3889 replies to this topic

[toxie]

MB stutters for me like hell..

But i'm on it, trying to find out why..

[bolt]

Thank you unclewilly for this fantastic MB WIP. Looks very very good.

Runs smoth at me, only one big stutter at the Monster Bush Multiball if I got a Freegame.

After the Freegameknock the table runs good again.

Don't forget the three lights for the Multipliers(D,I,G under the Mummy), does not work at this moment.

[unclewilly]

Thanks toxie. I only get a little but vsync smoothes it out. I know i still need to change a bunch of things over.
i also get wierd slowdown when the ball hitS the right wire ramp.

Strange cause its smooth in vp 991
Yeah im still converting it bolt. But thanks.

@toxie. I think in dt mode its a little smoother with the internal dmd/textbox

[fuzzel]

UW, I guess the slowdown on the wire ramp comes from the fact that the wire primitive is collidable. The primitive has a lot of polygons so checking the collision for so much polygons is quite expensive.

[unclewilly]

Ill see if i can decimate it a bit

[fuzzel]

I fear that won't help. The only real improvement is to use a normal hidden ramp instead and uncheck the collidable flag of the primitive.

[unclewilly]

Strange that there is not slowdown on the left ramp.

Good thing i left all the vp ramps in place

[chepas]

My wireform is ok for me but who knows on a different system. I know they've have acted differently before this. The ramp is fine, even when it comes off the ramp there's not too much funky spin, unlike previous, where it had trouble making it up and odd spin.

 

Any of my targets , drops & stand ups are using normal walls. There was too much air ball coming from the primitives with an array of different friction settings. The black rubber posts have been left as primitive collisions and not noticing a problem with them, but I still may end up changing them to std walls.

[toxie]

Could be we have to re-tweak the acceleration structure used for the collision detection. There are quite some heuristics in there that might "fail" with so many primitives.

 

btw: of course the hellish stutter doesn't occur on my "big" machine, only on my laptop.. :/  so maybe its as simple as my old rusty laptop swapping gfxmem all the time..

btw2: good news is that MB runs almost as fast as the default table (if both 4xAA and AO are enabled on both) on the non-laptop. i didn't expect that one. 

Edited by toxie, 03 February 2015 - 10:31 AM.

[Nidwin]

I did have some stuttering too past Sunday on one of the cleaned up material demo tables. I was although also running 2 high def graphical animations while testing the empty table for ball movement and stuff.

 

The strange part is when I hit the space bar the stuttering went away. Could just be a coincidence of course.

[blashyrk]

I just tested MB and I get crazy stuttering on the wire ramp, and ball stutter all the time, it's not smooth as with Vp9.9

Edit: Oh enabling vsync fixed the ball stuttering for a little while then i get like 5 fps

Edited by blashyrk, 03 February 2015 - 10:48 AM.

[unclewilly]

When i update my sig in a few days. Ill put the vp ramps back in and see if that smoothes things out a bit for people

[fuzzel]

I would expect stuttering on the left ramp too, but the poly count isn't that high because the ramp itself isn't as long as on the right side. The problem for wire ramps is that you always need more polys than for normals ramps otherwise the wires would look blocky. Someone could test it if the slowdowns are noticeable on normal primitive ramps if you increase the poly count by subdividing the mesh in Blender.

[chepas]

This is a long shot but fuzzel or anyone know about this error? Skip to 1:45

 

 

It's fine to just skip pushing enter or OK, but wondering if you know anything about it? It's fine as it is because I usually launch game with a script that deals with that for me.

Edited by chepas, 03 February 2015 - 11:50 AM.

[LadyLuck]

@unclewilly

 

Thank you for Monster Bash VP10. I tested and after only changing the physic of flippers I feel the game plays very nice!

[toxie]

This evening i'll do some changes to the acc structures, and then you guys can retest if that helps in comparison..

[unclewilly]

What were the flipper settings?

[toxie]

As for Monster Bash slowdowns: Seems like its a pretty simple reason actually: The steel rings in the primitive ramp to the right are pretty well tesselated, and as the collision routine handles the ball "temporarily" as a cube (for simplicity during the rough collision phase), each ball needs to collide with all these triangles it passes through. So as a workaround try to decrease the number of triangles for these rings.

[LadyLuck]

My flipper settings for Monster Bash are

 

 

I don't know, why pictures not possible, so the data in written form:

 

Mass 0,3

Strength 2200

Elasticity 0,6

Elasticity Falloff 0,43

Friction 0,35

Return strength 0,15

Coil Ramp Up 0

Edited by LadyLuck, 03 February 2015 - 01:00 PM.

[unclewilly]

As for Monster Bash slowdowns: Seems like its a pretty simple reason actually: The steel rings in the primitive ramp to the right are pretty well tesselated, and as the collision routine handles the ball "temporarily" as a cube (for simplicity during the rough collision phase), each ball needs to collide with all these triangles it passes through. So as a workaround try to decrease the number of triangles for these rings.

If its the rings i can do even better. Ill just redo the mesh without the rings. Then add seperate ring meshes in set to non collidable.

If not Much improvement ill just go back to non collidabe. Its easy enough just a few check boxes