CompuServe Messages

#Siggraph/MAX/WOW!!

    13-Aug-95 21:17:26
Fm: MARTIN G FOSTER 70303,456
To: ALAN IGLESIAS 70262,1240
dear all, I'd like to add my thoughts on Siggraph. It certainly was a great show for a new generation of software on the cool NT platform. The title "MAX" which looks good in text but is confusing to discuss verbally in computer circles (I call it 3dsNT in conversation, or "Bruce" to avoid confusion <g>) simply stole the show. Even SGI diehard sceptics at my company were floored. Alan had to be sedated after getting a tour of the product <g> and my mind was racing wildly over the possibilities. My feeling is that much of the promise of the product will be fulfilled by 3rd party participation and it will take until New Orleans, at least, to see a good x-section of this. Although, this is already the case now with r4 ipas development, it should be many orders of magnitude more applicable and powerful with MAX. The architecture seems to allow this to the fullest extent and should attract new, world-class developers AND allow C++ savvy users to write custom code. I, for one, will be dusting off my C skills, and learning the C++ way. I still love "keyscript's" quick and dirty basic BASIC, though, and hope that will be incorporated into MAX. I was also delighted to see that most of my 3 pages of wishlist items seemed to be have been incorporated PLUS things I could not have even imagined. MAX seems to out-Softimage Softimage and looks to be a great environment for character animators. Glenn Melenhorst should wet himself when he sees this <G>. Hardware-wise, there was manna from Nirvana aplenty – if Adesk decides to support the other RISC processors (especially DEC alpha) which they would be foolish not to. I priced out some Alpha boxes called Alphaservers and they are SGI killers in price and performance. For example, one with 4 cpu's (21167 300mhz available soon), 4mb cache per CPU, 1gb RAM, 9gb disk which would be used for a "render-server" with a street-price around $120k. The performance is about 28 pentium 100's nominally, but with large scenes which take buckets of RAM (500mb+) the payoff would be enormous. I call it an SGI killer because it's about 80% faster than MetroLight's 8 cpu (R4400 200mhz) SGI Challenger XL for 25% of the price. Just like the Challenger these Alphaservers work very effectively with multiprocessors and don't have a lot of overheard per CPU. You maintain most of the performance even after adding several CPU's, unlike with the Intel m/boards I've seen. This is probably due to the huge, fast caches dedicated to each CPU. The useful maximum number of Alpha CPU's with DEC's architecture seems to be 12 <choke> <g>. Caution: these are initial estimates and may not be highly accurate but the point is that the Alpha's could be a cost-effective high-end alternative. However, I still haven't counted out Intel. They could still give the RISC boys a run for their money with the P6 – I'd like to be able to compare render-servers based on 4-12 P6's with similar design to the Alphaservers (i.e. large individual cache, shared RAM). In any case, I'd much rather have 1 render server than a bunch of individual boxes for a render farm – it's so much easier to work with. Well, so much for the hardware/software situation. It looks like we've got those bases covered for the future. I wanted to discuss some issues in the entertainment industry, specifically film/TV/commercial – one small area of CGI – so please don't take offence if you're not focused in that area. There is a real problem of talent – or rather lack of it. There's an industry-wide shortage of 3d talent generally (everybody was bitching about it), and there's a chronic shortage in 3ds-land. I think this will only get worse when MAX ships because I think more entertainment companies will jump on that product (maybe leaving SGI's completely) and will be looking to hire qualified 3ds people. This is one area where the SGI world has 3ds beat. (this should come as no surprise since that's where SGI concentrates). They have a trained and experienced talent pool, many with 5-10 years experience in _production_. Whereas, 3ds has lots of mainly self-taught folks with 0-2 years, a few with 4-5. Looking at the 3ds tape and looking at Softimage's and Alias/Wavefront's there are still much higher production values overall in the latter 2. They have the money and they have the talent. There is also world class animation on the 3ds tape, but overall IMHO, there is more of it on the SGI tapes. Ironically, there was some of MetroLight's 3ds animation (artist: Lars Fields) on the Alias/Wavefront reel and they didn't even know it. It was done entirely in 3dsr4 with no Alias/Wavefront at all – however Lars was working with many of the top people in the business, including Mark Dippee of ILM, the director. So he certainly wasn't working in a vacuum, but this does reinforce my point that 3ds has the quality if placed in the right hands. Because 3ds is newer and by it's nature of being a less-expensive product that individuals can afford, there is a lot of very poor animation being done. Poor lighting, awkward composition, bad timing, awful bumpy cameras, no use of ease to/froms. These are the things I see all the time in 95% of the demo reels I receive. Certainly, 3ds is very empowering for individuals, and I'm a product of that, but people should not ignore the need for education and training. This need not be formal and it's not a mystery: there are books aplenty on film-making, photography, art, and animation techniques. I study these and continue to do so – the need never ends. Hey, I'm not even that good and would like to work with people much better than myself, but they are hard to find. My hope is that SGI artists will start to defect to MAX, but it's a long shot. SGI apps are pretty deeply entrenched and many would have a hard time with that kind of a move. Hopefully, I can convince them otherwise. Overall, what I'm saying is we're going to have the tools (hardware (Alpha, P6) and the software (MAX + 3rd party plug-ins)) to go head to head with, or even beat, the best of them. However, cultivating the people to use them is going to take a couple of years. That's just some personal notes from my point of view, which I admit is kind of a narrow one. It's going to be an interesting year between now and New Orleans. I welcome your comments 🙂 Martin