> I haven't met a low-end Pentium yet which is > capable of saturating 100MHz ethernet, even in > applications where disk IO is not part of the > equation.
Is this running Windows or Linux? If so, that is quite possibly why (although I would still expect a low end pentium to be able to saturate 100mbit with say static http requests that are cached in ram). Much as we all love linux, it is well proven that NetBSD, FreeBSD, and OpenBSD have a much more efficient TCP/IP stack, and thus are better on low end machines, and for some tasks high end machines. This is why I run NetBSD on my servers (well, for now. I'm tempted to set up Solaris on one machine for FDDI support). I still use linux on the desktop though (along with MacOS, one Windows machine, and hopefully soon an Irix machine or two).
Gah. If you are paying anywhere close to $500 for an indigo, you are being seriously ripped off. If the indigo had a good amount of ram (128+megs), a good processor (r4k, not r3k), and the elan graphics, and it came with the keyboard and mouse, it might be worth $80.
$500 will easily buy you a HighImpact or MaximumImpact machine, which significant money to spare. $500 will also buy you a very nice O2, again with money to spare. $500 is even just enough to buy a decent Octane with SI (which means no hardware texturing) and a decent amount of ram and harddrive space. If you are luck my might end up with a better octane for $500, but it isn't guaranteed.
The only way would be incapable of meeting his needs is if his needs are that he is a lamer who needs a create pretty, imaginative model button, or he needs to not use Windows (perhaps BillyG terminated his licenses and refuses to sell him any more), or he doesn't have any money.
In any of those cases, except the not able to use windows one, there is just no help. Max can do everything he'd want with a little programming (got to create the game engine as a plugin, which should be easy to do if he wrote the game engine in the first place) and a little inginuity (use multiple layers of vertex coloring, polygon coloring, and UV maps to store different pieces of information). And if he can't handle that, then he should stick to something simpler. There are still many great games waiting to be made that don't need any texturing at all.
Possiblly a modeling package more geared to hardware capabilities, or some way of adapting an existing modeler to make it more hardware friendly by blocking or modifying features that 3d hardware can't handle. It would seem such a package could be cheaper too, since it wouldn't have to support as many fancy features.
Both max and maya can be adapted to do some or all of what you want. It is fairly easy to build you game engine into Max or Maya so that it runs with the model in a preview window at all times. In the past year or two, Game Developer Magazine has run articles on doing both things. You might be able to find the articles at gamasutra.com (some articles from GDM end up being published there as well).
Along with this, you can add tools for carefully controlling the blending and hardware shading as data attached either by uv maps or stored per vertex.
As to fans and strips, it definately would be possible to make an exporter that would automatically do this, but not nescesarily in an optimized manner. I'm fairly confident that it would be possible to write plugins for both that add fans and strips as a new object type, but I'm not sure if it would be worth the effort. You also might be able to do it by storing extra information along with each triangle saying what strip or fan it belongs to.
The great thing about Max and Maya isn't their built in tools. It is how programmable they are. Thus they are best used when they are being used as the foundation for writing your own production pipeline you can do everything from integrate in your own asset management system, to adding modelling plugins to help, to storing all sorts additional information per vertex or per triangle (or per whatever object type), and then integrate in whatever animation tools you want, whatever special effects you need, what ever exporting, rendering, and post processing you might need. There really is pretty much no end to what can be done too those programs.
The z-buffer dump feature would be really usefull for people wanting to use povray in a more production oriented way.
Also, I used to find myself wishing it would write images using 32bit floating point TIFFs. I didn't see that it was added, but I haven't really had time to thoroughly examine all the new options.
At one time I used to wish that which object is shown at each pixel could be tagged, but then I hit on using the simplest rendering options and unique colors for each object as a seperate pass to generate that information.
Ahh, that brings back memories. I haven't used Povray in such a long time. A few years back I switched to BMRT, and now am looking at writing my own and switching again.
Ahh. A day to dream of. In a few years people will have come to agreement except a few lusers who want to know why that wierd ATA stuff isn't supported by new OSs.
According to the book, "The Art of Blue Sky", Blue Sky (the people who made Ice Age) use Rhino for their initial NURBS models. The process outline was that first the characters are modelled in clay. Then they are digitized into Rhino using a microscribe. They are split into pieces and modelled in Rhino, then they are taken to a proprietary tool they call suction cup that basically is along the lines of SoftImage's Zipper tool (ie, if blends two NURBS models to look like one).
