, that cannot happen.

Propietary software is not without its problems, thou. On one hand, you need to hire smart people that will be willing to work on it for long periods of time. That's usually not a cheap investment.
A competitive render engine to mental ray or Prman, for example, can take about 5 years to develop (and that's assuming you know what you are doing). A competitive 3d animation package, problaby more.
Propietary software requires that you keep investing on it so that your software remains competitive to what's available commercially or even, as open source these days.
Good technology is also not enough, as you also need to make sure your software is intuitive enough for artists to use (who more likely than not come from a background of using commercial tools).
Finally, if your main reason for preferring propietary software is to obtain a technical advantage, you need to also make sure your software not only copies what's available in other software but your programmers must come up with new ideas every day. This is an additional burden on top of doing a good or perfect implementation of already well-known ideas.

beaker
12-13-2006, 11:20 PM
Too add to gga's list:

1. Fixing bugs is a little faster with in house software. Depending the software, a patch might take weeks to months with commercial packages where in house software you can get a major bug fixed in a week or so.

2. With huge scenes like what we do in film, all commercial software likes to break.

3. Large fx companies get really paranoid about packages being EOL(end of life) along with support and bug fixes ending. So much to the point where Dreamworks Glendale build a custom pipeline around commercial software. They hired Alias with their own in house development team for Maya and added a bunch of custom tools to it. Essentially built their own version of RAT called Luigi to connect Maya to Prman. Someone up high became very paranoid about Apple buying Alias and had them move to the PDI tools. They also dropped Shake and started building their own compositor.

Robert Bateman
12-15-2006, 02:52 AM
They also dropped Shake and started building their own compositor.

Shake being a good example of proprietry software....

http://en.wikipedia.org/wiki/Shake_(software)

Normally though, in house software is not full blown packages. Normally you are talking about small tools, plugins etc. It's done simply because

a) when you start pushing the limits of a 3D package, you normally hit those limits very quickly - so you need to extend the package to fit your needs.
b) computers are good at repeating monotonous tasks. Artists start complaining after 5 days that a small tool could be written in an hour to automate things....
c) Maya/Max et al don't have any tools to facilitate 100 or so people working together, so the vast bulk is organisational pipeline stuff - tools you just can't buy off the shelf.

beaker
12-15-2006, 03:51 AM
Shake being a good example of proprietry software....

http://en.wikipedia.org/wiki/Shake_(software)? What do you mean, Shake is commercial software, not proprietary.

Normally though, in house software is not full blown packages. Normally you are talking about small tools, plugins etc. It's done simply becauseIt is at large houses, PDI, DW, ILM, Sony, R&H, Blue Sky, Pixar, etc..... Any company over 150-200 people.

Robert Bateman
12-15-2006, 09:58 AM
? What do you mean, Shake is commercial software, not proprietary.


It's commercial now, but started as proprietary software at sony imageworks.....

jude3d
12-15-2006, 11:53 AM
shake was created by a french guy If I remember, many company use in house software for many reason, the mosstly is to control the tool and devellop in parrallele to the needs, it's better to know a core of an engine to devellop it faster. Most studio use also their own phylosophy about tools and want most simple and robust tool to use.
Of course the trust of comapny is a deal, look at maya 8.0, it's the first really buggy maya version because of autodesk, alias was good, autodesk have to be in the next release. those things are very bad for production if the tool is not stable. Is all about money, trust in software, and involving for the future and the producton.

leigh
12-15-2006, 03:40 PM
a big company may end up having 50+ artists and more than 100+ machines in their render farm.

