834 replies to this topic

[mjr]

Hi - I know the VP10 code is preliminary and isn't expected to work well with .vpt tables yet, but I'm curious if that's supposed to work at all.  I've compiled the latest version from the SourceForge trunk and it loads .vpt tables fine into the editor, but when I run them, the table is rendered at the wrong scale.  Most tables are rendered about halfway down the screen, with the top half of the screen black.  Some tables look like they're magnified to about 2x size, and they're also moved halfway down, so I can just see the bottom right quarter or so of the screen.  (I've only tried full-screen cabinet tables, by the way.)

 

Is there a video prefs setting or table setting I need to change to get the proper scale/position?  Or is it just too early to be trying to run .vpt tables at all?

[fuzzel]

The reason for this is that the backdrop settings/calculation (things like scaleing, inclination, layback) is different in VP10. You have to set new values for VP9 tables. Try to start with scalex/y/z = 1.0 this should nearly fix the layout. But VP10 won't play VP9 tables out of the box. VP10 is not backwards compatible to VP9. You can load VP9 tables though but this is just a helper for those who wants to convert a VP9 table to VP10.

[mjr]

The reason for this is that the backdrop settings/calculation (things like scaleing, inclination, layback) is different in VP10. You have to set new values for VP9 tables. Try to start with scalex/y/z = 1.0 this should nearly fix the layout. 

 

Thanks - I'll play around with those a little.

 

 

But VP10 won't play VP9 tables out of the box. VP10 is not backwards compatible to VP9. You can load VP9 tables though but this is just a helper for those who wants to convert a VP9 table to VP10.

 

That's what I was inferring from earlier in the thread, but thought I'd give it a try.  I mostly want to compare the analog nudge physics, since I'm not getting very good results so far in VP9.9.  It looks like this has been overhauled in the new code so I wanted to try it out.

Edit: Actually, on closer inspection, it doesn't look like there are any changes after all.

Edited by mjr, 18 July 2014 - 10:02 PM.

[fuzzel]

Here is another rubber demo. I fixed the issue kiwi reported and found an ugly bug that happens if you rotate a rubber object. The only "problem" that is still in is that one control points looks not correct. The reason for this is that I reused the spline routine used for ramps and walls and connected the beginning of the spline with it's end. Special care must be taken with non smoothed control points depending how steep the angle from one control point to the next is the rubber could look strange in that area.

 

Happy testing!

Attached Files

•  VPX_RubberDemo3.rar   943.84KB   24 downloads

[jpsalas]

Thanks fuzzel! The rubbers look very nice, and the problem I was having abou a new point getting 0,0 as coordinates is gone.

[fuzzel]

[kiwi]

No problem if the start-end point does not seem right, I hide it behind a peg, surely you will find a solution, even with the ramps you can not close a ring perfectly.

 

When I add a new control point to the rubber, VP crashes.

 

Thanks

 

Max

[fuzzel]

When I add a new control point to the rubber, VP crashes.

 

Thanks

 

Max

Do you have an example project? I can't reproduce the crash. I can add as many control points as I like without a crash!?

[kiwi]

 

When I add a new control point to the rubber, VP crashes.

 

Thanks

 

Max

Do you have an example project? I can't reproduce the crash. I can add as many control points as I like without a crash!?

 

I have tried several times to a New table (or other tables) to add an rubber and then a point, VP crashes without any error message, or with the usual message: You have an error in Visual Pinball .... . The application will be closed.
However, VP allows me to move or delete control points.

 

Max

[lio]

much better, fuzzel!

one very slight problem I still have: width 5 seems already slightly too thick while width 4 is already too thin for a regular ring-aroung-peg situation (in FP we used 5 mm which would be ~9,6 vp units diameter or 4,8 vp units radius)

 

Also the rendered representation does not seem to match what we see in the editor (in the editor the rubber looks thinner - I'd say the editor draws the "width" value as if it where the diameter while when rendering it uses the value as radius).

 

Try it out: place two walls next to each other with 10 VP units of space between them - now place a long stretch of rubber of width "10" between these walls - in the editor it will fill the gap perfectly but when you press on "play" the rubber will be much wider than the gap.

Edited by lio, 21 July 2014 - 06:29 PM.

[fuzzel]

Ah thanks for all the tests. Those small issues always slips through

 

Here is a new demo. I found the memory leak which can happen when you add a control point and renamed the rubber "width" to "thickness". It's actually the diameter but to not baffle too many new users thickness is a bit clearer!? Also using this value is corrected now

Attached Files

•  VPX_RubberDemo4.rar   944.31KB   143 downloads

[lio]

thanks for the unbelievably quick fix! :-)

[kiwi]

Good job Fuzzel ! rubber works like a charm, thanks.
And there are new features are not working yet, but very interesting .
 

Playing with the settings I have noticed, in the "Editor Options", if I save the settings, the "Auto-Save Table Every" box is unchecked even if it is selected,
probably something wanted or not finished.

 

Max

[kiwi]

After another test, the control point with coordinates 0,0 is returned.
If I add a point, where the red arrow indicates, sometimes the point is put at 0,0, does not always happen.

 

 

Thanks

 

Max

 

 RubberTest.rar   446.34KB   6 downloads

[fuzzel]

Thanks Kiwi! I changed the control point handling a bit. It's better now but not completely fixed. It seems that the algorithm for finding the point on the polygon(rubber) where you want to add a control point returns 0,0 sometimes.

[J3SteR]

Ok I've been away from the scene for a very long time. I took a part my pincab but I want to at some point rebuild it. I prefer AMD video cards but VP has always favored Nvidia cards. Is there any chance vp10 might be less bias on video hardware? Maybe some optimizations for AMD hardware? Thanks

[sliderpoint]

Ok I've been away from the scene for a very long time. I took a part my pincab but I want to at some point rebuild it. I prefer AMD video cards but VP has always favored Nvidia cards. Is there any chance vp10 might be less bias on video hardware? Maybe some optimizations for AMD hardware? Thanks

 

I've seen this comment about not using AMD cards a few times on different forums.  I think it must be referring to older versions of VP or older AMD cards.  I have a Radeon R7 260X 2GB card that runs all three of my monitors in my cab.  I have had zero issues with performance or graphic anomalies on VP 9.21, 9.90, all the physmod versions and rev1081 of VPX.

 

-Mike

 

edit: that same card also runs my VPinMame and windows audio at the same time through HDMI Audio.

Edited by sliderpoint, 24 July 2014 - 03:56 AM.

[Arcade4]

Another satisfied AMD card owner here. Just a lowly 5670 and no problems at all. 

[fuzzel]

As slidepoint said the argument that only NVIDIA cards should be used for VP is really old. You can use any AMD card you like as long as the card has enough power for normal PC games especially if you want to connect multiple monitors to one card. I run a Radion HD7770 and I have no issues with VP. If you use the old DX7 VP versions (9.21 or below) you need a graphics card with really fast memory but with the portation to DX9 these issues are all gone.

[Kernel]

I also use a Radeon HD 6670 1Gb and no issues at all.

The only thing that's not clear (at least for me) is the "prerendered frames" option that maybe works only for nvidia cards.

AMD cards have the equivalent "flip queue size" but I don't know if the VP option works the same.