Re: Making 3-D fully textured, shaded graphics using just pixel routines?

From: TS (nomailhere_at_deinmeister.de)
Date: 11/08/04 

Date: Mon, 08 Nov 2004 10:53:20 GMT

>(...)...yes,
>difficult to know who is worse, eh? The person who wrote this "3D globe" to
>be that pedantic in their "3D modelling" to include something so minor as
>"cloud parallax"...or, that I'm such a sad geeky person, I actually
>_notice_ such things while watching the weather forecast, where any normal
>person really doesn't care or notice at all...hey, I'm a programmer and
>like to play around with graphics...it's a kind of "research", so to
>speak...oh, okay, I'm just weird ;)...

Well, maybe the "cloud parallax" was just added since it helps
avoiding z-fighting ;)
Anyway, I guess modelling the cloud layer as a slightly larger sphere
isnīt harder than just using a different method... but it depends on
how the globe is rendered...

Noticing these small things sounds familiar to me, since knowing how
to implement such 3D stuff spoilt me as well - I can hardly enjoy most
3D games or 3D animations on TV since I see bad triangulations and
parameters, mapping, banding, distortion, billboards and similar
effects which often happen.
On the other hand, things like a practically invisible LOD or nearly
perfect transparency impresses me - while most other people do not see
anything special at all.

>(...)always "bottleneck" to performance in a PC (and video is the task that
>typically guzzles the "bandwidth" like there's no tomorrow...well, you
>know: 1024 x 768 x 32 bbp x 60fps ~= 180 MB/s...in comparison, "CD quality"
>5.1 sound is only 44.1KHz x 16 bits x 6 channels ~= 0.5 MB/s (although, you
>can use "3D API" on the sound too and send over "mono" waveforms and let
>the card itself split it amongst the speakers with its built-in hardware
>for doing so...this obviously improves performance :)

About those "hardware accelerated" soundcards which mix audio streams
on their own: Imagine playing back dozens (or even hundreds - Iīve
found an .IT tune using up to 160 voices) streams at the same time -
this will put a severe strain on the PCI bus as well, thus such stuff
is often faster when done by the CPU itself.
Things get different when a lot of processing like filtering and
reverb is added - when the soundcards DSP is really handy (among its
reduced latency due to "real" realtime mixing).

>Consider that GPU...it's a _processor_...yes, a "graphics specialist"
>processor but still a processor...it has registers and is programmed with
>its own "assembly language"...as things have now turned out, though, you're
>not permitted to directly program in that assembly language but must use
>some "generic HLL" Microsoft has invented...indeed, the way Microsoft has
>tried to keep it as "assembly language-like" as possible, demonstrates, I
>think, that they also see the "issue" developing...but Windows has an
>"architecture" which mandates that all hardware is accessed via what is
>effectively a "hardware HLL"...now, adding things, removing things to an
>OS, that's relatively easy...but completely changing the fundamental
>architecture it's built upon _whilst_ still obliged to maintain "backwards
>compatibility"? Contradictory objectives...they might attempt a "slow
>change" over time but, well, we know the speed Microsoft moves technology

I donīt think that the DX shader language is like a HLL, especially
not the so-called assembler version. Especially when looking at the
first pixel shader versions, the instructions like pretty much like
the texture stage operations or the register combiners which were used
for texture processing before the pixel shader language was
introduced... Similar things can be seen in limitations or
possibilities offered in other places of the Direct3D API.

OpenGL is way more high level, there is only GLSL as a standard
shading language.

A different case is the DirectMusic API: I can hardly figure out whats
going on behind all those objects...

Something which is really missing today isnīt operating system
related, but compiler specific: SIMD instructions like MMX were
introduced to the mass market almost a decade ago, but there is still
no widespread language (or even an optimizing compiler) which uses
them. Sure, there are some languages intended for vector operations,
but these are targetted towards supercomputers rather than the
widespread small scale machines. Thus, using these instructions still
requires assembly (there are these "intrinsics", but these are more or
less the same...).

And now, there are not only multiple data instructions, but also
multiple core CPUs approaching...

>(1986 saw the '386 bring 32-bit protected mode, Microsoft only got Windows
>"mostly" 32-bit and using those '386 capabilities in 1995 with Windows
>95...and, as noted, that was "mostly", not "totally"...with XP, now we can
>begin to say "totally"...nearly two decades just to "go 32-bit"? We're in
>for a very long wait, folks ;) ]

Wasnīt NT introduced prior to Win95? AFAIR it was designed as 32bit
from the beginning... anyway, 64bit is right on its way down to the
desktop and home machines ;)

Hey, thatīs quite a lot of text, and Iīm still way beyond the amount
of letters typed by Beth *g*

Relevant Pages

  • Re: Thinking assembly?
    ... I simply needs to understand the CPU and how it works, ... > understand the instructions, be a good and experienced programmer, ... assembly language is the machine's language (or, at least, ...
    (alt.lang.asm)
  • Re: Python or PHP?
    ... > every language here and there more ways to do something. ... The best the programmer can do, as you imply, is to ... parse out into proper perl expressions. ... > lists, dictionaries, etc. etc. ...
    (comp.lang.python)
  • Re: Evolution
    ... > inline assembler and larger parts by linking to an assembler ... in the HLL. ... language, particularly in smaller projects. ... > or two instructions if you ...
    (alt.lang.asm)
  • Re: Python or PHP?
    ... >> every language here and there more ways to do something. ... The best the programmer can do, as you imply, is to ... I am curious of a list of extraneous methods in Perl (more about the ... I just had a glance on Python, ...
    (comp.lang.python)
  • Re: The War On HLA
    ... Take a look at the HLA compile-time language. ... Macros can be abused just like any other language feature. ... I find it amusing, however, that HLL programmers (e.g., Delphi, ... > feature is left to the programmer to invent his own uses for. ...
    (alt.lang.asm)