GoldQuake : Improved WinQuake

[Dr. Shadowborg]

The way DirectQ does it is actually by checking for movement of triangles between frame 0 and frame 1; if they move beyond a certain threshold value then those triangles only are not interpolated (other triangles in the model are). It's called from R_DrawViewModel so we can be certain that the model checked actually is a viewmodel (using models beginning with v_ is not valid as even an ID1 model that ain't a viewmodel begins with v_), and is set up so that it only gets called once per model.

Overall a little hacky but it works fine in every test case so far.

Interesting, does this work on models that have multiple variations of muzzleflash? ex. assault guns with a rapid fire frames and grenade launcher frames?

[mh]

The way DirectQ does it is actually by checking for movement of triangles between frame 0 and frame 1; if they move beyond a certain threshold value then those triangles only are not interpolated (other triangles in the model are). It's called from R_DrawViewModel so we can be certain that the model checked actually is a viewmodel (using models beginning with v_ is not valid as even an ID1 model that ain't a viewmodel begins with v_), and is set up so that it only gets called once per model.

Overall a little hacky but it works fine in every test case so far.

Interesting, does this work on models that have multiple variations of muzzleflash? ex. assault guns with a rapid fire frames and grenade launcher frames?

It should in principle because it's completely agnostic towards the model. Everything is done on a triangle-by-triangle basis (actually vertex by vertex, but they need to come from triangles), only back-to-front movement is checked (I forgot that bit in my initial explanation), and each result is totally in isolation from every other one.

The gotcha here is that it only checks frame 0 against frame 1, so it's obviously dependent on the data in those two frames being reasonably representative of the full thing. It should be easy enough to extend to check frame 0 vs all frames, accumulate a count of passes and fails, then fudge something based on that count.

Worst case is that it's either going to be no different from if it wasn't done, or stuff with large movement between the frames that's not a muzzleflash won't be interpolated (so no different than GLQuake).

The obvious correct solution is to build the model with interpolation in mind, of course.

[Sajt]

I've noted down all your requests!

By the way, I've named the engine "GoldQuake", which is short for "Good Old Quake".

[mh]

Excellent! I'm really looking forward to seeing what comes out of this.

[Sajt]

And the SVN is now up at http://svn.icculus.org/qshed/trunk/goldquake!

[Urre]

Urre approves.

[Dr. Shadowborg]

Groovilicious

[Baker]

Quake hasn't seen the amount engine development going on right now in years.

[metlslime]

The EF_MUZZLEFLASH trick is the way Fitzquake does it, actually. Works unless the modder decided not to use that effect (i think one of the Quoth enforcer variants doesn't have a muzzleflash for one of its attacks...)

[dreadlorde]

Svn? *shudders*

[Baker]

Cvar_SetValue ("r_maxedges", (float)100000); //NUMSTACKEDGES
Cvar_SetValue ("r_maxsurfs", (float)100000); //NUMSTACKSURFACES

aguirRe's software wizardry to render areas with a lot of surfaces.

Also: no big deal but in r_drawentities there is a minor bug in original WinQuake. It may be as simple as the fact it draws the viewmodel when it not supposed to. Been a long time.

[Sajt]

If you think that r_drawentities should not draw viewmodels (and DarkPlaces seems to agree with you), I can do that.

[Error]

you could ask FrikaC if he has the source for FrikQuake. it was a upgraded winquake engine. had frik_file and model alpha, water alpha, etc.

[Sajt]

FrikQuake...now there's a name I haven't heard in a while. I don't think I ever got my hands on it.

[ceriux]

i wish i could have.