Re: [Fenfire-dev] Libvob features needed for Loom

[Tuomas Lukka]

On Fri, Nov 21, 2003 at 04:32:52PM +0200, Benja Fallenstein wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Tuomas Lukka wrote:
> | On Fri, Nov 21, 2003 at 02:58:38PM +0200, Benja Fallenstein wrote:
> |>GL fog sounds just fine. I didn't know that it existed.
> |>
> |>The only quibble is that it uses depth... but I guess it can be made to
> |>work, it might just be a bit more difficult to program the view for it.
> |
> | Why? Usually depth correlates with fogging out just fine, i.e. the ones
> | that should go into background are deeper anyway.
> 
> Yes, but not necessarily linearly.
> 
> The major problem with getting depth to be equal to the desired fog
> level is this: I have the cs I want to translate from, as well as the
> polar coordinates for the translated cs, and I have the *absolute* depth
> I want to place the translated cs at -- well, actually, the depth
> relative to the cs *given to the view*. I don't have my parent cs's
> depth (relative to the cs given to the view). So it's hard to get the
> child cs to the right depth.
> 
> One possiblity would be a coordinate system type that takes the
> translation, rotation & c from one coordinate system, but the depth
> relative to another coordinate system?

Why not use the depth simply linearly and use the advanced fog mappings
if you want more control?

> Another problem is that depth is also needed to orchestrate connections,
> boxes, and box content: connections need to go under boxes, and box
> content needs to go above.

But those only need epsilon depth deltas, not macroscopic ones.

> |>Ok, what should the interface for this be?
> |
> | Somewhat similar to Irregu?
> | For awt, setting color through the RenderInfo objects instead of
> | Graphics.setColor()?
> 
> Can you go into more detail?

I.e. I'd like to keep this outside the VobScene objects for modularity.

A separate object that knows how to set up fogging for a given vobscene
would be good.

        Tuomas