2 coding ideas
category: general [glöplog]
First:
I've seen the wii remote and it rules. What I've thought is... could it be possible to do something similar with a webcam? I mean, getting the webcam in your hands, and, for example, some reference points in the screen (3 for example, full R, full G and full B in the squares of the screen) so when the camera get the points you would be able to calc position, distance, and rotations. I've already webcam support in my demosystem, so maybe I try one day.
Second:
Using something similar to voxlap for voxel volume rendering (not surface only, but all the volume). What voxlap do is near the same of a wolfenstein3d-doom engine, with a supercovering and subpixel accuracy raycaster. So, the lines are generated vertical always, but later a proyection is used (like when you use a cubic mapping). Well, doing the same with volume rendering would be great (not using a supercovering, but in a similar way), because storing vertical voxel lines in a correct way, memory access would be lineal the most of the time, and rendering in a vertical line would be cached, so it would be really fast. Usually raycasting a volume voxel is slow as hell because it is usually done in a non-linear way, but this way it could be as fast as the faster existing method wich is only for parallel representations, but with perspective. I think I'm going to code in these days...
Thats all.
I've seen the wii remote and it rules. What I've thought is... could it be possible to do something similar with a webcam? I mean, getting the webcam in your hands, and, for example, some reference points in the screen (3 for example, full R, full G and full B in the squares of the screen) so when the camera get the points you would be able to calc position, distance, and rotations. I've already webcam support in my demosystem, so maybe I try one day.
Second:
Using something similar to voxlap for voxel volume rendering (not surface only, but all the volume). What voxlap do is near the same of a wolfenstein3d-doom engine, with a supercovering and subpixel accuracy raycaster. So, the lines are generated vertical always, but later a proyection is used (like when you use a cubic mapping). Well, doing the same with volume rendering would be great (not using a supercovering, but in a similar way), because storing vertical voxel lines in a correct way, memory access would be lineal the most of the time, and rendering in a vertical line would be cached, so it would be really fast. Usually raycasting a volume voxel is slow as hell because it is usually done in a non-linear way, but this way it could be as fast as the faster existing method wich is only for parallel representations, but with perspective. I think I'm going to code in these days...
Thats all.
write an article to you random diskmag or i will not believe you
First) Yes it is. Webcam has retarded resolutions however but the technique is not new nor hard to implement. However to make it work in 3d you will have to have plenty of cameras and a clean environment.
Second) I just realized that youre one of those active in the MENSA thread. I think its rather pointless to go on writing to you.
Second) I just realized that youre one of those active in the MENSA thread. I think its rather pointless to go on writing to you.
telling about ideas here, are completely useless. If you can't figure that out, you should spend more time reading this bbs.
How does the Wii controller work anyway?
Quisten, not that useless... the most of the times you get a BASS thread, but sometimes it gets really useful. And if it is not the case, at least it is fun :)
Shut up and code!
howzabout raytracing that uses precalc'd voxel grids for faster omission of rays that don't intersect with anything? on second thought, it's probably already been done. sucks when you grow up in a time where all your good ideas have been snatched away a few years earlier :(
- noouch: Yes, it has been done already. Anyway, traversing a voxel is slow, and there are better structures for the boundings. When I say "slow" I mean "slow for realtime".
- linde: it has gyroscopes inside.
- broderick: I don't want a high precission position, just to do something for fun, so no camera calibration needed.
- photoshop: I just want to calc an aproximation of the position of the camera by capturing an image of the screen (maybe with lights off to make the job easier). So, only 1 camera needed, one webcam in your hand pointing to the screen.
- linde: it has gyroscopes inside.
- broderick: I don't want a high precission position, just to do something for fun, so no camera calibration needed.
- photoshop: I just want to calc an aproximation of the position of the camera by capturing an image of the screen (maybe with lights off to make the job easier). So, only 1 camera needed, one webcam in your hand pointing to the screen.
but why would you need to have a webcam in your hand? lots of games before wii were doing the same thing with a webcam over the display. you didn't need to have anything in your hand, just move your hand and the webcam sees what you're doing. before the game you have a calibration when you point to the four corners, and then you just control the game by moving your hand/body/whatever. if there's a static background (like, not playing this at an airport;) things should work pretty well.
I would like to be enlightened by the ways of fast raytracing...
the wiimote uses a 3-axis accelerometer and a IR sensor to do triangulation with the 2 IR light sources from the badly called "sensor bar".
el-bee: glovePIE is so efficient and easily scriptable :)
uhuh "NOTE! You may not use this software on military bases, or for military purposes, or in Israel (which amounts to the same thing). Violation of the license agreement will be prosecuted."
uhuh "NOTE! You may not use this software on military bases, or for military purposes, or in Israel (which amounts to the same thing). Violation of the license agreement will be prosecuted."
Ger: I don't want to capture the motion of moving objects, I want to do the inverse, to capture the 3d position and rotation of the webcam when it captures 3 static points in my screen. Why? Because it is much easier than capturing motion, and because I've already done it (two eyes in the screen that followed me). Well, I will try to do it, let see how it works, if it is cool I will post it.