 News | Forum | People | FAQ | Links | Search | Register |
Log in |
|
|
found the links
http://photo.net/learn/technology/mflash/merl-non-photo http://www.merl.com/people/raskar/NprCamera/ Seems they haven't yet discovered the application I had in mind for it. now that I think more of it, it might not be so good for normalmaps, since it's more for discontinuity detection, but techniques might be tunable...
Lasers cost a lot more than humans, and they don't do 100% of the job, so you still need artists to 'clean up' the maps and make them tileable and they still have to match the diffuse maps and bla bla
Motion capture saves exactly 0 animator time. It works the same way. Heheh, look at any mud texture in almost any game. Go to Google images, type mud and there you've got the base of the texture.
Lunaran is right - an artist is always needed to modify whatever image to make it actually usable. Its the same with outsourcing, the stuff always needs a few weeks of someones time because management won't extend the contract to have it done properly. Remedy sending some of thier devteam to take photos around New York sounds like a tax loss situation - they spunk some cash on something effectively useless to drop thier tax rating. Ok, maybe not in that case, but I gauruntee that Remedy itself didn't pay. Heheh, look at any mud texture in almost any game. Go to Google images, type mud and there you've got the base of the texture.
Lunaran is right - an artist is always needed to modify whatever image to make it actually usable. Its the same with outsourcing, the stuff always needs a few weeks of someones time because management won't extend the contract to have it done properly. Remedy sending some of thier devteam to take photos around New York sounds like a tax loss situation - they spunk some cash on something effectively useless to drop thier tax rating. Ok, maybe not in that case, but I gauruntee that Remedy itself didn't pay. cleaning up can take a lot less time than creating from the raw photo. If the team knows what to look and has trained in the local environment before, it could be quite effective in asset "stealing" I imagine.
I also imagine it's quite time consuming in creating some normalmaps by hand, people model stuff in max and create it from that then etc... He specifically showed some floor texture for example: the photo showing the hotel corridor, then the face-on photo of the tiles, then the texture in the texture browser and finally the texture in-game (I must say that considerable vividity (?) loss in colors was observable along the process). Of course it was an old game and the world hadn't even heard of normal maps back then. I don't know of funding, and there might have been other reasons to go there too, like to get some feel for what the city is like and get inspiration and even possibly just for fun. Of motion capture, I assume there are still reasons for using it, like getting more realistic animations in some sense. Or maybe it's just a retarded trick used by EA for marketing reasons, to unfairly tread on small cute studios! cleaning up a normal map gotten by some hardware method would take less time than making one with an nvidia hackjob approximating tool from a picture and then tweaking and editing that ad infinitum to reach the equivalent quality... Now, of course if you're willing to do with less quality or realism, then...
the easy part is getting the basic shape of the objects down on paper (or screen :))and the more difficult part is getting the material surface of those shapes to appear realistic.
If you are using a laser to render a map of a surface of a real object, most of the work it is doing is the same thing that an artist would do in a few minutes. The subsurface levels a laser could fine tune are less important than the fidelity of the entire shape. Also, the majority of what you would want a normal map to capture is in the shape that you created whereas the material surface you add to it creates visual noise and this produces a less precise height map. http://www.katsbits.com/htm/tutorials/creating_bumpmaps_from_images.htm If youre doing a texture which is organic then you might as well get a computer to do it for you, but if you are doing a texture which is in-organic i.e. has lots of artificial-looking right angles and would have a much more precise 3D map, then you might as well do it yourself, because although it might take longer to do it manually, the end result will be better.
Or am a chipping into something I only half understand?!?! at http://www.quaddicted.com/?p=42
Thanks for the submissions! if you guys insist.
field-portable 3D laser scanners are still pretty goddamned expensive for something approaching the resolution you'd need. it reminds me of the advice Dustin Hoffman got when he was using sleep deprivation and exhaustion to get a grip on his role in Marathon Man,
"Why don't you just act." No lasers, learn to draw. No lasers, learn to draw.
not aimed at Lunaran, and I don't assume Bambuz can or cannot draw either. Just saying it would be far cheaper to train a half dozen people to draw assets then it would to buy a laser. My site needs a new map reviewer as Im really busy and Tronyn is also.
you will write reviews for the latest Q1SP releases and have them posted with screenshots at my website. If you wish to help out please email me at: uwfan1 [AT] gmail.com
|
|
| You must be logged in to post in this thread. |
|
| Website copyright © 2002-2025 John Fitzgibbons. All posts are copyright their respective authors. |