New demo

Humus

Crazy coder
Veteran
This time I'm showing soft coronas through occlusion queries.

softcoronas.jpg


http://esprit.campus.luth.se/~humus/
Enjoy :)
 
Very cool. I could see occlusion queries being quite useful in BSP+Portal type game engines. You're working on one of these, right? How's it coming along?
 
(Screenshot) is looking nice. (R9700P laying right next to me but keeps crashing my current PC, so I can't watch the demo until I get a new PC.)

As I'm too lazy to look in the sorce right now: From what frame do you get the occlusion info? Same, last or even older? Have you tried the different versions, and seen how it affects performance (and looks)?
 
I'm getting a request for "vorbis.dll". A quick google search doesn't yield links I have confidence in, and the vorbis page doesn't list a dll file except a winamp plugin (I'm pretty sure I had that installed already).

So...what simple thing did I miss? Should I scan the demo website some more for a file with that?

EDIT: Oh, libvorbis. That download page didn't parse very well in my brain, I guess. :oops:
 
I'm getting errors, like the corona lighting area effect is showing through at random points. Occlusion query failures? Could it be related to Basic's question?

EDIT: Looks like it is occuring at geometry edges...the effect has even appeared as a line once that I noticed...are those blocks on top of the columns supposed to be solid? They look so from underneath.
 
nice to see you released another great demo. but dam your advanced features.:( Need to upgrade to see all these shiny new demoes. any chance of more screenshots. :) Keep up the good work.

later,
 
Mintmaster said:
Very cool. I could see occlusion queries being quite useful in BSP+Portal type game engines. You're working on one of these, right? How's it coming along?

That project has been one of the most troublesome projects I have had. While pretty much every other project have gone from beginning to end without a hitch that project have been braked by driver bugs several times. So it has been just lying waiting for newer drivers over and over. Right now there's no driver problem braking it, but it has been laying untouched for quite some time anyway. Not sure when I'll pick it up again. There are other things I prioritize higher now.
 
Basic said:
As I'm too lazy to look in the sorce right now: From what frame do you get the occlusion info? Same, last or even older? Have you tried the different versions, and seen how it affects performance (and looks)?

It's from the same frame. I haven't tried squeezing out the maximum performance, so no frame-to-frame reuse of data, no async etc. I could use most of that, but performance was not what I looked for here.
 
demalion said:
I'm getting errors, like the corona lighting area effect is showing through at random points. Occlusion query failures? Could it be related to Basic's question?

EDIT: Looks like it is occuring at geometry edges...the effect has even appeared as a line once that I noticed...are those blocks on top of the columns supposed to be solid? They look so from underneath.

Do you have a screenshot?

And yes, everything should be solid.
 
Yes, I was just hoping you wouldn't make me overcome my laziness and upload it. ;)

I'm using an 8500, latest drivers.

Here it is

That is a mild form of the errors, and as the images progresses the pattern of the dots shifts erratically, but where they appear follows the outline of geometry edges as I said.
 
demalion said:
Yes, I was just hoping you wouldn't make me overcome my laziness and upload it. ;)

I'm using an 8500, latest drivers.

Here it is

That is a mild form of the errors, and as the images progresses the pattern of the dots shifts erratically, but where they appear follows the outline of geometry edges as I said.

Well at least you got your 8500 to run at all. Mine fails & the app says I have none of the OpenGL exstensions that Humus' lists on his download page. I am using DX9 with Omega's Catalyst 3.0 drivers. Anybody else with an 8500 getting this to run?
 
Runs fine, but gives me an 'invalid suffix on number' error and everything is transparent.

Ti4400 with 42.70 dets.
 
demalion said:
Yes, I was just hoping you wouldn't make me overcome my laziness and upload it. ;)

I'm using an 8500, latest drivers.

Here it is

That is a mild form of the errors, and as the images progresses the pattern of the dots shifts erratically, but where they appear follows the outline of geometry edges as I said.

Ah, I see. Blame the Unreal Editor which creates geometry which contains lots of T-junctions, which causes pixel leakage. I have written a piece of code that tries to remedy this, which for example fixed the pixel leakage in the shadows that rocks demo, but it doesn't seam to work here.
 
BoardBonobo said:
Runs fine, but gives me an 'invalid suffix on number' error and everything is transparent.

Ti4400 with 42.70 dets.

Hmm ... I had written "0.5f" instead of just "0.5" in the shader. Odd that ATi's driver accepts it anyway, I know the fragment shader does not accept suffixes. Anyway, I have uploaded a new one now. If you don't want to download it again you can just open Lighting.vsh and change the row
PARAM half = { 0.5f, 0.5f, 0.5f, 0.5f };
to
PARAM half = { 0.5, 0.5, 0.5, 0.5 };
 
sumdumyunguy said:
Well at least you got your 8500 to run at all. Mine fails & the app says I have none of the OpenGL exstensions that Humus' lists on his download page. I am using DX9 with Omega's Catalyst 3.0 drivers. Anybody else with an 8500 getting this to run?

Yep runs fine on my 8500 (apart from the pixel leakage)

It seems to be quite a strain on the core though. This is the first thing I've run thats produced artefacts on my card running at 315Mhz core :oops:
 
Back
Top