Page 6 of 6

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Mon Nov 07, 2016 11:59
by Graf Zahl
There's no blocker, I just wasn't sure it's finished.

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Mon Nov 07, 2016 18:37
by Nash
To my eyes, this is as "finished" as it can get at this point in time... with the exception of models, but I'm understanding that that has to come later, because the models' normals aren't being used yet so the model lighting stuff has to come first before model AO can be added... right?

No more bugs in the rounds of testing I've done and it's as good as it can look even when compared to modern games with a similar rendering feature.

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Mon Nov 07, 2016 19:22
by Rachael
I agree with Nash, but I guess dpJudas has to give the final word.

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Mon Nov 07, 2016 20:06
by Graf Zahl
I'll have a look later. First I need to do a little fixing on ZScript, as it turned out that casting non-constant strings to class types is totally borked.

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Fri Nov 18, 2016 13:27
by Nash
Can this be merged over the weekends? It's been added into QZDoom for a long time by now and I've been playing with it ever since... haven't noticed any problems. :D

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Fri Nov 18, 2016 13:46
by dpJudas
This one is currently blocked by Graf being busy with ZScript - I think. I wouldn't worry. It will be looked at once the workload lessens a bit. :)

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Fri Nov 18, 2016 14:51
by Graf Zahl
Nash wrote:Can this be merged over the weekends? It's been added into QZDoom for a long time by now and I've been playing with it ever since... haven't noticed any problems. :D

No. I need at least an entire day to properly review it, and that's what keeps it out for now.

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Thu Dec 22, 2016 12:07
by Graf Zahl
I finally found some time to test this a little.
Even with SSAO off it's still slower than without this code. Overlooking the bridge in Frozen Time I get 43 vs. 49 fps. With the various SSAO modes the frame rate will be between 36 and 33 fps but a slowdown here was to be expected. I'd still like to find out what causes the slowdown with SSAO off, though.

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Thu Dec 22, 2016 12:39
by Graf Zahl
Never mind, it was just one test with and one without dynamic lights being loaded. This thing looks ok, I think it can be merged.

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Thu Dec 22, 2016 18:37
by dpJudas
Excellent! About the earlier FPS drop that you found in the original testing round, that was because I changed the frame buffers to be textures. After you reported that, I changed it so that if SSAO is off it uses render buffers, and if its on it switches to textures.

By the way, since I know you got an upcoming release planned, if there's anything in GZDoom you'd like me to adjust for it please let me know.

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Thu Dec 22, 2016 18:47
by Graf Zahl
I have to admit that I lost track of it because I was sick last week. So the planned Christmas release won't happen, I first need to get an idea what's still needed to do.

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Fri Dec 23, 2016 5:34
by Nash
Thank you Graf Zahl and dpJudas! Does the implementation of this pave way for lighting for models, for after the planned release?

Re: Ambient occlusion and dynlight shaders with point light math

Posted: Fri Dec 23, 2016 10:56
by Graf Zahl
No, this has nothing to do with how to process dynamic lights.