The answer back from Softimage makes for interesting reading (keep in mind that “Sumatra (codename)” wouldn’t be released for another couple of years):
Subject: RE: Sumatra… revisited
From: Dan Kraus
Date: Mon, 22 Dec 1997 10:45:21 -0500
>I think we should all think about it and be a little concerned that
>after SIGGRAPH SI has not even whispered the word ‘Sumatra.’
Although we’ve been coding hard since well before Siggraph ’96, we
haven’t spoken too much about it, except at the yearly Siggraph users’
group, because we want to be certain of our ship date before starting to
set concrete user expectations.
Sumatra is a complete replacement for the currently 3D product –
modelling, animation, rendering, particle, mental ray, etc, all
integrated into a single, seamless multi-threaded environment. We’re
coding Sumatra simultaneously both on IRIX and NT – there’s no ‘port’
involved this time, which also means that we get to take max advantage
of the hardware on both sides. Of course, there’s a lot of new tools –
performance, modelling/animation, etc – but our first priority is the
v3.7 toolset, to guarantee that you can use Sumatra for exactly the same
thing for which you use SI3D today.
Sumatra will actually be preceded by Twister – a standalone rendering
product which uses the Sumatra interface/architecture, and also
incorporates the next-gen of mental ray (v2.0). Twister is designed to
be used in tandem with SI3D, so you can start using/learning the new
interface as you’re comfortable, and integrate it into your current
We currently expect Twister to ship in Q3 (Calendar) of ’98, and Sumatra
(Q4). This is behind our original target dates, but we want to be
completely certain that Sumatra is a true replacement for the current 3D
product. From the upgrade point of view, we’ll be treating Sumatra as
the release version of SI3D, which means users under maintenance will
recieve an automatic upgrade, just as you would to a point release or
>>I wanna know what the interface will look like
Can’t blame you 😉 One of the most time-consuming tasks of the
Sumatra/Twister effort has actually been understanding and replicating
the existing user model. This extends way beyond pure interface issues,
and it’s taken us almost 2 years of work with our PM and internal
development teams (including several professional animators) to
guarantee that we understand why and how data is passed through Soft,
and propose an interface re-design which improves on what we have today.
We also have the benefit of having a true in-house production team (the
Softimage Content Group), who works closely with us on tool design,
putting things into immediate practice as soon as they’re coded.
Here’s a peek at a few of the key UI issues, and what’s happening:
Speed of Access – things like parent, cut etc are not available in all
the modules in Soft. One of the things you’ll notice when working with
Sumatra is that the right-hand panel provides you with all the general
controls you need – all the time.
Tools Organization – The Sumatra UI puts things in more sensible and
intuitive places, yet respecting where the most important controls (ex keyframe)
Quick Selection Model – Sumatra has filters and presets which make life
much easier by not just making them ‘unselectable’ as is the case in
v3.7SP1, but actually letting you pre-select the type of objects you
want to grab. Makes repeated actions on a certain object type a whole
Existing Workflow – The Sumatra UI has been designed with a constant
preoccupation (‘obsession’ is probably more accurate, actually 🙂 with
maintaining the existing workflow. Specifically, things like keeping all
the major tools two clicks away, providing contextual menus (ok, that’s
new :-), work-centric focus (manage your character, not the tools) – and
most of all, pure interface speed.
Please keep the comments coming, and keep an eye on our web page early next year – we’ll start rolling out the info as we draw closer to ship.
Dan Kraus Softimage/Microsoft
Product Manager, 3D Montreal, Quebec
There was also a side-discussion of whether or not a context-sensitive UI would be a good thing; surprisingly (to me at least), opinion seemed to be split on that.