[K3d-development] Lurker is back: SoC ideas

[Rene J. <cen...@ta...>]

Hi people

I'm upgrading myself to lurker again.
And as the annoying, curious busybody I really am, that will not happen 
in a very silent manner... Ach so.

Here are a few suggestions that I myself had hoped to be able to 
contribute with in a time where the K3d code was a huge beast with a 
massively changing API. It never happened, but the need is still there. 
Seriously, these are not whimsical thoughts of little value, but 
features that I daily miss in the FOSS world!

They all focus on the concept that K3d - unlike all other projects - HAS 
an ontological foundation which is object oriented, supports 
interconnection/visual programming out-of-the-box etc. etc. Such a 
framework would really be a dream to extend on.
Sorry if these suggestions had already appeared on the list. Did scan it 
though. If you want me to go further with writing this stuff up, just 
ask me. I'm not a student.

#1... CAD SYSTEM

Using K3d as an implementation bed for a CAD subsystem. I've worked with 
SolidWorks which is a high-priced CAD system, and given the number of 
bugs and stupidly implemented features, I'd say that a clean 
implementation from scratch by a mathematically wiz-kid on top of a 
pre-existing OO-framework would give something to the FOSS world WHICH 
DOES NOT EXIST YET, AND IS **HIGHLY** NEEDED.

Important concepts:

- You model several PARTS which is then put together in an ASSEMBLY

- You don't work with vertices and edges like in Blender, but with 2D 
SKETCHES made up from 2D PRIMITIVES (Circle, Rectangle, Line etc.). Once 
a sketch has been made, you apply a 3D FEATURE to it (Extrude, Spin, 
Cut-Through etc.). Now comes the scoop: You now have a 3D spaceous body. 
This body will have faces, which themselves (if they are flat, such as 
the sides of a rectangle extruded into a box) can become the drawing 
canvas of a new sketch.
So to make a house with a door-frame: Make a Rectangle on the base 
XY-plane. Apply an Extrude. Apply a Hollow-Out/Shell-Out feature leaving 
a box which is not a solid thing, but rather six thin plates of iron. 
Select  a face on the house and use it as a sketch-base where you can 
draw a Rectangle the size of your door. Apply a Cut-Through feature to 
that sketch, and tell it to cut all the way through to infinity in both 
directions  just to be sure.

- Parametric modelling taken to the extreme. You can supply a ton of 
CONSTRAINTS between all sorts of elements, and these constraints can 
often take hours to evaluate! Examples: Align the top face of a 
door-part with the top-face of the door-hole in the above example. Once 
you have done that, you have constrained the ability of the door to 
move. This shows when manually moving and rotating. The face that must 
keep aligned STAYS aligned, and you looses degrees of freedom.

What is needed:

- Most importantly a geometric analyzer, which can figure out the 
degrees of freedom of various elements

- Hard to code features like: Shell-out, Fillet (rounding edges with, 
say 5mm fillets)

- More stuff. Ask me if needed. I have given the subject some thought.


#2... ARCHITECTURE SYSTEM

This is actually different from #1. I suggested this years ago. Since 
K3d is fully object-oriented in a sense that even shows through into the 
UI (something is a mesh if it implements an interface, but it can be 
something else under the hood), making a "smart" set of objects and 
dedicated UI-widgets that helps modelling houses and cities a lot faster 
than if you had to deal in vertices, edges, faces. If you put in a 
window somewhere then both cutting the hole and attaching a child window 
object is automatically done. Cutting should NOT use booleans - that 
wrecks havoc on the topology - but something that cuts the target in 
vertical and horizontal parts).


#3... GPU-BASED REALTIME RENDER ENGINE

A recent idea. I have just finished rendering 800 frames in Blender. 
Although I used 7 machines and 3 friends, I still managed to push my 
deadline almost a full day. After the war, one of my friends (a gaming 
freak) asked me why that could take so long since he didn't think the 
result looked better than what he had seen in a recent game. I told him 
that in FOSS/Linux the developers are nerds, that have no clue what is 
really happening in the rest of the world.

He was right of course. Anyone tracking NVidia and ATI can easily see 
that one powerful GPU and a lot of trickery could deliver almost in 
realtime the same or better result than what 7 powerful machines 
couldn't do in 24 hours.

And so: Since K3d has a perfectly ready RENDER ENGINE PLUGIN 
architecture starting to make a juicy GPU realtime renderer would be 
quite easy given a certain amount of mentoring.



Curse it I'm not a student :)

Regards,
Rene Jensen