Geometry Shader based Motion Blur Patent Application

[Jawed]

Is this novel? I'm kinda struggling to see how this could be patented.

System and method for creating motion blur

An embedded, programmable motion blur system and method is described herein. Embodiments include applying displacement on a vertex level and amplification on a primitive level within a graphics pipeline. Embodiments include receiving a plurality of vertices in a first component of a graphics processing unit (GPU), displacing each of the plurality of vertices, receiving a primitive that includes at least one of the displaced vertices in a second component of the GPU, and transforming the primitive. In one embodiment, transforming comprises at least one of translation, scaling, and rotation. Further included are generating a plurality of primitive samples over a time aperture, and outputting the plurality of primitive samples to further components of the GPU for further processing for display of the scene with motion blur.

It just seems to me that GS in D3D10 (or OGL) supports the data types and provides the performance to make this technique viable.

Jawed

 

[nAo]

one word: slow

 

[K.I.L.E.R]

Why are things like this allowed to be patented?

 

[Cal]

I think Lost Planet used the same algorithm to do the motion blur, which was based on a published paper, albeit without GS.

 

[nAo]

I think Lost Planet used the same algorithm to do the motion blur, which was based on a published paper, albeit without GS.

No, LP uses a completely different algorithm

 

[CouldntResist]

Since this description is barely comprehensible to me, I'm not sure how this technique makes motion blur...

...unless it's actually just emulation of 3dfx' T-buffer, done with help of GS.

 

[Npl]

Since this description is barely comprehensible to me, I'm not sure how this technique makes motion blur...

...unless it's actually just emulation of 3dfx' T-buffer, done with help of GS.

I undertand this as horribly complicated but generic explaination of a very obvious technique:
Take vertices of an object, sample multiple positions of those vertices in a timespan and process them further (like just rendering the object multiple times semitransparent).

Wouldnt be the first damn obvious patent.

 

[silent_guy]

Why are things like this allowed to be patented?

You must be new to the world of patents...

 

[NocturnDragon]

You must be new to the world of patents...

Well since when are software patents allowed in europe? Or is it an hardware one?

 

[Simon F]

Well since when are software patents allowed in europe? Or is it an hardware one?

Software patents are allowed in Europe but there are restrictions on what is patentable.
The crazy "one-click" patent in the states, AFAICS, would not qualify.

 

[CouldntResist]

This patent is stupider than I have dared to imagine...

The description simply stops at the stage when the geometry is amplified. It doesn't propose how the samples are going to be combined. There is no motion blur part. Not to mention, a blur in a way "completely orthogonal to other graphics features". This patent doesn't provide a solution, just claims rights to a potential one, on the basis of using GS for geometry amplification.

Or is it an hardware one?

Dunno, it does mention "quantum devices".

 

[zsouthboy]

I hate the patent system.

 

[Simon F]

This patent is stupider than I have dared to imagine...

Isn't it still just a patent application? It's not actually been granted.

The description simply stops at the stage when the geometry is amplified. It doesn't propose how the samples are going to be combined.

Agreed, that aspect all seems quite obvious. I'd considered proposing something very similar some years ago but just using fixed function hardware.

 

[[maven]]

I've just heard what sort of money you earn as a "newbie" in the European Patent Office.
For that sort of money, there are still way too many obvious patents being granted...

 

[Simon F]

I've just heard what sort of money you earn as a "newbie" in the European Patent Office.
For that sort of money, there are still way too many obvious patents being granted...

Can you define "that sort of money"? I'm curious. (Not that I want to change jobs. Writing patents is tedious, but reading patents is dulll)

 

[zsouthboy]

It doesn't sound like a hard job though - you just get a big rubber stamp with the word "APPROVED" on it and a stack of applications.

If they use big words in describing something, you are to immediately stamp it.

If its an obvious thing they're trying to patent it, you are expected to stamp it even quicker!

 

[Simon F]

It doesn't sound like a hard job though - you just get a big rubber stamp with the word "APPROVED" on it and a stack of applications.

If they use big words in describing something, you are to immediately stamp it.

If its an obvious thing they're trying to patent it, you are expected to stamp it even quicker!

Here are the words if someone who has clearly never tried to get a patent granted in the EPO and definitely never tried in Japan!

 

[zsouthboy]

How did you know?

 

[silent_guy]

Here are the words if someone who has clearly never tried to get a patent granted in the EPO and definitely never tried in Japan!

I don't know about Japan, but you're right about the EPO: in general, they are indeed way more rigorous when analyzing a patent than in the US.

 

[soylent]

There's one way these kinds of frivolous patents can be stopped IMO, what I'd love to see is an anti-patent.

* Application cost taken from public money(anyone interested in the public good can apply for an anti-patent freely).
* The anti-patent grants free use to anyone and prevents patents that overlap the patent.
* Any significant discovery or invention of a publically funded project must be anti-patented.

This allows a means to fight back against moronic patents. This allows pre-emption of simple patents without producing expensive and time consuming "prior art" that can be used to void a patent that has been granted or prevent a new patent.