Past and future

[Angel Jimenez]

Hi all.

 I am going to talk about myself, so you can skip to the technical details
if you get bored :-)

 Well, I must apologize for my silence lately, but, as some of you know, I
have been working full time on different things, and I couldn´t save time
for Panorama.

 To be honest, that´s not the whole truth. I have been progresively
dissapointed by the lack of real usability of the project as it was. Nobody
was using it for anything serious. It was a nice platform for trying out
some ideas, but nothing more.

 Some time ago, I started thinking about my future, so I decided to work on
some nice ideas, and keep the code for myself. I worked on physics, doing
cloth, rope, hair, and fluid simulation. I also worked on some global
lighting stuff, researching photon maps and other Monte Carlo techniques,
and then, one day, I discovered real time graphics...

 Then came a lot of changes. For example, I don´t have Linux at home now.
Not even a backup on some HD. Only Windows 2000, and soon maybe XP. I have
been happily working on VC++ for almost 2 years, using OpenGL and DirectX,
and I have learnt a lot on that time. If I had to write Panorama now, it
would be very different.

 Ok, you can stop skipping now :-)

 On one hand, I think most of the techniques used in real time graphics
could be used to reduce rendering time on a photorealistic renderer. On the
other hand, I don´t think anybody is using a raytracer in production at
all... Most of the people are using Maya, 3D Studio, Softimage, etc, so, for
them, the world is a bunch of triangles and little more...

 Well, I will give you my idea in two words: kill panorama. It´s mostly dead
now anyway, and, although some people is interested in working on it, I am
sure most of you will agree that the world has changed a bit, and the main
idea behind Panorama is not that good anymore...

 What I propose is: why don´t we start from scratch, give a more realistic
goal to the project, and use as much code as we can from the old one?

 My proposed goal is simple to describe. A system that takes Maya (or any
other 3D soft) input, and make an image/movie out of it. But, instead of
forcing this program to use our format, we will use its own one. Instead of
having every possible geometric object, we will have the fastest possible
triangle raytracer. Instead of creating a whole new material system, we
could use the existing ones...

 I am talking about Maya because it is probably the most common 3D soft, and
it is both in Linux and Windows. As you can imagine, I would be working on
windows now. But the project could be multiplatform from scratch (it is not
hard, if you do it that way). So, instead of getting a lot of messages from
people asking if we are going to make a Windows version, we will have them
as users from the start.

 I won´t give many technical details now. It would be a simpler design,
taking a lot of code out from the project (my apologies to the authors, but
most of them is mine anyway...). It could be split in two parts: front-end,
with a scene input plugin (modular desing, of course) that creates a
database of the scene/film. And a back end, with a renderer that uses some
image/video output plugins.

 From the begining, I would separate both layers, so we could have a
front-end working with several back-ends (a render farm), or even the
opposite, with many front-ends attacking a single back-end (for things like
a rendering server, for example).

 I would keep the modular design of Panorama, so these systems would be
dynamically loaded plugins. But, about the renderer... I would take another
approach. Instead of having every possible rendering technique, which won´t
get us anywhere... why not using a multi-pass rendering, using the most
appropiate renderer for every task, and blending outputs together to make
the final image? Many people are already doing that by hand in computer
graphics.

 Well, I think that´s enough for this message. I would like to listen to
your opinions. What do you think about the idea, what would you add, remove
or change, etc. I have been working on the design, and I will give you a
document once it is complete, and I have arranged the issue of intellectual
property of my work with my current employer (lovely british companies...
everything you do, no matter when or where, is property of the company...).
This document would be just a draft, so you can read, and make changes to
it.

 Regards,
 Angel

PS. BTW, another possibility, in case somebody is pissed off about this, is
keeping Panorama as it is, and starting a new project...



_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com