Then the models are taken to Maya to be animated, where they have a thousand and one proprietary plugins and tools to help them.
OC-3, OC-12, OC-24, OC-48, and OC-192 can go 155.52Mbps, 622.08Mbps, 1.244Gbps, 2.488Gbps, and 9.952Gbps, respectively. So, there I shown you 5 links that can go faster than 12Mbps and further then a few kilometers, and all are serial.
Gear for those speeds is fairly expensive still, although OC-1 stuff can sometimes be affordable, assuming that you control everything for the kilometers that you want the connection run (as opposed to other people controlling property in between forcing you to go to the phone or cable company for the cabling run).
CCIR-601 specifies 720x525x60 for NTSC and 720x625x50 for PAL. For NTSC, only 486 lines are usually used despite the standard. [1]
720x480 is the resolution for DV.
While most DV cameras follow the DV standard carefully, very few people are really all that carefull about making sure that everyone interprets CCIR-601 and NTSC the same. Probably because it doesn't matter than much in the analog world.
Ironically, we are less sensitive to vertical information than we are to horizontal information,
I suspect that a bad safe combination is no more secure that a bad password. I know I don't use very good PIN numbers.
To make matters worse, cheaps safes are usually easy to crack. Expensive safes are large and unwieldy, and certainly not suitable for putting in my cubicle just to hold passwords that I forget. And having to run to leave work and run to the bank is also a fairly bad idea.
For systems that need security, I think the best plan is to force periodic password changes, disallow reusing old passwords, enforce a few simple rules on password creation (like the password can't appear in the dictionary), and then routinely copy the password file to a seperate machine and continually run password crackers against it. When they find something, have the user change their password, and if they don't, assume that along with the simple rules, a good password has been chosen.
I can't remeber ever seeing a CD that was 70 minutes long, except for mix CDs.
The last CD I saw that was over 60 minutes long was a Greatfull Dead Jam CD (official, not bootlegged, unfortunately). Even 50minutes isn't really all that common. Most CDs seem to be 40-50 minuts with about as many being only 30 minutes as there are 50+ minute discs.
Ahh, but grass hopper, just like one once could write data to audio cassete, we too can write data to vinyl. In theory one might be able to build a recordplayer that would work in a car. But the physical size makes like difficult.
That said, I still think CDs use superior technology. So, they don't sound as good as the good. If only they had gone with an analog standard then we could have had the best of both worlds.
Some of us have better things to do than memorize random (or psuedo random) strings. Give me two days of not using a password like that, and I'll forget it, just like I forgot the 1st 20 digits of pi when I stoped reciting it every hour.
> If you see a standard 35mm print of the film, > you're seeing a digital->analogue conversion > which willn not be as crisp and vibrant.
It is my understanding that for some time, feature films have edited digitally. Not like they digitized it, came up with an EDL, the used the EDL to cut the film, but rather that they digitize it, edit it, the print back to film.
I don't know what exactly was done for star wars. But a lot of digital work isn't done at the highest quality possible. I think that film is a much more flexible medium for the forseable future, but most people don't work in a way to take advantage of it. Starwars would have been one film that should have though.
That sounds to me like they are taking 4 IDE busses and 1 scsi bus, then claiming that IDE must be faster. Of course, since they are ATA100 drives, that means that there either they are using drives faster than their bus, or they are loosing about 133megs per second.
On the flip side, IDE is cheaper when you need a low end solution.
UPSs are cheap. I bought a cheap switched APC UPS from Staples for $40 that is good enough to keep a PC based server from resetting for a few minutes. Some hardware requires an active UPS (such as a Smart-UPS instead of a BackUPS), but PC hardware usually doesn't. I use the UPS on my file server, dhcp/boot server, and on my lanplex network switch, and it irons out a lot of the small resets that occur in my area.
"It's like shopping at JC Penney or Sears and the salespeople hound you EVERY 3 MINUTES. Thanks, but I already know how to shop and you just turned me off to your store."
I used to hate that, but I've come to appreciate it. When they offer to help, I ask them to show me the range of items that I'm interested in, list the benefits, downsides of them, and give them a chance to try to convince me why the next model up is worth the extra money.
