You're mostly right but what's different with CG is that you can build it faster. What would have taken a model builder 1 year to manufacture can be done on a computer in a fraction of the time.
We're slowly moving towards the Avatar model of doing things even for romantic comedies. You don't need generators. You don't need street clearances. You don't condors. You don't need grips and gaffers and camera assistants. If the DP wants a big soft light source 6' behind the actor. Click, Click, bam. There it is. No getting another light out of the truck. No running cable. It's finished.
The quality isn't quite there yet for quick CG but the process is. I could model a room in 1/100th the time it would take 4 carpenters. And I could change the color of the paint in real-time. Once we make achieving photorealism a little bit easier it'll look as good as the one made by carpenters as well.
No need for constant make-up touch-ups. You have a makeup artist do it once and it stays exactly the way you want it between takes since it's just a texture.
We aren't more than 10 years away from the first non-fantasy photoreal movie imo.
The article cited an average 7 hour render time per frame.
7 Hours = 420 minutes = 25,200 seconds @ 24fps = 604,800x faster in order to render in real-time.
That even makes an enormous (and inaccurate assumption) that GPUs can handle PRMan quality sampling/rendering. It doesn't. Especially not at 100x faster than CPU speeds.
I find it amusing that half of the comments on this story have devolved into one of two diametrically opposite opinions which neither party find contradictory to one another:
1) "Films are looking all CG and crappy." 2) "So what if games are looking great the graphics don't matter."
Because you can never anticipate every situation you'll need to photograph.
Let's say you want a walk cycle of your character. Let's say a loop of about 1 second at 30fps. Now you have: 30 frames. Now you want your character to turn so you need it to rotate. Now you have to shoot about 1080 more angles. So now you're up to: 32,400 frames. Oops but now you also need to see this walking character from above or below. Let's say 200 degrees and assume nobody will ever see it from right below the ground. Ok now you are up to: 6,480,000 frames. Now let's say you want your character to walk but you also want to be able to rotate the torso of your character. Let's say you want 100* of rotation at 1* increments. Now you're up to: 648,000,000 frames. But you also need a dark and a light version for even the most crude of lighting views you're up to: 1,296,000,000 frames. But you also need a transition in and out of walking to standing still. That's 2 more animation cycles: 3,888,000,000
There you now have the equivalent of about 50 years of video to fit onto a disk somewhere AND it looks like crap since it's shot under a single lighting condition and doesn't fit into any environment. Also you'll never get good blending between animation cycles because you can't mix them. You can't have your character yawn while walking. Or aim both to the left and the right while walking... Every possible angle just multiplies on top of all the previous.
One interim solution is to instead of taking a 2D picture take a 3D picture. Instead of pixels you capture voxels. These voxels are in 3D space and then you can render your character from any angle. Trouble with voxels is the same as the trouble with your pictures though. If you just capture a single scenario you are then faced with game scenarios where you need the character doing something you never filmed. So then you have to animate your voxels. You also have to light your voxels since you need the character to look like they're in the right 3D space.
Before too long you're exactly back where you started. Some form of CGI which uses a 3D mesh or Voxel volume rigged to a skeleton lit by 3D lights and rendered by a 3D camera.
And then you have things like reflections. To really get a realistic scene you need reflections. Most of the modern world is reflective. If you take a picture from one angle you need the reflections to change based on your viewing angle, you can't bake everything onto a surface. So then you need ray tracing. Also not all objects in the world are just opaque you have something like skin which changes its hue and brightness based on backscattering and reflections so you need a high quality approximation.
What happens if your character jumps forward and then stops and lands. If they have a cloak or such on the cloak should have momentum. You anticipate every action your character will take so that cloak has to be simulated using physics.
Long story made only slightly shorter: yes you can just use video but you'll be extremely limited. For some applications it works spectacularly well but as soon as you want a third dimension or fluid animation which blends and can do a combination of things at once not pre-programmed it all falls apart.
It's also misleading because films can cheat. You can't see something from every angle and cameras don't always have to move through a space so a lot of what you see are flat cards carefully hand painted and positioned in 3D space.
