Newbie-alert! Nanny needed.
Carsten
I should have been here a long time ago. I've needed a wall to play my
ball up against.
I've eventually managed to make a workable 3d-editor for primitive
geometry like cubes and spheres and so on (finished some months ago).
They are scaled in orders comparable to a unit equel to half
clientarea.width or so. It works well, when I set
Device.Transform.World and my generated objectMatrix, like
Device.Transform.World = objectMatrix
This is conseptionally wrong, but ... where is the 'slot' to present
the Object-matrix to the device?
So now, three month later, ... confronted with a messy code ... I'm
hesitant to change anything, until I know that I'm doing it /Right/.
Doing a realistic 'Earth' with a diameter like 6000km, there's a lot
of rolling with the mouse-wheel from (0,0,0) to whatever coordinate
6000 taps away.
Though I can make it appear appropriately on the screen, I would like
the code to be understandable for a later revisit. By some miracle,
the general structure of the program is fairly easy to understand ...
it obviously needs a serious overhaul, cleanup and commenting ..., but
classes like Filemanager(I keep the generated shapes and projects
external), Shape(provides an objectmatrix and a ref. to a pair),
Project(a list of Shape), Pair(vertices and indices for cube/sphere-
primitives etc), Register(to keep track of where the vertices went),
and the final joy, the Generator where I can code new primitives (as
long as the function generates a Pair). There are 3 .dll's, one for
the Editor_form (the .exe), one for the specific classes mentioned and
one for presenting the graphics. The last one is a short one written
in C#, the rest in vb ... yes, I'm a vb-whimp (any consolation, that I
wrote the first edition in notepad, and the C# compilation in vb?).
I've coded with the fixed pipeline, so it's fairly primitive compared
to what it could be. I miss to distribute vertices to more than one
stream-source though, but optionally I can persist a little more
information in the current structure ... so far my Register-class has
done a fine job of keeping track of everything. And.. do I hate to see
my screen go black and the pc reboot..
My 3d-editor is a fine little thing(everything, including libraries
can fit within a 1.4 mb floppy), and I'm mighty proud.
Carsten
I'm not accustomed to this news-reader, so any follow-up probably ends
out of place.
Carsten
> Howdy all
>
> I should have been here a long time ago. I've needed a wall to play my
> ball up against.
Ok. I'll keep playing
> I've eventually managed to make a workable 3d-editor for primitive
> geometry like cubes and spheres and so on (finished some months ago).
It's taken an obscene amount of code with about 1/1000 of the calls to
DirectX3D.
One of the neighbouring messages here puts a link
http://www.xmission.com/~legalize/book/download/index.html
for what I've been looking for at the ordinary libraries while beeing
off the net, and in vane. The author puts a smile on my face by naming
a chapter "Concrete classes". It's the DirectX.vectors, matrices etc.
Yep ... the rest of ..X3D is something of an abstraction.
... I could have avoided a rather elaborate class, the cShape, where
I've put all realtime manipulation, if I can count on one tiny bit of
information: Does Device.Transform.World1 , 2, 3 etc concatenate, or
are they separate and just assures, that it will be followed by
Transform.View ?
It would be easy to test, but if anyone of you knew ...
> So now, three month later, ... confronted with a messy code ...
> ..'snip'
> The general structure of the program is fairly easy to understand ...
> it obviously needs a serious overhaul, cleanup and commenting ..., but
> classes like
> 'snip'
> ... Pair(vertices and indices for cube/sphere-
> primitives etc),
I'm adding a hint to this class, as to the Render-type (indexed or
not) of the vertices. The code has obviously been wrong, but still
working.... I have a bad feeling as to what's going to happen, when I
use the correct code!
> 'snip'
> I'm not accustomed to this news-reader, so any follow-up probably ends
> out of place.
Is it ordinary standard, that entering the 'visual' below and press
'Send' consequently takes two attempts?
Cheers