pouët.net

But never use a costable like the following:
http://www.clansofscotland.org/photos/burns/2004/pages/COSTable.html

chock:

I'm too lame to figure out which prog to use to view .mesh files. Blender doesn't like it, and directx mesh viewer thingie doesn't like them either... :( what format is this? Some random 3ds max thing? Help.

GbND: guess you'll have to use notepad for those.
.mesh is tha lobstah mesh format, and no
3d-app can boil it.

Quisten: Yeah, that was the first thing I tried... it made me think that it was a binary or binary compressed mesh. Sadly this is what they look like in notepad:

MESH1 _Б  À¬kDÀdú¿PM GcV<ÿÿÿÿ _Б  À¬kDÀU €¿ ÿÿÿÿ _БÂ
 @¬kDÀuÍ¿PM ßÕ =ÿÿÿÿ _БÂ
 @¬kDÀÍÌÌ3 €?àƒ«°ÿÿÿÿ ?¬‘  ÀÕ[Á©s¿PM îô…½ÿÿÿÿ ?¬‘  ÀÕ[Á)øM2 €¿gã3ÿÿÿÿ ?¬‘Â
 @Õ[Á~Í¿PM Ç ½ÿÿÿÿ ?¬‘Â
 @Õ[Áíâ˜3 €?§ûl³ÿÿÿÿ Ö‘  ÀŒâ@í~¿PM ôa»=ÿÿÿÿ Ö‘  ÀŒâ@‰$3 €¿¢3ÿÿÿÿ Ö‘Â
 @Œâ@±9~¿ïÖœ2vºð=ÿÿÿÿ Ö‘Â
 @Œâ@ƒÔ3 €?kO²ÿÿÿÿ

GbND: I actually meant the argh.png

If you additively blend this, I'm pretty much sure you'll get the resulting colors from the effect. I haven't tried it, though.

GbND: it's basicly just a binary dump of the d3d- vertex/index-buffers with a small header, and they are produced by my 3dsmax-exporter.

oh, and i really wonder how the 6dof-voxel in retroficial was done.. grrr... :(

it`s a raytraced heightmap.. bhuu :(

yeah, but how do you actually do that efficiently?

Yes, I am wondering how these kinds of voxels are running fast enough (while thinking of how the algorithm works). I'd like to try coding one alone and figure out myself. Quite impressive to see this in a middlepowered hardware..

Optimus did i mention twice its not about voxels?

the answer might be pretty simple but Im pretty knew to 3d coding and wanted to know how the overlay in
a deepness in the sky works. would you make it with a render with a bit noise on the vertizes position, or with a normal render of the scene rendered again on some vertizes and then put this render on the scene?

philipptr: it looks to me like they're simply distorting the x-coordinate with some function to y and then feedbacking...

The voxel tunnel in Stars: The Wonders of the World.

I would say if you have the code for the normal tunnel, you just find out which height the pixel you want to draw has and move it +(height/z*f) where you can find out z through the lookup table (x²+y²) and f depends on your second lookuptable. but that tunnel would have holes then (even if you draw the tunnel from far to near and always draw a line from the changed x,y to the x,y, without heightmap, there still would be holes I think.

I tried something like that years ago and yes. it had holes. I didn't spend much time on it though.

YES ! that voxel tunnel. I also thought of the same, and also dismissed it straight away as it would give you holes.

They DO something with a lookup table, but how exactly..

Also , the effect in deepness in the sky is also a questionmark. I tried different things like 1) median filter and hypnoglow 2) even that crazy scalex2 used in mame (after having posterized the image) and hypnoglow.. but didn't come much close. Maybe I have to watch it again.

The voxel tunnel in Stars: isn't it simply a tunnel using a voxel landscape as a texture for the tunnel ?

Navis : i remember ryg talking about mipmap-based blur in deepness in the sky on ex-flipcode.
here's a cache link. if it still works.

p01 : if the voxel landscape was simply a texture, then it would be flat. The difference is that this one was not !

A good way, maybe, to see how the effect works is to record it on AVI and then replay it again and again and look at the imaginary 'edges' that the 'nearest neighbour' type of texturing would create. In normal flat tunnels, these 'edges' are fixed - which means that the mapping is also fixed, and we only change the uv coordinates.

That would be a good start to determine what is happening in that respect I think. I'll do it whenever I find some time..

oh and thanks for the link

I'm no coder, but wouldn't that tunnel have been done by drawing lines from the texture coordinate towards the screen centre, length based on height map?

Also, I can think that rendering it as a flat surface without perspective (so the heights are drawn vertically up the screen) then converting to polar coordinates would have the same effect.

Ah, with the second method, you'd also want to scale the height offset to be smaller at the top of the image and bigger towards the bottom, or the mountains would get smaller as they got closer. And you'd want to fade towards the top too.

psonice: exact, and to obtain twisting one can simply add a variable x offset to the heightfield map depending on depth, wrapping to the other side of the map when needed (the heightmap has to be tileable along x).
Then the polar mapping does the magic... and what a magic =D

navis are you a coder?