pouët.net

3D-flyby Demos Impact (1996)

category: general [glöplog]
 
Bomb by Impact Studios

Bomb by Impact Studios
youtube

Well known demo? For me it was not, i just saw it today.

Some commented it was made by Bomb-members. I dont know if Nytrik made any of the 3D. I think it features some nice low-poly 3D scenes and texturing for that time. I think the music also is cool (maybe some wrong chords and some places, but who the fuck cares since its from 1996. when they worked it was really cool to listen to).
Contrast / Oxygene was released at the same party and won over this.

Some notes on my analysis of the 3D-demos back then:
For me it seems like the era of 3D texturemapped/envmap, lighting flybys etc.. took off in early 1996 and onwards. Seems like the Pentium-chip took care of some of it, being fast enough to process 3d maths for rasterization rendering. i wasnt really coding much own stuff myself back when i was 15, so this i am just guessing. Would have been nice to see some graphs and analyze and maybe make some assumptions based on the history of the computer chip. Its totally possible to trick the 486 doing fast 3d-graphics rendering as well of course, but after looking at a bit of history on the computer chip market it seems like when these people got faster machines it was easier to adapt/improve their code to do more advanced graphics rendering using more advanced math and trickery.

Theres alot of flat- and gouraud-shaded objects in demos before 1996 of course, i know a bit of history, i saw alot of demos! And I appreciated those moments seeing some new, better and faster realtime 3d graphics rendered objects on the screen and of course faster (optimized of cache, improved tables etc) 2D effects.

Just have not much time to research on the history of the processor-chip and the demoscene. One does not need to guess what machines everybody had back then, just the requirements of a what particular demo needs to make it run normal. Based on this info one could make some transparent graphs and maps for the relationship between the development of better 2d/3d-graphics and the processor and ram. Get some sense of how the demos developed through time. I dont know if any of this makes sense. Comments are welcome.
added on the 2016-07-31 21:32:20 by rudi rudi
Quote:
I dont know if Nytrik made any of the 3.


well, according to demozoo he hasn't... i believe that 3d flybism is sth new for this era with chips having more computing power, so more and more demos will use this art style as time goes, a style that later passed to games as well, such as half life, quake and system shock 2.
added on the 2016-07-31 22:41:27 by Defiance Defiance
Defiance: ok. yes, definitively. I also recall Carmack mentioned the demoscene in some of his speeches i watched on youtube some years ago. Refering that to raycasting in games like Wolfenstein etc. which came before Doom.
added on the 2016-08-01 00:41:42 by rudi rudi
Quote:
a style that later passed to games as well

errrr, right. rather the other way :)

and dunno what's so wowz0rs about this particular french demo either? there are plenty of demos in 1995 already that use the power of pentium and have textures and shading?
I guess it's just a natural progression/evolution.
At first, you moved single 3d objects around on screen ('object show').
Then you started to make more complex scenes with multiple objects moving together, which evolved into small worlds, such as in Phenomena's Enigma demo, or Future Crew's Unreal, and later the city scene in Second Reality.

I suppose the influence of games such as Wolf3D/DOOM/Quake/Descent is that instead of 'outdoors', demos started to do indoor scenes, making use of raycasting/BSP trees for efficient visibility determination.

Related to this is the game Comanche, which inspired a lot of voxel landscapes in demos.
added on the 2016-08-01 09:27:13 by Scali Scali
Explicit has a nice 3D-object-history-up-to-1996 part. :)

1994 demos mostly featured texture mapping and flat or gouraud shading. 1995 demos featured env-mapping and phong (usually faked by env-mapping). 1996 demos featured complex 3D scenes with camera movement instead of single 3D objects. There were some late '95 demos that had simple 3D scenes, like Reve and Caero. In those days, it was all about CPU and RAM. From 486-50 and 4 MB in 1994, to 486-100 and 8 MB in 1995, to Pentium-166 and 16 MB in 1996.
added on the 2016-08-01 20:26:56 by phoenix phoenix
Nope i did not. At the Time i was Still pushing pixels together on my falcon for syndrome "amiga". I had no clue about 3D. This demo was made by the french all stars at the Time : gengis , made, titan and claws. I was at the Saturne party where it took the second place after oxygene. Those guys were like gods to us.
added on the 2016-08-02 00:13:17 by nytrik nytrik
Gengis was the 3D coder, kinda fast routines he did at the time.
Compared it to NooN demos, mighty Karl coded absolutely fast code.
Dunno where hes now, prolly work for Ubisoft like many French-demo coders.

Sad thing on the other side is, gurus like chaos and ryg, does not produce nothing to the demoscene. A little glenz with a style will do :D
added on the 2016-08-02 11:00:05 by moredhel moredhel
phoenix: I remember Explicit, I liked that demo. I totally forgot about Reve and Caero also. " In those days, it was all about CPU and RAM" - that does interest me a bit today, took look back and find out more the connection between demo effects and the processor and ram evolution.