I get in, I get out, and I don't have to stress myself out trying to summon the attention span to look over the shelves.
Now if only there would be computer stores around here with people clueful enough to be helpful.
If the monitor signal suffered so much, then you were doing something wrong. In the Sun and SGI worlds, cable runs of 50" require nothing more than a long set of cables, and will still look nice. A large part of this is the usage of the superior 13w3 connector (modern O2+s use a regular VGA connector, but the rest of the SGI lineup still uses 13w3). I wonder what would happen if just used adaptors on either end and ran a 13w3 cable the whole length. Or better yet, get a monitor with a 13w3 connector or BNC connects, and use it so that you only conver cable types once, immediately after leaving the machine.
Of course, it isn't any real solution for just using a proper machine in the first place.
What the heck. Shake and After Effects have nothing to do with Final Cut Pro. Those two programs are for special effects, compositing, rotoscoping, etc, while FCP is all about editing.
At school we have a robot arm. It is small. I think they were mainly made and sold for education purposes rather than industrial usage, based on the size and information video I've seen about it. Anway, in one of the videos, they show the gripper replaced with a suction cup (and a computer controller pump to provide the suction). With this attachment it would be pretty easy to do what you want.
The caviat is that supposedly this arm cost $20,000. I'd don't know what you would have to do to get a reasonable piece of equipment cheaper. If you used a vision system, you wouldn't need an arm that was quite a precise or stuck to it's calibration as well. But cheap arms (like the Radio Shack armitron line) aren't going to be as easy to get suction grippers for.
Their render farm consists of hundreds of 14 processor Suns. I shudder to think of the space, electricity, and cooling requiredments needed to replace that many highend suns with x86 boxes.
Suns on the otherhand typically use less electricity and produce less heat, and are just as good at FP tasks, and don't really cost too much more, especially since Pixar is probably getting a special deal. The only people who pay list for Suns (or any other major workstation really) are people who only buy one or two.
Racks of x86 hardware don't scale so well past a few hundred units without extra help.
For the same reason that SGI can ask $50k to $1 million (or more) for a machine and call it a workstation. Of course, these are machines that would kick the butt of many a server.
Slashdot Mirror
> I haven't met a low-end Pentium yet which is
> capable of saturating 100MHz ethernet, even in
> applications where disk IO is not part of the
> equation.
Is this running Windows or Linux? If so, that is quite possibly why (although I would still expect a low end pentium to be able to saturate 100mbit with say static http requests that are cached in ram). Much as we all love linux, it is well proven that NetBSD, FreeBSD, and OpenBSD have a much more efficient TCP/IP stack, and thus are better on low end machines, and for some tasks high end machines. This is why I run NetBSD on my servers (well, for now. I'm tempted to set up Solaris on one machine for FDDI support). I still use linux on the desktop though (along with MacOS, one Windows machine, and hopefully soon an Irix machine or two).
Real men use real emacs on win32, not that microemacs crap. Works well for me, never crashed, etc, etc.
Gah. If you are paying anywhere close to $500 for an indigo, you are being seriously ripped off. If the indigo had a good amount of ram (128+megs), a good processor (r4k, not r3k), and the elan graphics, and it came with the keyboard and mouse, it might be worth $80.
$500 will easily buy you a HighImpact or MaximumImpact machine, which significant money to spare. $500 will also buy you a very nice O2, again with money to spare. $500 is even just enough to buy a decent Octane with SI (which means no hardware texturing) and a decent amount of ram and harddrive space. If you are luck my might end up with a better octane for $500, but it isn't guaranteed.
The only way would be incapable of meeting his needs is if his needs are that he is a lamer who needs a create pretty, imaginative model button, or he needs to not use Windows (perhaps BillyG terminated his licenses and refuses to sell him any more), or he doesn't have any money.
In any of those cases, except the not able to use windows one, there is just no help. Max can do everything he'd want with a little programming (got to create the game engine as a plugin, which should be easy to do if he wrote the game engine in the first place) and a little inginuity (use multiple layers of vertex coloring, polygon coloring, and UV maps to store different pieces of information). And if he can't handle that, then he should stick to something simpler. There are still many great games waiting to be made that don't need any texturing at all.
Both max and maya can be adapted to do some or all of what you want. It is fairly easy to build you game engine into Max or Maya so that it runs with the model in a preview window at all times. In the past year or two, Game Developer Magazine has run articles on doing both things. You might be able to find the articles at gamasutra.com (some articles from GDM end up being published there as well).
