666 replies to this topic

[toxie]

actually he wanted to be back when he was finished, and AFAIR he had finished some time ago.. but who knows..

[mukuste]

Does anyone know when Cupid will be back?  (After college?) - he did have that DX10 table render and player working with vp9 tables.  Maybe he would want to work on the rendering engine for VP10?

 

Wow, so a DX10 renderer already existed at some point? Why did nobody tell me How far along was that, and is the code available? Not that it really matters anymore, the DX9 port is pretty far along. But I'm still curious.

[kruge99]

 

Does anyone know when Cupid will be back?  (After college?) - he did have that DX10 table render and player working with vp9 tables.  Maybe he would want to work on the rendering engine for VP10?

 

Wow, so a DX10 renderer already existed at some point? Why did nobody tell me How far along was that, and is the code available? Not that it really matters anymore, the DX9 port is pretty far along. But I'm still curious.

 

 

I could be wrong, but I think it was just a standalone player, written from scratch.

 

See here:

 

http://www.vpforums....showtopic=11534

[mukuste]

I could be wrong, but I think it was just a standalone player, written from scratch.

 

See here:

 

http://www.vpforums....showtopic=11534

 

 

Oh wow, so the idea there was a reimplementation from scratch. That's... immensely ambitious. I'm kind of glad I aimed lower right now

[The Loafer]

At the speed you have been going at, so are we!

[DreamTrap]

just downloaded 908 how damn cool are the ball trials. not sure if they when they were added but still cool! vp just keeps getting better and better

Edited by DreamTrap, 08 February 2014 - 03:32 AM.

[GSGregg]

Hi; downloaded 908 and when I click the shortcut I immediately get Error: unable to load SciLexer.DLL When I click on OK or 'x' it out, the Editor comes right up, but any selected table is rendered as only a static image, with no VPM splash, DMD, sound or attract mode lightblinks. Likewise, keyboard presses produce nothing.

 

Having downloaded 9.2.0 minimal, I do have SciLexer.DLL 3.3.6.0 in my Visual Pinball folder; have I missed some other vital component?

 

Gregg

[cupid]

 

Does anyone know when Cupid will be back?  (After college?) - he did have that DX10 table render and player working with vp9 tables.  Maybe he would want to work on the rendering engine for VP10?

 

Wow, so a DX10 renderer already existed at some point? Why did nobody tell me How far along was that, and is the code available? Not that it really matters anymore, the DX9 port is pretty far along. But I'm still curious.

 

 

Actually i'm done with my MBA-Study, but I'm not done with my master-thesis. After a break for personal recovery after the main phase of my study, I startet to write the thesis and i'm right in the middle of writig it and that unfortunately consumes very much of my personal time and energy.

 

I look very rarely at the vp forums, but i know what you guys are doing right at the moment. I still get the revision-updates and look for them in my e-mails and i heard of the DX9 project from my brother, saw some changes "under the hood" (Index- and vertex-buffers used now), saw some VERY NICE work on the primitives (which was my last project before starting my MBA). I asked dev team that time if i should take primitives out again, because they were buggy, but we decided to let it in.

 

Very nice work and keep it up!

 

The DX10 port of the player was totally written from scratch in C# using the XNA framework, but most of the code was written to circumvent the XNA stuff in it (could not load textures, when not loaded into dev-studio, etc.). Everything media-related was a Pain In TA. The current changes in the main trunk (using vertex and index buffer for almost everything) are way better than my port. So i would not recommend to look at it (and translate all this XNA specific stuff out).

 

I learned a lot from it, and i learned not to use a GFX-Framework for a Pinball Engine because it either has too much or not enough for our needs - depending on the subject. Maybe the Unity-Project will come to another conclusion. I don't know.

 

I have to say that i did not have these framerates with my player that time (Although it was only a graphics display and no physics or what else...). But i believe that i measured real gfx frames and you measure the fps-display in VP (which is more or less a "physics frame per second display", if i remember correct).

 

I don't have any clue how much this DX9-project and the main dev-team work together (and if there are problems to include the dx9 port some time in the main Dev-Trunk), but i believe that a DX9 port will stabilize the performance for us all (even for those who seem to have lower FPS at the moment).

 

So be sure, that i will be back someday, but just give me some more time for my thesis. I will still see thay you great guys do great things.

[mukuste]

Thanks for the explanation, quite interesting. i will say that the FPS numbers shown in VP are actually the graphics frames per second,(albeit averaged over 1s or so). However if, as you say, you didn't use vertex and index buffers extensively, then I can imagine that the framerate would be quite a bit lower. Also don't forget that even in the DX9 port, the background and static elements are still prerendered, while in your project I assume everything was dynamic, so that is a major difference.

 

Good luck for your thesis! Believe me, I know how much time these things take. Hope you can finish that successfully soon and return to VP.

[fuzzel]

So be sure, that i will be back someday, but just give me some more time for my thesis. I will still see thay you great guys do great things.

 

Hmm what about writing your thesis about VP then you can kill two birds with one stone

[cupid]

Mukuste:

 

Of course, i used Index- and vertex buffers (as XNA needs them, btw), but i think that todays buffers in VP are better sorted from what i have seen in the code.

 

If you read something about FPS-measuring, you will see, what i mean with different methods of FPS-Calc (Frank D. Lunas 3d Game Programming with dx11 is great).

 

A nice read about misleading frame time is: dunlop Frame time (just google it, second hit). Dunlop explains, why a drop from 1000 to 500 fps is actually better as a drom from 100 to 75 fps, although a 50% drop rate seems shocking at first.

 

I think (and i hope that this is right, which might not be!), that VP measures the Count of physics Frames per graphic frame. So graphics are not measured directly but indirectly with the time between each frame. This is why FPS drop so drastically, if many things (especially in Multi Ball) happen in VP. Actually there is no stutter in graphics, but you feel microstutter because physics are not updated as often as in other graphics frames.

