Wednesday, September 29, 2010

Looking for an alternative - II.

While looking for a software to substitute 3DS MAX, I have considered the following options:
All of them are great software and I deeply respect the all companies that produce these. I'm only writing here about a choice we made for our pipeline.

Besides trying out each software with our existing game engine, it was also useful to know about the software's background.  How is the relation between the user and the developers, how are bugs dealt, updates, support, etc.

A good starting point for my research was a survey lead by CGenie.com two years in a row, about what users think about the software of choice and it's developer company.  Other places too analyse these aspects are on user based (not moderated by the developers) CG forums like CGTalk and od[force] or of course in the developers support forum (this way you can see how a developer treats it's costumers).

Reading reviews is not always the best way to judge a software as a lot of articles like these are a bit biased in my opinion.

I have used Softimage for a couple of years while I worked in the U.S. back in 1990's, I only have great memories from the software, the company and the SGI computers.  I did play with XSI a little on Windows and have to admit that almost bought it a few years back when they had the US$ 499,00 version. Unfortunately, having Autodesk behind it now, was a big disappointment for me.

Softimage has always shined on animation, rigging and Mental Ray integration in my opinion.

I haven't had the chance to use it much, but again, and trying not to sound like a broken record, it is a great software, and production proven. But having Autodesk behind it is a turn down.  I am really afraid of investing money, time and work in a software that I don't know what it's company's plan are. Will they continue all 3 software that are relatively the same and competing with each other?

Probably the youngest in the list, can be a good thing in my opinion. Although it is a relatively new software, the people involved in its engineering came from NewTek, the producers of Lighwave.  I'd think that they would have a good insight about what worked and didn't work on the old Lightwave, and by forming the new company they had the chance to avoid some mistakes that sometimes is rooted on the core of a software.  Whereas a complete rewrite would be prohibitive expensive in a business plan for some, for Luxology engineers this new formation was the opportunity for a fresh and clean restart.  Again, I'm not a programmer, so I was really disappointed when I read that not even operating systems are rewritten from the ground up when ever a a big new version comes out (i.e. win xp -> vista -> 7).

That is why I have such a good expectation about Modo.  I have downloaded two months ago a trial version and had it running here for a while.

Poly modeling was probably the best all around from bunch, although it took me a while to start feeling comfortable with the way the tools work, but once you understand its concepts, you notice that it is designed that way for speed.

Modo user interface is wonderful as well, you can just customize it as you wish. 3DS MAX and Maya probably have the most rigid interfaces in this list. Some may think that this is not important or aesthetics only, but it is not. Probably the premise of a good work environment is comfort. Just like when you are not comfortable in your chair and you cant focus on your work, if you're not comfortable to work on your software, you wont be able to do your best and will probably give up.  So it is imperative that, instead of making one interface and try to please everyone or even worst, try to conform people to accept your style it is better to let each one work the way it feels better for them.  It is never to much to let people opt for color scheme, layouts and sizes.

Here is the father of customization. Not only you're free to chose the place you work (schemes, layouts, sizes) but how you work.  It would probably be a smaller list of things that you can't do in Houdini that a list that what can be done.  I had to say smaller but not easier, as I can't imagine something CG related that can't be done in it. But unfortunately my search has to deal with 'what we are need versus what we can afford'.  And when I say afford, is not only monetarily, because the version of Houdini that we would need is not as expensive as MAX, Maya or Softimage but because of Houdini's 'openness' of workflow it is surprisingly difficult for many people to understand or accept it :-|  I guess that some people find it easier if they have someone telling them what or how to do stuff on a systematic way.  I have heard that people who never had contact with a CG software does a better start at it, that people who has a previous CG software knowledge. Anyways, we had a greater resistance to adaption to Houdini here, and finding a Houdini user in Brazil is almost impossible.  We haven't rulled out Houdini as an viable option yet, but it wont be it for now.

I personally have Houdini as my preferred software, specially for making tools prototypes and proof of concepts.  And have been using the Apprentice version at home since version 8.  SideFX, the developers of Houdini is probably the best company to deal with, and have always responded to my contacts even though I'm not even a paying costumer.

For the longest time, I have never counted Blender, an open source CG software as a competitor on the list. But as you can see on the survey , Blender has done really well.  Version 2.5 has great improvements, specially where it is so important and mostly needed interface.  We have all tried it in the past here, and since we needed to find a way out of MAX this year, we were really surprised when we tried this new 2.5 version. All tests have been doing so well, that we even started an internal wiki about its integration on our workflow, with best practices, etc.  Besides some small tests of character rigging and animation, we have been rewriting some of our early games to be played online this time, and so we have been using Blender for our models. We still have some time in our schedule before committing to one solution for our next project.

Things that we liked a lot in Blender, pecially comparing to MAX our previous solution:
  • Lightweight, on system resources:  We haven't done complex scenes, but again, all the levels are built inside the game engine, so our modeling never gets too complex and one place.  Also these new games are supposed to run on the web, so it can never get too complex. 
  • Cusomizable UI: I haven't change the colors scheme but UI layout is extremely fluid and easily changeable.
  • Exportable: Our models are easily integrated into our game engine via .FBX format. I know .FBX is Autodesk's format. But again, we were stuck with 3DS MAX v9, and this version only accepts an old FBX format.  We would have to upgrade MAX to be able to use multiple UV layouts, for lightmaps for example.
  • Bonus features: This version of Blender has features that we had to cross out as an option for us on previous games.
    • 3D texture map painting: we previously were using Photoshop's CS4 mesh painting, but it was really heavy and cumbersome on our computers. Although we still use Photoshop for texture finishing.
    • Mesh sculpting: probably not as powerful as ZBrush, but is definitely useful for us. We never used mesh sculpting on our games, since we don't have tools for that before.
    • Integraged compositing: I haven't have the chance to give it a try yet.
Stability is something that is missing a bit, but this version that we are using are still in the beta stages so hopefully it will get better, meanwhile we just have to get used to saving our work more often.  I'll keep posting more thoghts on Blender on the course of it implementation on our productions.