27 replies to this topic

[toxie]

Thanks for the build koadic. I still see the haze in build 362 though. Honestly I'm not sure I'd run FSAA even if the haze was fixed, it's a subtle upgrade, but really cuts into performance in a noticeable way.

The haze problem was basically only fixed for the hyperpin startup case (plus one "bug" introduced with the new 9.1.4 that could bring the haze back during gameplay)..
As for performance: It will affect performance in the sense of bringing down insanely high frame numbers like in the 1XXX-2XXX range, but this won't change a thing gameplay wise..
Honestly i even go for enabled adaptive vsync nowadays (especially as it now works as expected with rev 362 ), as it provides much better/smoothed gameplay, too, IMHO, but brings down the framerates to "just" 60!
It affects the overall "precision" of the physics update code in theory though, but i didn't see any strange stuff happening from that..

[Slydog43]

What the No Need to 4Gig patch mean? I don't have to use program to patch VP anymore on win7x64?

[hmueck]

I don't have to use program to patch VP anymore on win7x64?

That is correct, but only when you use the new VPinball rev 362.

[thewool]

First off toxie and koadic - many thanks for the ongoing update work!!! I've just tried build 362 with the following results:

- the adaptive vsync seems to work great, it has a really noticeable effect of tearing. This is amazing how you have acheived this after all this time

- if I disable the fix, so in HP settings set useloadscreen 'false' then I get the haze when using FXAA. If I set it to 'true' I get no haze.

However.... this version definitely seems to improve things. For example, before when using the fix, if i viewed a flyer or instruction card during a game the haze appareard after I closed the flyer. Now I can do this no problem

Cheers again for the work on this

[toxie]

As said the FXAA haze fix is merely a fix to the hyperpin specific symptoms you described here, nothing more unfortunately..
So if FXAA didn't work for a specific setup before, it still won't now..

But at least for the average user that uses default hyperpin with an updated (default) fplauncher, FXAA now works flawlessly on WinXP.. Win7 will need the mentioned software implementation, i fear..

I don't have to use program to patch VP anymore on win7x64?

That is correct, but only when you use the new VPinball rev 362.

It's not fully correct, as there can still be tables (in theory) that would require it, but at least so far there is no existing table that needs it AFAIK..

[thewool]

I see what you mean toxie. In that case yes FXAA seems to be running on XP 'almost' flawlessley To me it makes a big difference to the overall look of a table.

...I say almost as I'm still having problems with Rosve's BTTF B2S version. Before it was the only table that would be 50/50 if it would launch with the haze or not, even with the fix. Now with the 362 build the haze seems to be there every time, but if I display a flyer and then close it again, the haze disappears. Weird... just glad you mananged to help with this.

Cheers again for the dev work toxie.

[toxie]

hmmm.. the main problem is that i still didn't fully understand where exactly the driver puts the FXAA code into the VP rendering pipeline, as i would have thought that it simply just does it on the final outputted picture, but instead it somehow is triggered elsewhere before in the pipeline, leading to these weird sideeffects.. i'll maybe dig deeper into this, especially as it's related to all the 2.5D blitting madness anyhow, and that i wanna redo at some point anyway as i think there is way too much work done on an average table nowadays, so potential speed improvements hiding there..

[Aurich]

Thanks for the build koadic. I still see the haze in build 362 though. Honestly I'm not sure I'd run FSAA even if the haze was fixed, it's a subtle upgrade, but really cuts into performance in a noticeable way.

Really Aurich??? I see next to no impact.

Yeah, it's hard to know how much stock to put in that 'framrate' counter you get with f11, but I see a serious impact on it just from turning on the FSAA. My system isn't a BEAST, but it's decent enough, i5 3.2GHz, GeForce GTX 560 Superclocked 2048, running recent drivers.

The haze problem was basically only fixed for the hyperpin startup case

Ah, hmm, in that case I may retract, I think I'm only seeing it with straight up VP, and not when I launch through Hyperpin. Which is fine really, I only truly care about Hyperpin launched games, so thanks. Still not sure about the performance impact, but maybe I'll mess with it some more. I just don't want to give up any smoothness of the ball.

hmmm.. the main problem is that i still didn't fully understand where exactly the driver puts the FXAA code into the VP rendering pipeline, as i would have thought that it simply just does it on the final outputted picture, but instead it somehow is triggered elsewhere before in the pipeline, leading to these weird sideeffects.. i'll maybe dig deeper into this, especially as it's related to all the 2.5D blitting madness anyhow, and that i wanna redo at some point anyway as i think there is way too much work done on an average table nowadays, so potential speed improvements hiding there..

By the way, and I'm sure I speak for many, thanks for taking the time to dig into this code, it's exciting to see improvement rolling out again.