Doing shaders I felt like distanced away from the per-pixel framebuffer mindset.

nytrik: Ok. The 3D low-poly work back in the day by you and other frenchmen was very cool in demos for me that is. Before the shift to Windows, i remember i watched No exit / Nomad many times, because the 3D and texturing was up to the best I saw. The music in that demo also made the difference.
added on the 2016-08-02 11:43:51 by rudi rudi
Quote:
"In those days, it was all about CPU and RAM" - that does interest me a bit today, took look back and find out more the connection between demo effects and the processor and ram evolution.


The demoscene is (or was) all about that actually: Ever since the early days of crack intros, coders were trying to squeeze the maximum out of the hardware, try to make it do things that were seemingly impossible.

On the PC you had little to work with in terms of graphics hardware, so things indeed often boiled down to the CPU and RAM, and trying to push as many polygons as possible, with as much detail and visual quality as possible.
So you should really evaluate these early demos in the context that they were released in.
Older demos may look more primitive, but that is not because their coders sucked. They just had less powerful hardware to work with.

There's some real gems hidden in old demos, with very clever solutions to work around hardware limitations. In the PC world, you'd get a new generation of CPUs and videocards every year, so these hardware limitations disappeared over time, and the beauty of older demo code got lost.

One of my personal favourites is the spacecut-effect in Crystal Dream 2. It worked around the limitation of not being fast enough to do per-pixel z-buffering to handle intersecting objects. A few years later, everyone was able to do 'spacecut' with standard z-buffering. But the solution in Crystal Dream 2 is far more elegant and more difficult to implement.
added on the 2016-08-02 12:29:51 by Scali Scali
The funny part of this subject is that I know well all the people who did those demo since they are approximatively ALL french sceners :D

I don't know where is Karl actually, Gengis is working at a french company who do mobile video game (I met him some weeks ago), Skal (the coder of State of Mind demo) is something like TOP ELITE engineer at Google (back to France since some years so we met him a couple of time), and you probably know Oxbab, the coder from Oxygene since he is actually Naughty Dog's vice-president :)

Ha and we meet regularly Clawz too, he did a music for a demo recently (do remember wich one, I think it was for a big Oxygene Atari Falcon demo).

Also after some decades, we finally bring back Titan/Bomb at Revision 2016, and he will probably return next year :)

So, everybody is getting old, be alive and kicking :D
added on the 2016-08-02 12:33:57 by rez rez
ha and phoenix was speaking about Reve demo by Pulse (ex-Infiny), the code was done by LCA, who was working at Delphine Software some decades ago, he is now living in the USA working at Electronic Arts since ages.
added on the 2016-08-02 12:37:07 by rez rez
Quote:
you'd get a new generation of CPUs and videocards every year, so these hardware limitations disappeared over time, and the beauty of older demo code got lost.


I disagree that hardware limitations disappeared over time. It may look like that, but it surely wasnt the case. GPU's came and made it easier to code 3D, and it almost became a religion to use it, because it was easily available. Before GPUs had plastic-looking scenes, and they improved more and more by added extension etc., now you have shaders that can almost, if not, do everything. On CPU you could always improve the same old demoeffects in addition to the screen resolution, or create new ones.

Iirc Crystal Dream 2 did actual polygon clipping for the spacecut-effect (prolly not acutely accurate, but close to it). because z-buffering needed atleast an 1 byte (if not two bytes) buffer that required more fast access to other portions of RAM (if that was even available).

rez: I thought Skal worked as quantum physicist at CERN, or maybe that was another guy :P
added on the 2016-08-05 01:08:38 by rudi rudi
Quote:
I disagree that hardware limitations disappeared over time.


That is another discussion.
I meant that the hardware limitations specific for a demo would disappear.
Eg Second Reality was designed for a 486DX. But for a Pentium it's nothing special.

Quote:
Iirc Crystal Dream 2 did actual polygon clipping for the spacecut-effect (prolly not acutely accurate, but close to it). because z-buffering needed atleast an 1 byte (if not two bytes) buffer that required more fast access to other portions of RAM (if that was even available).


The 3d renderer in Crystal Dream 1 and 2 is super-fast, and renders up to 32 pixels at a time, so even on a simple 386SX it runs very fast. But if you would do a z-buffer, you'd have to render one pixel at a time. So it would totally destroy the performance of that part on 386 or 486 machines.
added on the 2016-08-05 09:21:57 by Scali Scali
so how exactly do you know so much about the inner workings of it? do you have the source? or disassembled it? :)
added on the 2016-08-05 09:34:02 by visy visy
Quote:
so how exactly do you know so much about the inner workings of it? do you have the source? or disassembled it? :)


Yea, I disassembled it.
I also fixed the Sound Blaster detection routines in Crystal Dream (that was the main reason for disassembling it, but I was also curious about the renderer, because it is the fastest flatshaded renderer ever made on VGA as far as I know. Really stands apart from other 3D demos of that era at least).
added on the 2016-08-05 11:10:40 by Scali Scali

login