Along with this, you can add tools for carefully controlling the blending and hardware shading as data attached either by uv maps or stored per vertex.
As to fans and strips, it definately would be possible to make an exporter that would automatically do this, but not nescesarily in an optimized manner. I'm fairly confident that it would be possible to write plugins for both that add fans and strips as a new object type, but I'm not sure if it would be worth the effort. You also might be able to do it by storing extra information along with each triangle saying what strip or fan it belongs to.
The great thing about Max and Maya isn't their built in tools. It is how programmable they are. Thus they are best used when they are being used as the foundation for writing your own production pipeline you can do everything from integrate in your own asset management system, to adding modelling plugins to help, to storing all sorts additional information per vertex or per triangle (or per whatever object type), and then integrate in whatever animation tools you want, whatever special effects you need, what ever exporting, rendering, and post processing you might need. There really is pretty much no end to what can be done too those programs.
The z-buffer dump feature would be really usefull for people wanting to use povray in a more production oriented way.
Also, I used to find myself wishing it would write images using 32bit floating point TIFFs. I didn't see that it was added, but I haven't really had time to thoroughly examine all the new options.
At one time I used to wish that which object is shown at each pixel could be tagged, but then I hit on using the simplest rendering options and unique colors for each object as a seperate pass to generate that information.
Ahh, that brings back memories. I haven't used Povray in such a long time. A few years back I switched to BMRT, and now am looking at writing my own and switching again.
I'm highly suspicious that he is never going to retire. I bet you he will keep going until he dies.
Ahh. A day to dream of. In a few years people will have come to agreement except a few lusers who want to know why that wierd ATA stuff isn't supported by new OSs.
According to the book, "The Art of Blue Sky", Blue Sky (the people who made Ice Age) use Rhino for their initial NURBS models. The process outline was that first the characters are modelled in clay. Then they are digitized into Rhino using a microscribe. They are split into pieces and modelled in Rhino, then they are taken to a proprietary tool they call suction cup that basically is along the lines of SoftImage's Zipper tool (ie, if blends two NURBS models to look like one).
Then the models are taken to Maya to be animated, where they have a thousand and one proprietary plugins and tools to help them.
OC-3, OC-12, OC-24, OC-48, and OC-192 can go 155.52Mbps, 622.08Mbps, 1.244Gbps, 2.488Gbps, and 9.952Gbps, respectively. So, there I shown you 5 links that can go faster than 12Mbps and further then a few kilometers, and all are serial.
Gear for those speeds is fairly expensive still, although OC-1 stuff can sometimes be affordable, assuming that you control everything for the kilometers that you want the connection run (as opposed to other people controlling property in between forcing you to go to the phone or cable company for the cabling run).
Except he really should have said:
"We do plan to fork() some child processes eventually..."
CCIR-601 specifies 720x525x60 for NTSC and 720x625x50 for PAL. For NTSC, only 486 lines are usually used despite the standard. [1]
e o_intro.htm
720x480 is the resolution for DV.
While most DV cameras follow the DV standard carefully, very few people are really all that carefull about making sure that everyone interprets CCIR-601 and NTSC the same. Probably because it doesn't matter than much in the analog world.
Ironically, we are less sensitive to vertical information than we are to horizontal information,
[1] http://www.disctronics.co.uk/technology/video/vid
I suspect that a bad safe combination is no more secure that a bad password. I know I don't use very good PIN numbers.
To make matters worse, cheaps safes are usually easy to crack. Expensive safes are large and unwieldy, and certainly not suitable for putting in my cubicle just to hold passwords that I forget. And having to run to leave work and run to the bank is also a fairly bad idea.
For systems that need security, I think the best plan is to force periodic password changes, disallow reusing old passwords, enforce a few simple rules on password creation (like the password can't appear in the dictionary), and then routinely copy the password file to a seperate machine and continually run password crackers against it. When they find something, have the user change their password, and if they don't, assume that along with the simple rules, a good password has been chosen.
I can't remeber ever seeing a CD that was 70
minutes long, except for mix CDs.
The last CD I saw that was over 60 minutes long was a Greatfull Dead Jam CD (official, not bootlegged, unfortunately). Even 50minutes isn't really all that common. Most CDs seem to be 40-50 minuts with about as many being only 30 minutes as there are 50+ minute discs.
