Had no internet last night, so bit of a monster post here.
Is the aim very random (eg, at shot that should go up the middle goes around the left loop)? or is it just not quite precise? I had the very random problem until I switched off one of the video options (something to do with sync, I don't have VP available in front of me right now).
Interesting in that context is also that actually there are two (not so known) settings that also control the physics 'randomness':
In the table itself (Global Difficulty Level: 0..1 0 switches off randomness (almost) completely) or in the registry ("GlobalDifficulty" 0..100 0 switches off again)
Just in case somebody wants to do some more testing with the physics, cause i actually was stumbled the first time i saw this in the code as this is not just influencing the table slope as i expected.
But it should be noted that if the scatter settings on the table and the flippers are 0, as they are with most tables, then there is no artificial randomness added to the flipper collisions, regardless of difficulty settings.
Our plan is to release a 9.2.1 asap. After that we try to migrate VP over to DX9 based on Mukuste's work. I don't know if we can maintain compatibility with VP9 tables but if that works out as expected we try to make a final VP9.3 version until we move over developing a fresh VP10 where we'll definitely break compatibility 
What specifically do you see or know of what may not work in the future versions of VP. I've seen members say reels will be a problem, which we have seen in the past with iterations of VP, but what else should we expect to run across as far as compatibility problems. I am asking for the sake of VP authors that have already released tables and am wondering how much work it will take to make older tables compliant with the newer versions of VP..
Image Reels should already work fine since Test2.
I think that we might release some sort of transition guide for the version changes.
mukuste
I've tested VP9_DX9_test3 on a number of tables, here is the result for LW3_Alpha_edition_FS_v1.1
under VPinball 9.2.1 rev921 I achieved a constant 60 FPS, however under VP9_DX9_test3 (in windowed mode - full screen) - I achieved 300 FPS constant.
Amazing, smooth as you like. Other tables were also much improved, excellent work!
I've tried following the thread here, but there are so many posts. Will the final version be supporting B2S as most of my backglasses are running on this server?
B2S should already work. Just make sure that you run in windowed fullscreen mode.
Yes, VP has a full simulation of angular velocity which is independent of linear velocity and which takes into account collisions and the inertial tensor.
As for the mass, yes, I was only talking collisions. Certainly it has an effect on the rolling speed. One of these days I have to do the math and see if changing the mass gives new effects which you couldn't also achieve by changing the playfield slope and the friction.
With the few tables I have made for VP9, gravity (mass) does have a different effect then that which can be recreated via friction and/or slope settings. I would argue that friction is an almost useless parameter as on a real well maintained table, it is virtually zero anyway. As for flipper physics I think they are fine with the correct setting each table, although perhaps too digital in nature and did wonder how many steps the flipper go through in a swing arc, could this number of steps be increased?
The flipper angle is simulated as a continuous value, there are no discrete increments (disregarding floating point accuracy of course, but that's no issue). It used to be that the flippers were rendered in discrete increments, but in DX9 I changed this too to be continuous. Some tables (SS, e.g.) do still use their own discrete flipper sprites, but this has no bearing on the physics.
I am surprised Clark is getting so much flack when he's not wrong. The aiming of the flippers leaves a bit to be desired and does take away from some of the fun of VP. Yes, you can get them close but it's a compromise between middle shots and outside shots. I've always heard that part of the problem is that the ball is actually a square in the physics engine and that adds to some of the challenge but I don't know if that is still true. It's loads better than FP but it still needs some work. I've been very impressed with all the work going into VP the last couple of months but I would by lying if I didn't say I was disappointed that this topic always gets brushed under the rug.
The ball is simulated as a ball, no squares there. I plan to get into the physics a bit for VP10; had a browse through the physics code last night and already saw some spots where small improvements could be made. I also wonder if VP correctly simulates the physics of a ball which is rolling vs sliding on the playfield, I think it does not and this would be interesting to try. This might be a way to improve the floaty ball physics that people sometimes complain about.
Edited by mukuste, 20 February 2014 - 09:10 AM.
Top
are all trademarks of VPFORUMS.