"Liberty is always dangerous, but it is the safest thing we have."
- Harry Emerson Fosdick
More pages: 1 ... 11 ... 21 ... 31 ... 41 ... 51 ... 61 ... 71 ... 81 ... 91 ... 101 ... 111 ... 121 ... 131 ... 141 ... 151 ... 161 ... 171 ... 181 ... 191 ... 201 ... 211 ... 221 ... 231 ... 241 ... 251 252 253 254 255 256 257 258 259 260 261 ... 271 ... 281 ... 291 ... 301 ... 311 ... 321 ... 331 ... 341 ... 351 ... 361 ... 371 ... 381 ... 391 ... 401 ... 411 ... 421 ... 431 ... 438
Query FailedHumus
Friday, September 1, 2006

Well, if I had a camera with 90 degrees field of view that also produced square pictures and had no distortion, plus I had equipment to aim it perfectly in the six directions, then that would work. No, I only have a regular P&S camera with about 62 degrees FOV horizontally and a tripod, so I have to shoot a whole bunch of photos. Usually I do one center 360 row of 8 pictures, then 8 or sometimes just 6 pictures in a row above it and below it. Then 4 in the topmost row, and 3 down without getting the tripod legs in the view. Then finally I remove the tripod and shoot the last picture with it handheld leaning forward to avoid getting my feet in the view. Actually, I shoot a bunch of down pictures to hopefully get at least one that fits decently. Then I load it up in RealViz Stitcher and render a cubemap. Finally, I edit it a bit in photoshop if neccesary if there's any artifacts or ghosting.
Twixn
Thursday, August 31, 2006

I'd assume it would be from 6 photos
(up, down, and along the lines of North,
South, East and West).

But then im assuming this from knowledge of
cubemaps, not panoramic photos :P

-Twixn-
Chris
Wednesday, August 30, 2006

How have you made those cubemaps? How many photos did you use?
NeARAZ
Wednesday, August 23, 2006

Yeah, I see the point. The alternative to padding could be just rendering lightmap without it, and then do some kind of dilation filter on the GPU (that only spreads out to ontouched areas etc.). Anyways, a nice demo!
Humus
Tuesday, August 22, 2006

That's one reason. The other is that lightmaps will require some padding around each rectangle to avoid artifacts with a linear filter. With a position map this is very easy, the padded pixels can just have the same positions in the position map. Using the world space position it would be pretty tricky to do this padding. I'm not sure what would be faster though. The position map naturally consumes bandwidth, on the other hand I can render everything with a single primitive, whereas using the position directly would save the bandwidth, but need to render not only the triangles in the scene but also triangles to pad the lightmaps so the final count is triple that of the triangles in the scene, plus that that will add inefficiencies at rasterization.

Btw, I've fixed the resize bug.
NeARAZ
Tuesday, August 22, 2006

Why exactly the position map is needed? Can't you just compute world space position at vertices and pass that to the fragment? Or is it this "push out a bit" thing to avoid intersections?
Humus
Monday, August 21, 2006

Doh, forgot about device resets in DirectX. I'll fix that later.
sqrt[-1]
Monday, August 21, 2006

Opps, not dynamic lights - I meant the dynamic lightmap textures.
More pages: 1 ... 11 ... 21 ... 31 ... 41 ... 51 ... 61 ... 71 ... 81 ... 91 ... 101 ... 111 ... 121 ... 131 ... 141 ... 151 ... 161 ... 171 ... 181 ... 191 ... 201 ... 211 ... 221 ... 231 ... 241 ... 251 252 253 254 255 256 257 258 259 260 261 ... 271 ... 281 ... 291 ... 301 ... 311 ... 321 ... 331 ... 341 ... 351 ... 361 ... 371 ... 381 ... 391 ... 401 ... 411 ... 421 ... 431 ... 438