In the end what really holds back video games is their memory. A small scene can consume in excess of 8GB of memory. That's fine on the CPU where you have a lot of RAM and you can swap back and forth from the HDD. With a GPU you have to load everything into memory which is extremely limiting.
Renderman which is one of the most popular renderers in feature film production is really a rasterizer with a raytracer slapped on top.
As long as games can't go through a post-process hand tuned by a team of artists for weeks they'll look inferior to something in which every frame is hand crafted. It's much harder to create a photoreal game than a photoreal movie.
Were you needing that memory for something else and when you did, did Windows 7 not give it up immediately?
I see these sorts of posts all the time and wonder what exactly it is that all these people want unused RAM for. I payed for it. I want it in use dammit! And unless you're on a notebook there is no reason to not have 4-8GB of RAM. Even DDR3 RAM is now less than $20 a gig. So what you're saying is no OS should use more than $5 of RAM?
The best line from a TV show on the subject in the last 10 years I think is from Battlestar Gallactica.
"There's a reason we separate military and the police: one fights the enemy of the state, the other serves and protects the people. When the military becomes both, then the enemies of the state tend to become the people."
This is why I always was nervous about the "war on terror". If it's a war then it's a civil war since extremists are also American citizens. The US Military an incredible effective fighting force. It's too easy in a 'global war on terror' for its sights to be turned onto itself. After all the US despite all the 'exceptionalism' is part of the globe. If terrorism knows no borders then that includes our own.
It's ok. Don't work to hard to understand the article because it's complete bullshit. I use flash all the time on my tablet PC using my finger. I've never ran into a problem. It would be a problem in a game where something follows your cursor but I personally find flash much more often used for video/interfaces etc.
If warming causes snow, and 1998 was the warmest year on record, why weren't there blizzards in 1998 on the east coast? I think the problem you have is that if warming can cause increased snow fall and reduced snow fall, then the same could be said about cooling. How do you differentiate between the two hypotheses? You mention Antarctica, which is interesting, since the claim there is that the ice is receding...but according to your theory, it should actually be expanding due to increased snow fall. Again, another "heads I win tails you lose" proposition. Still haven't seen a single warmer posit what evidence would falsify their hypothesis.
You just explained why freak storms AREN'T part of the evidence of global warming. It's a strawman to say that Global Warming will predict annual weather. Global Warming is based on GLOBAL averages over decades not whether or not it was sunny last Tuesday.
As to Antartic increased snow fall. I'm not even certain it's true, or that it's warmed enough to get anywhere near the prime temperatures or whether or not the wind patterns and water currents are conducive to increased snow fall. Weather is a fickle bitch. Changing predominant weather patterns can turn a forest into a desert and a desert into a rain-forest. So you're right Global warming can't be attributed to any one event or even locality. But even if my extreme broad over simplification of what increased snow fall due to global warming were true you still could see glacial loss. Let's say a region of Antartica previously got 2" of snow. Now let's say that it gets 2' of snow. Now let's it's so much warmer during the summer that it would love 3' of snow or 2' of snow and 6" of ice. You have a net loss but increased snow fall. By your reckoning NY should be under a growing glacier since they've seen so much extra snow this year. Considering Slashdot's pedantic level of mathematic correctness on most stories I can only assume that the abject incompetence of understanding subjects a 3 year old can understand to be intentional and willful stupidity.
"OMG it's snowing in my back yard! Proof that a statistical model of the entire earth with a warming trend is wrong!"
You know what was even better? Chronicles of Riddick.
Am I the only person who thought that Riddick had all of the qualities listed in the first post? It was different, original and entertaining. Pitch black was just an above average shooting gallery.
Is it ever too cold to snow? No, it can snow even at incredibly cold temperatures as long as there is some source of moisture and some way to lift or cool the air. It is true, however, that most heavy snowfalls occur with relatively warm air temperatures near the ground--typically -9 degrees Celsius (15 degrees Fahrenheit) or warmer--since air can hold more water vapor at warmer temperatures.
