3dfx patent interpretation...

3dfx patent interpretation...

Post by CitiZen » Sat, 26 Sep 1998 04:00:00



3dfx's patent :
http://www.patents.ibm.com/details?patent_number=5740343

Their patent essentially claims they own the idea of having a device
(apparatus) that combines textured pixel's (refered to as a signal) with
any blend algorithm, as well as the method it uses to do it. It also
mentions that one of the features is to be able to use it serialially, so
that you can do triple, quadruple, etc. multitexturing.

It sounds like a very lame (weak) patent. A texture-map is an image, and
image blending has been around much longer then their filed claims. They
laid claim to something merely because it has a diffrent name. If we
refered to textures as images, they would not have a valid patent, how many
apparatus's have been invented for blending (compositing) images. Keep in
mind an apparatus can exist in software in the form of a function or
algorithm as well as hardware.

CitiZen X

 
 
 

3dfx patent interpretation...

Post by James Sharma » Sat, 26 Sep 1998 04:00:00


CitiZen,

I looked carefully at the patent and to be honest I don't think they are
even trying to claim ownership of the methodology envolved. The patent
containes some pretty 3dfx chipset specific details which lead me to belive
that the primary purpose of this patent is not to protect any global
concepts such as texturemapping and blending but rather to pretect the
specific details of implementation on the 3dfx system.

This kind of patenting is relatively common and is designed to prevent over
vendors from producing 3dfx 'compatible' cards drawing on the marketting and
developer relations record without the expence.

I very mutch doubt (my thoughts, I am not a legal expert) that a patent
worded like this could be used to enforce royalties on a different
textureing or multitexturing design so it should not be great concern.

James,


>3dfx's patent :
>http://www.patents.ibm.com/details?patent_number=5740343

>Their patent essentially claims they own the idea of having a device
>(apparatus) that combines textured pixel's (refered to as a signal) with
>any blend algorithm, as well as the method it uses to do it. It also
>mentions that one of the features is to be able to use it serialially, so
>that you can do triple, quadruple, etc. multitexturing.

>It sounds like a very lame (weak) patent. A texture-map is an image, and
>image blending has been around much longer then their filed claims. They
>laid claim to something merely because it has a diffrent name. If we
>refered to textures as images, they would not have a valid patent, how many
>apparatus's have been invented for blending (compositing) images. Keep in
>mind an apparatus can exist in software in the form of a function or
>algorithm as well as hardware.

>CitiZen X


 
 
 

3dfx patent interpretation...

Post by CitiZen » Sat, 26 Sep 1998 04:00:00




Quote:> CitiZen,

> I looked carefully at the patent and to be honest I don't think they are
> even trying to claim ownership of the methodology envolved. The patent
> containes some pretty 3dfx chipset specific details which lead me to
belive
> that the primary purpose of this patent is not to protect any global
> concepts such as texturemapping and blending but rather to pretect the
> specific details of implementation on the 3dfx system.

  Which specific details? I'm trying to find the hardware specific stuff,
and it is very vague, with no refrence to the aparatus's form (I mean they
are not supposed to hint around).

  An example, another patent that was granted in the past was for hardware
that used digital video playback for interactive purposes (they covered the
hardware in detail), and years later (I guess they thought every one
forgot) was used to claim the invention of digital video playback in
software as well, and was persued through lawyers against 3D realms of
Apogee software because there game Duke Nukem 3D used digital video
playback interactively. If it works in this case, and frightens businesses
with costly and long lawsuits what could keep 3dfx from doing the same?
Might as well face it there patent says they invented "Texture compositing
apparatus and method", and that correctly fits there entire patent, with no
specifics on hardware.

Quote:> This kind of patenting is relatively common and is designed to prevent
over
> vendors from producing 3dfx 'compatible' cards drawing on the marketting
and
> developer relations record without the expence.

Except when we get the lawyers fees and/or patent fees, as part of buying a
graphics card.

Quote:> I very mutch doubt (my thoughts, I am not a legal expert) that a patent
> worded like this could be used to enforce royalties on a different
> textureing or multitexturing design so it should not be great concern.

Actually, how can any texturing and multitexturing (or image compositing)
exist with out having something that blends them in hardware or software?

CitiZen X

 
 
 

3dfx patent interpretation...

Post by Samuel P. Uselt » Sat, 26 Sep 1998 04:00:00





> > CitiZen,

> > I looked carefully at the patent and to be honest I don't think they are
> > even trying to claim ownership of the methodology envolved. The patent
> > containes some pretty 3dfx chipset specific details which lead me to
> belive
> > that the primary purpose of this patent is not to protect any global
> > concepts such as texturemapping and blending but rather to pretect the
> > specific details of implementation on the 3dfx system.

>   Which specific details? I'm trying to find the hardware specific stuff,
> and it is very vague, with no refrence to the aparatus's form (I mean they
> are not supposed to hint around).

It is common, if not absolutely standard, for the lawyers to put into
the patent a series of claims, starting so broad that it is almost
certainly not defensible from a serious attack, each claim becoming
narrower and more specific.  That way, if the patent is denied or
challenged, and they "lose," they don't lose everything, just the
broader claims.

Quote:> > I very mutch doubt (my thoughts, I am not a legal expert) that a patent
> > worded like this could be used to enforce royalties on a different
> > textureing or multitexturing design so it should not be great concern.