This leads to: and this is only a guess, too:

I don't know if VP measures the average Physics frame per second. and executes physics as often as measured (even if physics calc is longer) - maybe vp adjusts itself only some graphic frames later. This would couse even stutter rather than only microstutter.

 

In todays graphics especially in DX11 (but 11 is not a must), multithreading is used to overcome this issue. But I have not yet checked, if multithreading is possible with VP or not. I recommend The Book Game Coding Complete for such questions. They suggest, depending of the game you write, to use multiple threads, not just two (which is common since many years). They actually present their own multithreadingmanager in the book, but i heard other opinions -> why to use only two (or three, one for sound) threads in games.

 

I have read some books, while i studied, just to relax, but did not program anything because of time.

 

Another good book (although a little bit superficial) was "Game Engine Architecture" from Jason Gregory. While Game Coding complete (GCC) is too special in some cases, this book gives you some Background about what's goin on in GCC. I saw a pdf at Google.

 

@fuzzel

Maybe i should write my thesis about vp ;-), but i don't know how to combine it with Business Administration. I think i will stay with human capital measuring instead measuring FPS. :-o

 

Thanks for you wishes. As i always said: I will be back.

Edited by cupid, 10 February 2014 - 07:59 PM.

[mukuste]

I absolutely understand the distinction between frame time and FPS. That was actually exactly the point I made when people were concerned that adding dynamically rendered flippers dropped their framerate from 4000 fps to 2000 fps on the default table: it sounds horrible, but it's just an extra 0.5 ms per frame.

 

VP doesn't display the physics frames in its FPS counter. It's purely dependent on the graphics frame time. You'll have to either believe me about this, or check the code (it's in pin/player.cpp).

 

Multithreading in game engines is really only useful if you are CPU bound so that you can distribute the work across multiple CPUs. VP isn't very CPU hungry at all in my experience, so I don't think it's worth the hassle. As soon as you go multithreaded, you have to deal with a whole new set of headaches related to synchronization. Typically this will also introduce extra latency, something that you don't really want in something that has to be as responsive as possible, i.e., a pinball simulation.

Edited by mukuste, 10 February 2014 - 08:22 PM.

[fuzzel]

We should really try to understand the physics engine in VP. Right now the engine is more or less a mystery with a lot of magic numbers scattered over multiple files. I hope we can convert VP to DX9 without touching the physics engine first but we should definitely doing that later on.

 

Anyway, does anyone has found some "new" bugs on the 9.2.1 lately or wants to change some minor things? Otherwise we should start the release phase on this version and migrating VP to DX9 after that. Any opinions?

[unclewilly]

I'm ready for the release as I'm about to do the final updates on mb

[jimmyfingers]

@fuzzel - There was one last thing I was thinking might be useful for having a more neat and organized collection manager and that was to add the same ability to the collection groups as their individual elements have regarding ordering them higher or lower in the lists.  I know that for the elements it has a very meaningful aspect if utilizing their index values for coding / subroutines and I imagine that's why it exists, however, for the groups themselves it would just be nice to be able to re-order them for grouping similar groups or even list them alphabetically as right now the order is simply the order in which they were created. 

[dup3d]

hi late to the party but first i want to say thank you for all the great work on the VP Versions. Really amazing how it evolves and i´m suprised everyday by news/features/bugfixes/dx9migration and other things.

also want to thank all testers and table builders who test the new rev and report bugs or suggestions. great to have such a community working together 

 

so as i´m not really a table builder but modding tables from time to time i have something collections related also.

think it´s not mentioned before but if this idea is reported before simply forget my whish ^^

don´t know if it´s easy to do but it would be great to have a "right click option" on objects for "add to collection". I think it would be a great timesaver for a lot of people when building and modding tables.

something like the "Assign to....> Layer1,Layer2... " would be nice but for the existing collections like "Add to collection > collection1, collection2..."

hope you understand what i mean if not so let me know then i pm you in german to describe it a bit better

 

keep your great work up....

Edited by dup3d, 11 February 2014 - 04:06 AM.

[teppotee]

+1 for dup3d's suggestion. I requested the same thing sometime in the past. Of course only if it's relatively easy to add.

 

Feature would be very useful when you need to create collections for example for rubbers where you want to set physics properties or hit sounds for the whole collection. Just right click -> "add to collection" (collections could still be created with the collection manager). Currently you need to basically write the objects down somewhere -> go to collection manager and then try to find the same objects again from the whole list. That's where I usually give up and just skip creating the collections

[martinoil]

I just installed the latest 9.2.1 rev 912.exe.And none of my db2s backglasses show up anymore ?. Also dont see any DMD s either ?.

[fuzzel]

rev915 adds a CurrentAngle function to the gate object. The usage is the same as for the flippers and is only available from the script.

so as i´m not really a table builder but modding tables from time to time i have something collections related also.

think it´s not mentioned before but if this idea is reported before simply forget my whish ^^

don´t know if it´s easy to do but it would be great to have a "right click option" on objects for "add to collection". I think it would be a great timesaver for a lot of people when building and modding tables.

something like the "Assign to....> Layer1,Layer2... " would be nice but for the existing collections like "Add to collection > collection1, collection2..."

Well I think that isn't that easy to implement and it could look really confusing if you have more than 10 collections as that would increase the context menu a lot. But I'll check that and maybe I can add it to the 9.2.1

Edited by fuzzel, 11 February 2014 - 08:54 AM.

[mukuste]

One idea: add only a single collection in the context menu, a special temporary selection. The user can put all their stuff there, and then the temporary collection can be renamed to something suitable from the Collections dialog.