Ahh, but grass hopper, just like one once could write data to audio cassete, we too can write data to vinyl. In theory one might be able to build a recordplayer that would work in a car. But the physical size makes like difficult.
That said, I still think CDs use superior technology. So, they don't sound as good as the good. If only they had gone with an analog standard then we could have had the best of both worlds.
Some of us have better things to do than memorize
random (or psuedo random) strings. Give me two days of not using a password like that, and I'll forget it, just like I forgot the 1st 20 digits of pi when I stoped reciting it every hour.
> If you see a standard 35mm print of the film,
> you're seeing a digital->analogue conversion
> which willn not be as crisp and vibrant.
It is my understanding that for some time, feature films have edited digitally. Not like they digitized it, came up with an EDL, the used the EDL to cut the film, but rather that they digitize it, edit it, the print back to film.
I don't know what exactly was done for star wars. But a lot of digital work isn't done at the highest quality possible. I think that film is a much more flexible medium for the forseable future, but most people don't work in a way to take advantage of it. Starwars would have been one film that should have though.
That sounds to me like they are taking 4 IDE busses and 1 scsi bus, then claiming that IDE must be faster. Of course, since they are ATA100 drives, that means that there either they are using drives faster than their bus, or they are loosing about 133megs per second.
On the flip side, IDE is cheaper when you need a low end solution.
UPSs are cheap. I bought a cheap switched APC UPS from Staples for $40 that is good enough to keep a PC based server from resetting for a few minutes. Some hardware requires an active UPS (such as a Smart-UPS instead of a BackUPS), but PC hardware usually doesn't. I use the UPS on my file server, dhcp/boot server, and on my lanplex network switch, and it irons out a lot of the small resets that occur in my area.
"It's like shopping at JC Penney or Sears and the salespeople hound you EVERY 3 MINUTES. Thanks, but I already know how to shop and you just turned me off to your store."
I used to hate that, but I've come to appreciate it. When they offer to help, I ask them to show me the range of items that I'm interested in, list the benefits, downsides of them, and give them a chance to try to convince me why the next model up is worth the extra money.
I get in, I get out, and I don't have to stress myself out trying to summon the attention span to look over the shelves.
Now if only there would be computer stores around here with people clueful enough to be helpful.
If the monitor signal suffered so much, then you were doing something wrong. In the Sun and SGI worlds, cable runs of 50" require nothing more than a long set of cables, and will still look nice. A large part of this is the usage of the superior 13w3 connector (modern O2+s use a regular VGA connector, but the rest of the SGI lineup still uses 13w3). I
wonder what would happen if just used adaptors on either end and ran a 13w3 cable the whole length. Or better yet, get a monitor with a 13w3 connector or BNC connects, and use it so that you only conver cable types once, immediately after leaving the machine.
Of course, it isn't any real solution for just using a proper machine in the first place.
What the heck. Shake and After Effects have nothing to do with Final Cut Pro. Those two programs are for special effects, compositing, rotoscoping, etc, while FCP is all about editing.
At school we have a robot arm. It is small. I think they were mainly made and sold for education purposes rather than industrial usage, based on the size and information video I've seen about it. Anway, in one of the videos, they show the gripper replaced with a suction cup (and a computer controller pump to provide the suction). With this attachment it would be pretty easy to do what you want.
The caviat is that supposedly this arm cost $20,000. I'd don't know what you would have to do to get a reasonable piece of equipment cheaper. If you used a vision system, you wouldn't need an arm that was quite a precise or stuck to it's calibration as well. But cheap arms (like the Radio Shack armitron line) aren't going to be as easy to get suction grippers for.
Their render farm consists of hundreds of 14 processor Suns. I shudder to think of the space, electricity, and cooling requiredments needed to
replace that many highend suns with x86 boxes.
Suns on the otherhand typically use less electricity and produce less heat, and are just as good at FP tasks, and don't really cost too much more, especially since Pixar is probably getting a special deal. The only people who pay list for Suns (or any other major workstation really) are people who only buy one or two.
Racks of x86 hardware don't scale so well past a few hundred units without extra help.
For the same reason that SGI can ask $50k to $1 million (or more) for a machine and call it a workstation. Of course, these are machines that would kick the butt of many a server.