-National Snow and Ice Data Center (2002)
That sounds vaguely like a prediction. An accurate prediction 8 years ago.
Also I would like to add that even this year thanks to El-Nino we DID have a lot of snow. Many ski areas opened early. The trouble though is that all that extra snow fall thanks to El-Nino (which this year is more significant than any global warming) has since been reduced by the warm temperatures. What we had was more snow during the coldest weeks of winter but sustained melting there after.
I feel stupid even responding these idiotic claims but I'll do it anyway.
Last year the pacific northwest got record snow. Why? Because of a freak cold front which brought in far more arctic air. We normally get *rain* but the shift in weather patterns brought about unusually cold air. It's normally too warm to snow. Any extra warming will cause the temperature to rise well above freezing. This is why I've been able to go for bike rides in a T-Shirt and Shorts this winter.
Meanwhile you have the east coast. The east coast is subject to dramatically different temperatures during the winter. What is normally very very cold. Bone chilling cold warmed up enough to hit the sweet spot.
So yes. Warming can cause reduced snow fall AND increased snow fall based on what temperature the warming starts from. For instance Antartica should expect an INCREASE in snow fall since its winter temperatures are so far below freezing that very little moisture can stay in the air.
If the optimal snow fall range is -10C to 0C but anything over 0C results in rain then even a retarted chimpanzee can understand why warming can change the average temperature outside of the "snow range".
This principle is even known by school children without understanding the principle. "We're never going to have a snow day! It always rains when it's too warm and then is clear when it's cold enough!" I can't count how many times in my life I've lamented this fact. "If only all this rain was here yesterday when it was cold enough we would have a snow day!"
I don't know that it would make that much of a difference after all we internally handle values logarithmically. So I'm not certain $999 to $1020 actually does make much of a difference.
People switch phones when their contract expires. That's often every 2 years. It's almost a whole new race every 2 years. If you miss one go just wait for the next one.
Well if Microsoft is pissed then they should move back from "Nevada" where "They make and sell their software" and start paying taxes.
It's a common misconception that Microsoft is a Washington based company. I know all of the employees, offices and work being done in Seattle would seem to imply that they're based in Seattle but in reality they're a P.O. Box in Nevada.
Seattle is about 600k. The greater Seattle area is a lot more but we're just talking about city residents.
Also the recently elected Seattle Mayor made a campaign promise of pushing for a municipal fiber network. So Seattle could offer to negotiate some sort of deal which shares the costs with Google if they expand their scope. If hypothetically everything went perfectly and Google hit its maximum goal exclusively inside of Seattle then the city could follow up and fill out the last 100k for much less.
Regardless of Google's actions the city of Seattle is moving to roll out fiber this would just possibly make the effort easier.
Again though. Have you ever actually tried USING the service? With 10 users in a prime location it probably works great. My second hand experience with Wimax is that it's better than 3G but significantly worse than crappy DSL.
Have you actually tried Clearwire? I haven't tried them in the last year or so but every single person I know that's tried Clearwire (about 4 or 5) quit. It's better than 3G where it's available but it's not better than DSL.
Yeah Clearwire. I love Clearwire. I especially love the 250ms pings. Inconsistent bandwidth and largely unreliable uptime which seems to cut out every time a plane flies over--which in seattle means just about every plane landing at SeaTac it seams.
The office above us has 6 T1 lines. It costs them something like $1500 a month. We spend $200 on two of Comcast's 50/5mbps business connections and laugh. Why do they spend 10x what they should? The IT guy had Comcast at home and they pissed him off.
The way we see it is you can spend either $1500 for completely reliable internet. Or you can spend $200 for internet that is significantly faster and pay an extra $50 a month for our failover internet which is 6mbps DSL. We've had Comcast go out for a few minutes here and there every few months. After about 4 or 5 seconds the router switches to the DSL automatically.
Slashdot Mirror
You're mostly right but what's different with CG is that you can build it faster. What would have taken a model builder 1 year to manufacture can be done on a computer in a fraction of the time.
