What are some retro games that were ahead of their time graphically by using almost never seen before techniques?
Picture related was certainly interesting. Characters had bump maps, but it was all done in software.
Magic Carpet had wavy water, reflective water, and realtime terrain deformation. Somehow it even ran on the PS1.
https://www.youtube.com/watch?v=-EiRSQPhtZs
>>3189904
Wasn't terrain made from voxels as well?
>>3190147
I wonder if Mechner ever is sad that the only game of his to be real popular was not THAT innovative - Prince of Persia - when his really innovative and mindlowing stuff - Karateka, Magic Carpet and Murder on the Orient Express - is almost forgotten?
>>3190286
... I can't fucking believe I misnamed The Last Express. Wow.
Outcast broke so much ground and received so little reward for it.
Depressing, really.
Giants: Citizen Kabuto was incredible and is still somewhat good looking today.
The gameplay itself is totally unique and the style of game hasn't really been replicated since.
>>3189904
bump maps are the least interesting bit of that game's visuals
>>3190147
it also had a transparent HUD and dynamic lighting. That game was the fucking Far Cry of its time
>>3190296
because it was done in software. So it ran relatively choppy and at low resolutions. Doesn't matter that the game practically pioneered the pixel shader, graphics whores preferred the nvidia approach, tons of stupid polygons and simple textures
>>3190147
Wetrix on the N54 had similar features, but since it's a puzzle game it's not half as ambitious.
>>3190510
I think its obvious nvidia had the right idea, considering how much more complex GPUs have become.
>>3190612
it also came out 5 years later, and used plenty of hardware acceleration
Obviously
>>3190616
nvidia fucked things up, badly. While 3dfx was going towards better lighting, high colors and effects, nvidia pushed plain polygons and minimal effects. Outcast showed what pixel shading is capable of, and 3dfx pushed hard in that direction, but lost in benchmarks to retards that only looked at numbers, instead of the output. The modern pixel shader exists despite nvidia's early attempts, not because of them.
>>3190621
The only 3dfx card with any sort of pixel shading support would of been their unreleased Rampage card.
>>3190626
3dfx introduced fsaa, sli and the t-buffer. Also 32-bit output. They also remained at 1024x768, putting the extra power into fillrate, for multitexturing, overdraw, and so on. They were constantly pushing for better looking visuals, not more polygons
>>3190621
Nvidia cards had hardware T&L before the competition.
>>3190629
The Voodoo 5 was only released as a stop-gap because Rampage was taking too long. It wasn't their tech of the future.
>>3190630
indeed, polygon crunching. About all they were good for. I'd rather have good visuals
>>3190629
None of that has to do with pixel shading. I'm a 3dfx fan myself but its pretty obvious they gave up the "speed is most important" fight since they were losing it.
>>3190646
You're being deliberately obtuse or what? I said "good visuals" and similar phrases in every single post for a goddamn reason. I don't know how much more simple I can make it for you.
3dfx: make the individual pixel look good
nvidia: POLY-fucking-GONS!
They didn't do the pixel shader yet, because tech was still in its infancy. Out of the two mindsets though, 3dfx was much closer to it, than nvidia.
>>3190653
What I specifically stated was pixel shaders, but then you brought up the Voodoo 5, which has none. Its you're fault bro.
>>3190656
I have not mentioned the voodoo 5, and I have to assume at this point the only reason you latched on to the pixel shader (which I mentioned on a conceptual level) is so you could disagree with someone on the internet
>>3190663
I don't give a shit about the voodoo 5. All these features are made to improve the image quality and fill rate
>>3190665
Not really. The AA only works thanks to the SLI, which they wouldn't even need if the chips weren't so slow. Hence why the Voodoo 6000 is capable of doing 8x AA with its 4 chips while the Voodoo 5 is limited to 4x.
>>3190668
>The AA only works thanks to the SLI
So what? Only thing I cared about is that they did FSAA, actually cared about dealing with the sharp edges. The t-buffer was worthless too, holding only 4 frames. But it was used for depth of field, soft shadows, soft reflections and much more. It ultimately failed, because the approach was the wrong one, but it was about improving the image quality, which is what matters.
You seem to have a problem with concepts
>>3190673
I can't even think of any games that used the T-Buffer.
>>3190673
3dfx sure changed their minds pretty quickly, after sticking with 16-bit so long until the voodoo 4/5.
>>3190510
ATI themselves had the earliest tessellation implementation in their Radeon 8500 cards before everyone else did. Wasn't used very much either.
