Then, we'll take 10% of the winning bid price and give it to me for the discovery. If I want to commercialize it, then I'd better be the
highest bidder too (and end up paying myself that 10%, so I have a built-in discount of 10% if I win - fair credit to the inventor).
This process would then *strongly* encourage patent licensees to actually commercialize and use their patent, rather than acquire
them only to sqelch use. (I.e. if you can afford $5 billion and not use the technology, you'll probably suffer the consequences).
Ugh! Think about it -- currently, the inventor
could sit on the patent and "squelch use."
Conversely, if you invent it, you can keep it from
begin squelched.
With your scheme, it could just as easily be squelched, but it would be up to the highest
bidder, not the inventor. Hardly an improvement!
> >Linux can give 10-30% greater throughput on the
> >same hardware.
>
> I'm speaking about *pure* rendering - number crunching.
I'm afraid that this isn't true. Renderers are surprisingly more like operating systems than you might imagine at first.
The easy part of rendering is the math. The hard part -- and what sets the really good, heavy-duty renderers apart from the toys -- is all about system resource and memory management, shuffling large amounts of data around, network distribution, filesystem performance, and the amount of performance tuning that the OS will let you do. All of this is very much OS dependent.
Hi, "jd". Your reply indicates that you don't know what I meant by "artifacts". The things I'm thinking of don't go away with motion, frames per second, radiosity, or any of the other things you suggest. Feel free to email me directly or to ask on comp.graphics.rendering.renderman if you'd like more details.
Though completely off-topic of the DVD issue, "jd" said:
"Special effects - BMRT blows Renderman away, for the simple reason Renderman doesn't ray-trace. It only simulates. Partly because it's very expensive on the computer to apply raytracing and radiosity to every frame, in high quality. Open Source the frame data, and collaboratively render the CGI. You will end up with infinitely cooler graphics than ANY organisation (with the exception of MS) could EVER pay for out of it's own pockets, and at practically zero monetary cost."
As sole author of BMRT, one of the lead engineers of PhotoRealistic Renderman, and reigning chief architect of the RenderMan Interface Specification, I feel the need to set a few things straight here:
While I'm quite flattered by your opinions of BMRT, there's no possible metric by which BMRT can be said to "blow away" RenderMan (by which I assume you mean Pixar's PhotoRealistic RenderMan product, a.k.a. PRMan). Yes, BMRT does ray trace and PRMan does not, and although that may seem sexy to amateurs, those of us who make film visual effects for a living know that PRMan's scanline algorithm has a plethora of advantages over any ray tracer -- speed being one, but probably not the most important one. The reason we don't ray trace our films isn't because it's slow, or because we don't know how. It's because the scanline methods handle much larger and more complex geometric databases, have fewer image artifacts, and are much more flexibile at getting certain artistic effects.
Besides, BMRT isn't open source (in fact, the source isn't available under any conditions), and neither BMRT's raytracing nor PRMan's lack thereof has anything to do with their open versus proprietary status, so I'm not sure how this is relevent.
As for "open sourcing" the frame data and collaboratively rendering, it's obvious that anybody proposing that has *no* idea where the bottlenecks are in CGI production, or for that matter, any clue about just how much data and bandwidth are involved or what the basic process entails.
ObTopic: My take on the actual DVD issue is as follows. The music industry has seen many media formats for duplicating their property: cassettes, DATs, CD-RW's, MP3's. Each time they think it's the end of the world, but each time they survive and the music industry grows as the public's apetite for high quality music grows. The movie industry (the distribution industry, not the production industry) has the same knee-jerk reaction to new media, and again, they just keep making more money despite the predictions of doom. This DVD encryption issue, too, shall pass.
Slashdot Mirror
Ugh! Think about it -- currently, the inventor could sit on the patent and "squelch use." Conversely, if you invent it, you can keep it from begin squelched. With your scheme, it could just as easily be squelched, but it would be up to the highest bidder, not the inventor. Hardly an improvement!
I'm afraid that this isn't true. Renderers are surprisingly more like operating systems than you might imagine at first.
The easy part of rendering is the math. The hard part -- and what sets the really good, heavy-duty renderers apart from the toys -- is all about system resource and memory management, shuffling large amounts of data around, network distribution, filesystem performance, and the amount of performance tuning that the OS will let you do. All of this is very much OS dependent.
-- lg
-- Larry Gritz
"Special effects - BMRT blows Renderman away, for the simple reason Renderman doesn't ray-trace. It only simulates. Partly because it's very expensive on the computer to apply raytracing and radiosity to every frame, in high quality. Open Source the frame data, and collaboratively render the CGI. You will end up with infinitely cooler graphics than ANY organisation (with the exception of MS) could EVER pay for out of it's own pockets, and at practically zero monetary cost."
As sole author of BMRT, one of the lead engineers of PhotoRealistic Renderman, and reigning chief architect of the RenderMan Interface Specification, I feel the need to set a few things straight here:
While I'm quite flattered by your opinions of BMRT, there's no possible metric by which BMRT can be said to "blow away" RenderMan (by which I assume you mean Pixar's PhotoRealistic RenderMan product, a.k.a. PRMan). Yes, BMRT does ray trace and PRMan does not, and although that may seem sexy to amateurs, those of us who make film visual effects for a living know that PRMan's scanline algorithm has a plethora of advantages over any ray tracer -- speed being one, but probably not the most important one. The reason we don't ray trace our films isn't because it's slow, or because we don't know how. It's because the scanline methods handle much larger and more complex geometric databases, have fewer image artifacts, and are much more flexibile at getting certain artistic effects.
Besides, BMRT isn't open source (in fact, the source isn't available under any conditions), and neither BMRT's raytracing nor PRMan's lack thereof has anything to do with their open versus proprietary status, so I'm not sure how this is relevent.
As for "open sourcing" the frame data and collaboratively rendering, it's obvious that anybody proposing that has *no* idea where the bottlenecks are in CGI production, or for that matter, any clue about just how much data and bandwidth are involved or what the basic process entails.
ObTopic: My take on the actual DVD issue is as follows. The music industry has seen many media formats for duplicating their property: cassettes, DATs, CD-RW's, MP3's. Each time they think it's the end of the world, but each time they survive and the music industry grows as the public's apetite for high quality music grows. The movie industry (the distribution industry, not the production industry) has the same knee-jerk reaction to new media, and again, they just keep making more money despite the predictions of doom. This DVD encryption issue, too, shall pass.