Plan for Drew Parker
by Drew Parker · 02/14/2005 (1:36 pm) · 23 comments
My first official .plan to the GG community. :)
Hello everyone, you may not have heard much from me but a small rumble on the GG forums, but I've been doing TGE development for almost one year now. It's funny, although most people don't know me, I feel like I know a lot of you pretty well already. .plans are a great thing, and it's been terrific to be a part of such a great, helpful community.
Update 3-2-05:
Due to complications that have recently arisen, the team of which I am a member is unable to release the previously mentioned resource at this time. We are still considering this option and may be able to do so in the future, but as of now the team cannot make any promises as to when or if that will happen. We apologize for raising expectations prematurely.
GarageGames is a great community, and without it so many projects wouldn't be possible, including possibly, ours! I know everyone on our team feels the same way about the GG community, and so we'll continue to look for ways to contribute to the community.
Thanks for your understanding. You all know what it's like in software / game development, it can be an ever changing, surprising road. :)
Hello everyone, you may not have heard much from me but a small rumble on the GG forums, but I've been doing TGE development for almost one year now. It's funny, although most people don't know me, I feel like I know a lot of you pretty well already. .plans are a great thing, and it's been terrific to be a part of such a great, helpful community.
Update 3-2-05:
Due to complications that have recently arisen, the team of which I am a member is unable to release the previously mentioned resource at this time. We are still considering this option and may be able to do so in the future, but as of now the team cannot make any promises as to when or if that will happen. We apologize for raising expectations prematurely.
GarageGames is a great community, and without it so many projects wouldn't be possible, including possibly, ours! I know everyone on our team feels the same way about the GG community, and so we'll continue to look for ways to contribute to the community.
Thanks for your understanding. You all know what it's like in software / game development, it can be an ever changing, surprising road. :)
About the author
#2
Nick
02/14/2005 (2:39 pm)
That's really exciting Drew! I remember you talking about it in another thread. Did you get the small object problem fixed? Keep it up!Nick
#4
02/14/2005 (2:41 pm)
That looks great! I want it ;)
#6
02/14/2005 (3:32 pm)
Good job! I did a little Newton Physics Work myself. Didn't amount to much tho.
#7
02/14/2005 (5:48 pm)
Wow! Is that complete ODE integration for multi-player, or just singleplayer?
#8
02/14/2005 (11:03 pm)
Will this be a free resource or will we have to pay for it?
#9
Thanks for the comments! I want to mention upfront that there was our whole team working on this, not to mention it buildling off the great work the authors of the ODEItem 0.1 resource already did. So even though the news is posted in my plan, it's definitely not all me :)
@Josiah
The editor is part of the resource.
@Nick
We found out the small object problem was not ODE, but something on the TGE side. It seems to be coming from the collision code due to precision errors, because we were using such small objects. I think most devs won't have that problem, as we were simulating very small objects, with a world scale already much smaller then the standard TGE scale.
@Jeremy
Alright :) If you get time to try it out, maybe you can give us your feedback. It'd be great to get more people involved in this so any remaining issues or extra features can get ironed out.
@Owen
Nice video. Owen is a part of our team and has worked a lot with ODE, he's been transferred to the promotional department just for today. :)
@Chris
How did you like working with the Newton SDK? I wonder how similiar it is to ODE, I haven't looked into it much
@Mike
The screenshot with the raining boxes is single-player with no frame drops. The last "mystery" picture is actually a network implementation, with the server ODE being the yellow wireframe, and the client-side ODE being the blue wireframe, using "p" for prediction. :) I know, bad hint, I couldn't think of anything else. When the object is waiting for a server update, it starts using the clientside ODE sim to figure out what it should do next. The network performance is pretty good, especially with 10 or less objects. Once you get above that, things are still smooth, but it's not ideal. This is one area where we could use some others' input.
@Burning
This will be a GG resource, so it will be free.