We're slowly moving towards the Avatar model of doing things even for romantic comedies. You don't need generators. You don't need street clearances. You don't condors. You don't need grips and gaffers and camera assistants. If the DP wants a big soft light source 6' behind the actor. Click, Click, bam. There it is. No getting another light out of the truck. No running cable. It's finished.
The quality isn't quite there yet for quick CG but the process is. I could model a room in 1/100th the time it would take 4 carpenters. And I could change the color of the paint in real-time. Once we make achieving photorealism a little bit easier it'll look as good as the one made by carpenters as well.
No need for constant make-up touch-ups. You have a makeup artist do it once and it stays exactly the way you want it between takes since it's just a texture.
We aren't more than 10 years away from the first non-fantasy photoreal movie imo.
100x or even 100,000x faster isn't fast enough.
The article cited an average 7 hour render time per frame.
7 Hours = 420 minutes = 25,200 seconds @ 24fps = 604,800x faster in order to render in real-time.
That even makes an enormous (and inaccurate assumption) that GPUs can handle PRMan quality sampling/rendering. It doesn't. Especially not at 100x faster than CPU speeds.
I find it amusing that half of the comments on this story have devolved into one of two diametrically opposite opinions which neither party find contradictory to one another:
1) "Films are looking all CG and crappy."
2) "So what if games are looking great the graphics don't matter."
Because you can never anticipate every situation you'll need to photograph.
Let's say you want a walk cycle of your character. Let's say a loop of about 1 second at 30fps. Now you have:
30 frames.
Now you want your character to turn so you need it to rotate. Now you have to shoot about 1080 more angles. So now you're up to:
32,400 frames.
Oops but now you also need to see this walking character from above or below. Let's say 200 degrees and assume nobody will ever see it from right below the ground. Ok now you are up to:
6,480,000 frames.
Now let's say you want your character to walk but you also want to be able to rotate the torso of your character. Let's say you want 100* of rotation at 1* increments. Now you're up to:
648,000,000 frames.
But you also need a dark and a light version for even the most crude of lighting views you're up to:
1,296,000,000 frames.
But you also need a transition in and out of walking to standing still. That's 2 more animation cycles:
3,888,000,000
There you now have the equivalent of about 50 years of video to fit onto a disk somewhere AND it looks like crap since it's shot under a single lighting condition and doesn't fit into any environment. Also you'll never get good blending between animation cycles because you can't mix them. You can't have your character yawn while walking. Or aim both to the left and the right while walking... Every possible angle just multiplies on top of all the previous.
One interim solution is to instead of taking a 2D picture take a 3D picture. Instead of pixels you capture voxels. These voxels are in 3D space and then you can render your character from any angle. Trouble with voxels is the same as the trouble with your pictures though. If you just capture a single scenario you are then faced with game scenarios where you need the character doing something you never filmed. So then you have to animate your voxels. You also have to light your voxels since you need the character to look like they're in the right 3D space.
Before too long you're exactly back where you started. Some form of CGI which uses a 3D mesh or Voxel volume rigged to a skeleton lit by 3D lights and rendered by a 3D camera.
And then you have things like reflections. To really get a realistic scene you need reflections. Most of the modern world is reflective. If you take a picture from one angle you need the reflections to change based on your viewing angle, you can't bake everything onto a surface. So then you need ray tracing. Also not all objects in the world are just opaque you have something like skin which changes its hue and brightness based on backscattering and reflections so you need a high quality approximation.
What happens if your character jumps forward and then stops and lands. If they have a cloak or such on the cloak should have momentum. You anticipate every action your character will take so that cloak has to be simulated using physics.
Long story made only slightly shorter: yes you can just use video but you'll be extremely limited. For some applications it works spectacularly well but as soon as you want a third dimension or fluid animation which blends and can do a combination of things at once not pre-programmed it all falls apart.
It's also misleading because films can cheat. You can't see something from every angle and cameras don't always have to move through a space so a lot of what you see are flat cards carefully hand painted and positioned in 3D space.