> Actually, how can any texturing and multitexturing (or image compositing)
> exist with out having something that blends them in hardware or software?

It is truly amazing what lawyers and corporations will sue over.  And
many times these things are settled out of court.  In fact, one friend
of mine claims that *most* of the time, things are settled by
agreements in which licenses to use patented technologies are
*exchanged*.  Therefore, it becomes very important to have some of
these "chips" available to exchange when the situation arises.

As for a patent on this technology ...

I have not looked at the description myself, but the claim quoted in
this thread about blending multiple signals sounds so general as to be
invalid as demonstrated by the existance of prior work.  I can think
of two different areas immediately.  MIPmaps for rendering filtered
textures seems to me to include combining multiple textures.  And MUCH
earlier, the original work on bump mapping (Blinn - around 1978?)
points out that it can be used in conjunction with image based texture
mapping (originally published only a short time before that).  Again,
this would seem to fall into the category of blending the influence of
multiple signals for rendering a single surface.


 
 
 

3dfx patent interpretation...

Post by CitiZen » Sat, 26 Sep 1998 04:00:00


Oops, it wasn't about digital playback, it was about digital playback of a
living actor.
 
 
 

3dfx patent interpretation...

Post by Jeff Kesselm » Sun, 27 Sep 1998 04:00:00



Quote:>forgot) was used to claim the invention of digital video playback in
>software as well, and was persued through lawyers against 3D realms of
>Apogee software because there game Duke Nukem 3D used digital video
>playback interactively. If it works in this case, and frightens businesses

Um...  where pelase?

I was heavily invovled with this product from the side of maiking it
work on the internet and I can think of now video other then cut
scenes.

This sounds like one of those garbled net-rumor things...

JK

P.S.  Oh and on
"What's to keep 3dfx from trying to..."

Nothing. And nothinng should.  The foundation of oru legal systen is
that nothing rpevents ANYONE from TRYING to file and win a lawsuit.

If you ask the right question., "whats to prevent tehm from winning?"
The answer is "prior art".

 
 
 

3dfx patent interpretation...

Post by CitiZen » Sun, 27 Sep 1998 04:00:00


Quote:> This sounds like one of those garbled net-rumor things...

Yes, a net-rumor spread by apogee...
Um, no as a matter of a fact, Apogee had it on their web page, and a
protest page was linked to, as well as a place to download the patent. If
you worked on it, then you have not kept up with info on it because this
happened a while after its release. I still have the apogee web page, the
scans of that patent they uploaded, and an article that was in wired
magazine, I could send it to you if you want or still have problems with my
credibility on this?

Quote:> P.S.  Oh and on
> "What's to keep 3dfx from trying to..."
> Nothing. And nothinng should.  The foundation of oru legal systen is
> that nothing rpevents ANYONE from TRYING to file and win a lawsuit.

Wrong, sensibility. If the patent were worded better that would decrease
the chances that they would persue it at a future date (beyond the
provability of prior art), and if it was worded properly who ever they
persue would not be afraid to stand up to them.

Quote:> If you ask the right question., "whats to prevent tehm from winning?"
> The answer is "prior art".

Winning what? Are you talking about the lawsuit? If you are then the only
thing that could keep them from winning is if they can prove that there
patent doesn't cover Riva TNTs technology. Which by the fact that it is
baddly written could very well have been.

CitiZen X

 
 
 

3dfx patent interpretation...

Post by CitiZen » Sun, 27 Sep 1998 04:00:00


http://www.patents.ibm.com/details?patent_number=4662635
 
 
 

3dfx patent interpretation...

Post by Cory Bloy » Sun, 27 Sep 1998 04:00:00



> 3dfx's patent :
> http://www.patents.ibm.com/details?patent_number=5740343

> Their patent essentially claims they own the idea of having a device
> (apparatus) that combines textured pixel's (refered to as a signal) with
> any blend algorithm, as well as the method it uses to do it. It also
> mentions that one of the features is to be able to use it serialially, so
> that you can do triple, quadruple, etc. multitexturing.

> It sounds like a very lame (weak) patent. A texture-map is an image, and
> image blending has been around much longer then their filed claims. They
> laid claim to something merely because it has a diffrent name. If we
> refered to textures as images, they would not have a valid patent, how many
> apparatus's have been invented for blending (compositing) images. Keep in
> mind an apparatus can exist in software in the form of a function or
> algorithm as well as hardware.

> CitiZen X

If you want to see a really nutty one, referenced on the 3DFX patent
page is:
http://www.patents.ibm.com/details?&patent_number=5361386
Please correct me if I'm misinterpreting (it is late), but 5361386 looks
to me like Evans & Sutherland patented linear Gouraurd shading back in
`94...
 
 
 

3dfx patent interpretation...

Post by Kekoa Proudfo » Mon, 28 Sep 1998 04:00:00



Quote:> If you want to see a really nutty one, referenced on the 3DFX patent page
> is: http://www.patents.ibm.com/details?&patent_number=5361386 Please
> correct me if I'm misinterpreting (it is late), but 5361386 looks to me
> like Evans & Sutherland patented linear Gouraurd shading back in `94...

The patent you mentioned is not about Gouraud shading at all, it's about
interpolating parameters across a triangle.  And it's not nutty either.

There are many ways to interpolate parameters across a triangle, and the
use of barycentric coordinates is one of the ways.

-Kekoa