616 replies to this topic

[Patrick2610]

Fix the DT blank display for Windows 10 with this update please...

[fuzzel]

Thanks to all of you for your continued development of VP! I have one question does 10.3 now have the fix for the Metallica table?

Sent from my SAMSUNG-SM-G935A using Tapatalk

What fix do you mean?

[sliderpoint]

It was a bug found in a certain vpinmame developement thread. Toxie did that fix. He would know if it's in or not. But, I would guess 'yes'?

 

-Mike

[fuzzel]

VPM or VPX? If it's VPM then the bug is not fixed in VPX

[sliderpoint]

Nope, it was a VPX bug.  

[kiwi]

The table is automatically saved even with the checkbox "Auto-Save" disabled.

[Mark70]

Would it be possible to implement a "volume control" feature to the sound manager, which would let the user adjust the volume for every single sound?

For example, one could increase the volume for a ramp-rolling sound if it is too quit on a prticular table ... and so on.


And to all involved, thanks for the developing!

 

[Ben Logan]

About six months ago, we had some talk about an experimental ball physics development fork. Any interest in pursuing that further, Toxie and Fuzzel? I know you guys have a lot on your plate already.

[toxie]

Fix the DT blank display for Windows 10 with this update please...

 

??? Details?

It was a bug found in a certain vpinmame developement thread. Toxie did that fix. He would know if it's in or not. But, I would guess 'yes'?

 

-Mike

 

If i said that i did the fix back then already, then it is in 10.3 for sure (i do not even remember myself what i fixed there though ).

About six months ago, we had some talk about an experimental ball physics development fork. Any interest in pursuing that further, Toxie and Fuzzel? I know you guys have a lot on your plate already.

 

In principle, yes, and there is still a define in the code (search for NEW_PHYSICS, if someone else wants to have a look) that can remove some of the rather old hacks and heuristics, but this then leads to some subtle (or not-so-subtle if you ask others) changes in the physics behavior. I also have some more experimental code on my HDD that tries to improve some more of the collision code, but this is very unfinished and hacky (and most likely i do not understand most of it anymore nowadays ).

But as i will not have that much time on my hands in the next months, its very unlikely that this will happen in the 10.3 timeframe (unless somebody else wants to take over, see above).

Would it be possible to implement a "volume control" feature to the sound manager, which would let the user adjust the volume for every single sound?

For example, one could increase the volume for a ramp-rolling sound if it is too quit on a prticular table ... and so on.


And to all involved, thanks for the developing!

 

As the script has full control over the volume (as this is playing the sounds and not the VPX code), it's not that easy.  :/

[Thalamus]

@toxie : Mike is talking about this ... which DJRobX found ..
In kdtree.cpp there is a member variable m_org_idx.   It is only cleared if the vector size grows, which it rarely will since the number of total balls is relatively constant.    Later, in FillFromVector, the idx keeps getting added to every time a ball is created or destroyed.    I don't think we want that!   After changing this I haven't been able to reproduce a crash (but see previous comments, this one is a timing sensitive bugger).   I also don't claim to be an expert in this code, I'm still trying to understand what all is going on.   I think the bug triggers when a ball goes to the outer loop because the ball Y rect goes below 0 (-9 in every crash I've seen), which I think is making the invalid entries contenders for a hit test.

[toxie]

ah, okay, thanks.. yup, this is fixed in 10.3..

[mjr]

Would it be possible to get you guys to have a new look at the nudging. It is pretty good already. But I have a feeling, that it could be improved if there were some kind of adjustable hold-off timer available. It might be me and my settings that ain't correct, I haven't seen any good explanation of what the Gain parameters do.

 

The Gain settings are just multipliers for the raw input from the accelerometer.  If you double the gain, you'll double the amount of acceleration applied to the ball in the simulation for a given amount of physical acceleration read on the accelerometer.

 

The reason the Gain settings exist is that raw input from the accelerometer is in arbitrary units that vary by device, so we need some way to normalize the scale so that it produces sensible results in the simulation.

 

I'm not sure what you have in mind with "some kind of adjustable hold-off timer".  Could you be more specific?

Edited by mjr, 03 April 2017 - 07:51 PM.

[Thalamus]

@mjr Thank you for replying to this thread. As I said, I'm not sure if my settings are optimal or anything like that. But, from what I've noticed on test tables where there is almost no gravity is that a nudge to lets say left often gets followed automatically by a counter nudge to the right. My though about this was the if possible once a reading from the accelerometer hit a value that gets reacted upon it would not take into the account the next readings until a adjustable time had passed. We're talking a few ms here.

 

If that was possible to put in the code I would be very interested to see the result of such a change.

[fuzzel]

rev3006 is up:

 

- remove most copy/pasted code from new portrait patch and other tiny cleanups
- add SMAA
- fix audio preferences dialog not showing up
- rename BackdropFSS to ShowFSS to test for the FSS setting more consistent (e.g. similar to ShowDT)
 

[toxie]

..so here finally is SMAA, for all you that complained for so long..   was a bit of a pain to get that running correctly..

[mjr]

@mjr Thank you for replying to this thread. As I said, I'm not sure if my settings are optimal or anything like that. But, from what I've noticed on test tables where there is almost no gravity is that a nudge to lets say left often gets followed automatically by a counter nudge to the right. My though about this was the if possible once a reading from the accelerometer hit a value that gets reacted upon it would not take into the account the next readings until a adjustable time had passed. We're talking a few ms here.

 

If that was possible to put in the code I would be very interested to see the result of such a change.

 

You might try UN-checking Enable Nudge Filter if you have that enabled.  Doing that might reduce the effect.

 

The nudge filtering artifically injects a counter-acceleration of the type you're seeing.  If you think about it, the counter-acceleration is physically necessary: the basic Newtonian mechanics of the situation are such that a single acceleration in one direction would make the machine keep moving in that direction.  Now, we know that in reality the machine doesn't actually fly off sideways never to return; it stays where it started, brought back to its resting position by spring-like forces in the legs.  The accelerometer reads this reversal just like it reads the original nudge.  So what's the filter for?  It's there because even though the accelerometer sees the reversal, VP often misses it, due to sample timing in the input stream.  There are several contributors to the data loss - it's a combination of the accelerometer's ADC sampling frequency, the frequency of device-side USB report generation, the frequency of Windows-side USB report polling, and the frequency of VP's input polling.  In my testing, the net result is that VP only receives from 30% to 60% of high-frequency fluctuations in raw accelerometer readings.  The filtering is there to correct for the data loss by artificially reconstructing what must have happened based on an internal model, essentially supplying the simulation with accelerations that keep the simulated machine from flying off sideways even when the real readings of those forces get lost in the input stream.

 

Based on the experimenting I did that led to the filter in the first place, I'm afraid I'm not very optimistic that putting in some kind of anti-filtering that intentionally ignores real readings would be beneficial.  If anything I think it would magnify the problems that the filter was meant to reduce, namely excessive asymmetrical accelerations on the simulated ball in response to nudges.

[Umpa]

ah, okay, thanks.. yup, this is fixed in 10.3..

Yes thank you! Metallica here I come!

Sent from my SAMSUNG-SM-G935A using Tapatalk

[Patrick2610]

 

Fix the DT blank display for Windows 10 with this update please...

 

??? Details?

 

 

While Not in Windowed Fullscreen (1280x960 below)

[fuzzel]

Wow that is detailed

[fripounet]

How to integrate the property window into the main window ?