In the end what really holds back video games is their memory. A small scene can consume in excess of 8GB of memory. That's fine on the CPU where you have a lot of RAM and you can swap back and forth from the HDD. With a GPU you have to load everything into memory which is extremely limiting.
Renderman which is one of the most popular renderers in feature film production is really a rasterizer with a raytracer slapped on top.
As long as games can't go through a post-process hand tuned by a team of artists for weeks they'll look inferior to something in which every frame is hand crafted. It's much harder to create a photoreal game than a photoreal movie.
Were you needing that memory for something else and when you did, did Windows 7 not give it up immediately?
I see these sorts of posts all the time and wonder what exactly it is that all these people want unused RAM for. I payed for it. I want it in use dammit! And unless you're on a notebook there is no reason to not have 4-8GB of RAM. Even DDR3 RAM is now less than $20 a gig. So what you're saying is no OS should use more than $5 of RAM?
The best line from a TV show on the subject in the last 10 years I think is from Battlestar Gallactica.
"There's a reason we separate military and the police: one fights the enemy of the state, the other serves and protects the people. When the military becomes both, then the enemies of the state tend to become the people."
This is why I always was nervous about the "war on terror". If it's a war then it's a civil war since extremists are also American citizens. The US Military an incredible effective fighting force. It's too easy in a 'global war on terror' for its sights to be turned onto itself. After all the US despite all the 'exceptionalism' is part of the globe. If terrorism knows no borders then that includes our own.
It's ok. Don't work to hard to understand the article because it's complete bullshit. I use flash all the time on my tablet PC using my finger. I've never ran into a problem. It would be a problem in a game where something follows your cursor but I personally find flash much more often used for video/interfaces etc.
Wait. You had record cold winter? But this is an El-Nino year (above average temperatures). Do you reject the theory of El Nino too?
If warming causes snow, and 1998 was the warmest year on record, why weren't there blizzards in 1998 on the east coast? I think the problem you have is that if warming can cause increased snow fall and reduced snow fall, then the same could be said about cooling. How do you differentiate between the two hypotheses? You mention Antarctica, which is interesting, since the claim there is that the ice is receding...but according to your theory, it should actually be expanding due to increased snow fall. Again, another "heads I win tails you lose" proposition. Still haven't seen a single warmer posit what evidence would falsify their hypothesis.
You just explained why freak storms AREN'T part of the evidence of global warming. It's a strawman to say that Global Warming will predict annual weather. Global Warming is based on GLOBAL averages over decades not whether or not it was sunny last Tuesday.
As to Antartic increased snow fall. I'm not even certain it's true, or that it's warmed enough to get anywhere near the prime temperatures or whether or not the wind patterns and water currents are conducive to increased snow fall. Weather is a fickle bitch. Changing predominant weather patterns can turn a forest into a desert and a desert into a rain-forest. So you're right Global warming can't be attributed to any one event or even locality. But even if my extreme broad over simplification of what increased snow fall due to global warming were true you still could see glacial loss. Let's say a region of Antartica previously got 2" of snow. Now let's say that it gets 2' of snow. Now let's it's so much warmer during the summer that it would love 3' of snow or 2' of snow and 6" of ice. You have a net loss but increased snow fall. By your reckoning NY should be under a growing glacier since they've seen so much extra snow this year. Considering Slashdot's pedantic level of mathematic correctness on most stories I can only assume that the abject incompetence of understanding subjects a 3 year old can understand to be intentional and willful stupidity.
"OMG it's snowing in my back yard! Proof that a statistical model of the entire earth with a warming trend is wrong!"
You know what was even better? Chronicles of Riddick.
Am I the only person who thought that Riddick had all of the qualities listed in the first post? It was different, original and entertaining. Pitch black was just an above average shooting gallery.
Is it ever too cold to snow?
No, it can snow even at incredibly cold temperatures as long as there is some source of moisture and some way to lift or cool the air. It is true, however, that most heavy snowfalls occur with relatively warm air temperatures near the ground--typically -9 degrees Celsius (15 degrees Fahrenheit) or warmer--since air can hold more water vapor at warmer temperatures.