>>3190864
Tesselation is just throwing more polygons at the problem, which is a dead end
>>3190864
Nvidia had some decent tech for shadows called shadow buffers in their Geforce 3.
http://www.nvidia.com/object/feature_shadowbuffers.html
>>3190629
Nvidia had a true 32 bit framebuffer since RIVA TNT and TNT2.
Meanwhile 3dfx played around with 16 bit buffers, dithering and ramdac filtering tricks to output at "22 bits" because whatever, can't bother putting 32 megs on board like TNT2, that would cost money.
>>3190626
Actually even the first Voodoo had a color combined unit. Sure, it's not as flexible as 'modern' pixel shading, but it's still a pixel shader.
>>3190629
Actually the N64's GPU (SGI RCP) introduced FSAA.
>>3190630
It also introduced one chip hardware T&L, before Nvidia did.
>>3190864
ArtX's Flipper (GameCube GPU) is the first chip that I know of with hardware supported tesselation, and it's a literally a month older than Radeon 8500. Although it had some serious practical limitations that prevented much use.
>>3191218
>Actually the N64's GPU (SGI RCP) introduced FSAA.
not on the IBM-compatible desktop computer, or the 3D accelerator. Also, wasn't the N64 edge AA?
>>3191221
Actually yeah, it was edge AA. I was thinking of how the earlier Voodoo cards had edge AA, while the poster meant the Voodoo 5's FSAA which is a different thing.
>>3191231
that poster was me. Frankly, I consider FSAA a bit overkill. The solution of the V5 to practically everything was fillrate, it was crazy, but also kinda cool. In my understanding though it turned out finding the edges and only doing AA on them is more expensive than just doing FSAA.
By the way, you're definitely right that the N64 debuted a lot of stuff that would only make it to the desktop years later. It had plenty of shortcomings as a system, but that GPU was positively nuts
>>3191240
>but that GPU was positively nuts
Well SGI did promise Nintendo that the console GPU they would provide will support every single SGI workstation feature. That's the reason RCP's feature-set is absolutely top of the line for 1996 standards. (After SGI crashed and burned, the engineers of RCP left to form a company called ArtX and created the Gamecube's GPU Flipper)
The original Voodoo isn't actually very far off that feature-set either (little surprise considering 3dfx were ex-SGI engineers). I think the only thing it is missing compared to RCP is hardware T&L and (the potential for) 24 bit color output.
Voodoo 5 is a fascinating design though. Totally different to what everybody else was doing in the industry, but creative and innovative. I kinda feel that 3dfx's engineers were getting a little crazy at that point in regard to meeting industry standards.
Under Nvidia's ownership they developed the GeForce FX which is actually regarded as one of the worst GeForce series for one really big reason - the shader floating point standard of 16-bit differed from the DirectX 9 standard of 24-bit. So the cards were slow as fuck since they had to do 24-bit as 32-bit (doubling 16-bit) instead. Ironically this was a glorious moment for ArtX who were under ATi at that time, who developed the Radeon 9700 Pro, considered to be ATi's best designed GPU of all.
>>3191253
>I think the only thing it is missing compared to RCP is hardware T&L
kind of a big miss, sadly. As much as I loathe nvidia from that period, hardware T&L is the correct thing to do.
>I kinda feel that 3dfx's engineers were getting a little crazy at that point in regard to meeting industry standards.
Yeah, I love it. Also, let's not forget the low level API they tried to push as a Glide successor. Now, much later, low level graphics APIs are a reality
>>3190286
Were there any game prior to PoP that used rotoscoping for sprite animation? Because I'd argue it's a pretty good achievement.
>Hey look! An interesting topic!
>Half the replies are retards arguing about gay graphics cards.
>>3191268
Nvidia's first bad experience with NV1 soured them from being innovative (for the most part) and turned them into what they are now - OEM focused whores.
But I don't mind them too much from that period. Their 'Detonator' drivers were really good compared to the competition. And their GPUs had good backwards compatibility, all the way up to GeForce FX their cards supported fog table emulation for older Direct3D games which was neat.
Glide was cool as fuck though. It was the only way developers could tap into the features on the Voodoo. Like the color combiner unit. That's where the misconception of the Voodoo card not having pixel shaders comes from.
The DirectX API isn't compatible with the color combiner implementation, that's why people act like it doesn't exist. But with Glide you can tap into it.
>>3191280
Does that explain why, for example, Sub Culture on Glide had colored light sources, while in Direct3D they were just white?
The color added so much to it. The normal headlights were yellow-white-ish, the strobe light was blue, the ore glowed orange. Really gorgeous
>>3191285
That's certainly possible, but of course you don't need a color combiner to do colored lighting.
>>3191274
first case of mo-cap, it's when game development went downhill