Hehe, multiply those numbers by a few other numbers and you'll be closer to the mark ;) A 100 processor render farm is a very small render farm for a "big" company :)

beaker
12-15-2006, 05:24 PM
look at maya 8.0, it's the first really buggy maya version because of autodesk, alias was good, autodesk have to be in the next release. those things are very bad for production if the tool is not stable.bah, what a load of crap. It's the same exact team as it was before the aquisition. FYI, most people in production consider 8 the most stable version out there.

beaker
12-15-2006, 05:32 PM
It's commercial now, but started as proprietary software at sony imageworks.....No it didn't. NR guys wrote Sony's in house compositor called Bonsai before they made Shake. Many of them worked on Wavefront Composer and at TDI before they merged.

hprogrammer
12-17-2006, 08:12 AM
Thank you,
but what's your idea about this question:
which one is better:
writing plugins for commercial software packages.
OR
developing our own in-house software from scratch.
my reason is to obtain technical advantages.
again,Thank you very much

jude3d
12-17-2006, 08:44 AM
bah, what a load of crap. It's the same exact team as it was before the aquisition. FYI, most people in production consider 8 the most stable version out there.

yeah it's the same devvelloping team but not much the alias policy so autodesk launched maya 8.0 without debugging much before, the P10 the fifth patch available to correct bugs is not really full of uncrash, we have many crash and srange bug with maya 8.0 and all the patched version , no problem with maya 7.0 so we continue to work on the seventh one.
but autodesk told maya 8.5 will be stable, so we will see that.

ThE_JacO
12-17-2006, 09:21 AM
yeah it's the same devvelloping team but not much the alias policy so autodesk launched maya 8.0 without debugging much before, the P10 the fifth patch available to correct bugs is not really full of uncrash, we have many crash and srange bug with maya 8.0 and all the patched version , no problem with maya 7.0 so we continue to work on the seventh one.
but autodesk told maya 8.5 will be stable, so we will see that.

Autodesk or not maya8 and its quickfixes followed exactly the same pattern of 9 out of 10 previous releases.
I'm not a fan of autodesk, but this is just griefing for the sake of griefing, without knowing any of the factual elements or context of what's going on in Alias/AutoDesk.

What will happen in the next year will start showing some AD footprints, but it's not like because Alias got bought everything they do gets immediately comformed to the buyer's previous practices.

Matter of factly the Maya team was left alone for a fairly long time, other then some reshuffling, after every time Alias was bought out (which happens every other week these days).

mummey
12-17-2006, 03:31 PM
Of course the trust of comapny is a deal, look at maya 8.0, it's the first really buggy maya version because of autodesk, alias was good, autodesk have to be in the next release. those things are very bad for production if the tool is not stable. Is all about money, trust in software, and involving for the future and the producton.

Yeah, I gotta agree with beaker. You've obviously never used Maya 1-6.0 (6.5 was _finally_ good enough to be called "stable")

mummey
12-17-2006, 03:35 PM
Thank you,
but what's your idea about this question:
which one is better:
writing plugins for commercial software packages.
OR
developing our own in-house software from scratch.
my reason is to obtain technical advantages.
again,Thank you very much

In short, BOTH!

- If you're more interested in the structure of a vfx or 3D app, design and write an app.

- If you're more interested in the math and code behind the 3D and effects themselves, then don't worry about the app and go straight to writing plugins for existing applications.

dlanier
12-18-2006, 07:06 AM
Hi,

If the 3D package fits more than 75% of your needs, you don't need to build a in-house application from scratch... So developing plug-ins might be enough.
But if some of the main features you'd like to use in the 3d package don't work because you overload its capabilities, developing in-house software is probably the best solution you have... More over, you may want to sell this package if its successfull.

But as said previously, all is a question of cost and time, and biggest companies have time and money to build their own in-house software.

One of the best example is probably the Massive software for crowds simulation and rendering. At first it was an in-house tool then become a commercial package.
Regards,

hprogrammer
12-19-2006, 12:36 PM
Thank you very much,
now i can solve my problem
now another question:
I want to develop the polygon modeling process of maya.
the purpose is to do modeling as fast as i can.
but i can't understand which part of modeling process takes much time of work.
chould you help me in it?
thank you again and sorry for my bad english :thumbsup:

janimatic
12-28-2006, 05:21 PM
this last question should be the 1st one :
- which tool would be really usefull ?
Unless you are surrounded by skilled 3d graphists and you understand their needs, you should focus on using existing tools first, to find the real gaps to fill.
For example, dynamics is an area that could be very interesting for the programmer, and the production needs vary very often. You could then get fast feedback from users.
- On the other side writing a simple subd modeler from scratch can take a whole life and never success...

CGTalk Moderation
12-28-2006, 05:21 PM
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.