-National Snow and Ice Data Center (2002)
That sounds vaguely like a prediction. An accurate prediction 8 years ago.
Also I would like to add that even this year thanks to El-Nino we DID have a lot of snow. Many ski areas opened early. The trouble though is that all that extra snow fall thanks to El-Nino (which this year is more significant than any global warming) has since been reduced by the warm temperatures. What we had was more snow during the coldest weeks of winter but sustained melting there after.
I feel stupid even responding these idiotic claims but I'll do it anyway.
Last year the pacific northwest got record snow. Why? Because of a freak cold front which brought in far more arctic air. We normally get *rain* but the shift in weather patterns brought about unusually cold air. It's normally too warm to snow. Any extra warming will cause the temperature to rise well above freezing. This is why I've been able to go for bike rides in a T-Shirt and Shorts this winter.
Meanwhile you have the east coast. The east coast is subject to dramatically different temperatures during the winter. What is normally very very cold. Bone chilling cold warmed up enough to hit the sweet spot.
So yes. Warming can cause reduced snow fall AND increased snow fall based on what temperature the warming starts from. For instance Antartica should expect an INCREASE in snow fall since its winter temperatures are so far below freezing that very little moisture can stay in the air.
If the optimal snow fall range is -10C to 0C but anything over 0C results in rain then even a retarted chimpanzee can understand why warming can change the average temperature outside of the "snow range".
This principle is even known by school children without understanding the principle. "We're never going to have a snow day! It always rains when it's too warm and then is clear when it's cold enough!" I can't count how many times in my life I've lamented this fact. "If only all this rain was here yesterday when it was cold enough we would have a snow day!"
I don't know that it would make that much of a difference after all we internally handle values logarithmically. So I'm not certain $999 to $1020 actually does make much of a difference.
Yes. But HE doesn't find biker gangs more dangerous. He isn't a teenage girl. Being a selfish twit none of that concerns him.
People switch phones when their contract expires. That's often every 2 years. It's almost a whole new race every 2 years. If you miss one go just wait for the next one.
That would make more sense if Washington actually had income taxes.
Well if Microsoft is pissed then they should move back from "Nevada" where "They make and sell their software" and start paying taxes.
It's a common misconception that Microsoft is a Washington based company. I know all of the employees, offices and work being done in Seattle would seem to imply that they're based in Seattle but in reality they're a P.O. Box in Nevada.
Seattle is about 600k. The greater Seattle area is a lot more but we're just talking about city residents.
Also the recently elected Seattle Mayor made a campaign promise of pushing for a municipal fiber network. So Seattle could offer to negotiate some sort of deal which shares the costs with Google if they expand their scope. If hypothetically everything went perfectly and Google hit its maximum goal exclusively inside of Seattle then the city could follow up and fill out the last 100k for much less.
Regardless of Google's actions the city of Seattle is moving to roll out fiber this would just possibly make the effort easier.
Again though. Have you ever actually tried USING the service? With 10 users in a prime location it probably works great. My second hand experience with Wimax is that it's better than 3G but significantly worse than crappy DSL.
Have you actually tried Clearwire? I haven't tried them in the last year or so but every single person I know that's tried Clearwire (about 4 or 5) quit. It's better than 3G where it's available but it's not better than DSL.
Yeah Clearwire. I love Clearwire. I especially love the 250ms pings. Inconsistent bandwidth and largely unreliable uptime which seems to cut out every time a plane flies over--which in seattle means just about every plane landing at SeaTac it seams.
The office above us has 6 T1 lines. It costs them something like $1500 a month. We spend $200 on two of Comcast's 50/5mbps business connections and laugh. Why do they spend 10x what they should? The IT guy had Comcast at home and they pissed him off.
The way we see it is you can spend either $1500 for completely reliable internet. Or you can spend $200 for internet that is significantly faster and pay an extra $50 a month for our failover internet which is 6mbps DSL. We've had Comcast go out for a few minutes here and there every few months. After about 4 or 5 seconds the router switches to the DSL automatically.
As long as it's being sold by Google I wouldn't expect a lot of marketing. This is google we're talking about.