02/15/2005 (6:30 am)
Hi all,Thanks for the comments! I want to mention upfront that there was our whole team working on this, not to mention it buildling off the great work the authors of the ODEItem 0.1 resource already did. So even though the news is posted in my plan, it's definitely not all me :)
@Josiah
The editor is part of the resource.
@Nick
We found out the small object problem was not ODE, but something on the TGE side. It seems to be coming from the collision code due to precision errors, because we were using such small objects. I think most devs won't have that problem, as we were simulating very small objects, with a world scale already much smaller then the standard TGE scale.
@Jeremy
Alright :) If you get time to try it out, maybe you can give us your feedback. It'd be great to get more people involved in this so any remaining issues or extra features can get ironed out.
@Owen
Nice video. Owen is a part of our team and has worked a lot with ODE, he's been transferred to the promotional department just for today. :)
@Chris
How did you like working with the Newton SDK? I wonder how similiar it is to ODE, I haven't looked into it much
@Mike
The screenshot with the raining boxes is single-player with no frame drops. The last "mystery" picture is actually a network implementation, with the server ODE being the yellow wireframe, and the client-side ODE being the blue wireframe, using "p" for prediction. :) I know, bad hint, I couldn't think of anything else. When the object is waiting for a server update, it starts using the clientside ODE sim to figure out what it should do next. The network performance is pretty good, especially with 10 or less objects. Once you get above that, things are still smooth, but it's not ideal. This is one area where we could use some others' input.
@Burning
This will be a GG resource, so it will be free.
#12
02/15/2005 (5:39 pm)
I know a bit about Torque's netcode and I'd be more than willing to help test this out. Our project calls for large numbers(200+) of dynamic objects over a network, so it seems like a pretty good stress test. We've been making do with the basic Item physics, which though efficient, suck. So, yea, if you need some sort of performance reading from a situation with four crossplatform clients and lots of objects, my AIM screenname is in my profile.
#13
We are getting closer :)
@Alex
That would be great! I'd love to get the network performance really fine-tuned. It comes down to some interesting issues with the physics engine itself, and how to best optimize the client side. 200+ objects sounds like a pretty intense test. :)
02/16/2005 (9:39 am)
Hey everybody,We are getting closer :)
@Alex
That would be great! I'd love to get the network performance really fine-tuned. It comes down to some interesting issues with the physics engine itself, and how to best optimize the client side. 200+ objects sounds like a pretty intense test. :)
#14
02/16/2005 (9:46 pm)
Optimizing client side... does ODE have any way to adjust the accuracy of the simulation? Maybe you could hook that into the update priority or something. Can either do it so high priority objects get the nicest simulation so if anything goes wrong, they look fine, or so that low priority objects get the nicest sim because they're doing much more client side work. Personally, I'd code in both and make it a preference.
#15
My email is: igpgames@tiscali.it
Note: it's URGENT.
Thanks.
Hi, Ivix.
02/20/2005 (1:39 pm)
May you contact me? I wish to talk about this.My email is: igpgames@tiscali.it
Note: it's URGENT.
Thanks.
Hi, Ivix.
#16
02/20/2005 (10:13 pm)
No! They're busy fixing the bugs.... quiet now ;-P
#17
02/21/2005 (1:41 pm)
Nice one Drew!! Just what me and my team have been seaerching for recently :D YAY! 
#18
02/22/2005 (1:31 am)
your ressource looks so awesome !!!
#19
Are joints and constraints being implemented too?
Paul
02/22/2005 (3:41 pm)
Drew, very exciting! I think a general physics implementation was the key item missing from TGE. Especially good to see a server version.Are joints and constraints being implemented too?
Paul
#20
Anyway, I think I can share this with the people (in fact is not "so confidential")...
May you send me the source of your integration? If so, I'll work on it with my team for a TSE implementation.
Because we are using TSE for our project and we need a working ODE integration.
What are the known bugs/todo?
If you agree to send me something, you can find my email address in my profile or contact me on igpgames@tiscali.it
Thanks in advance.
Hi, Ivix.
P.S.: suggestion for the network side: Isn't better to share only the world position (xyz) and the local rotation (xyz) values of all objects? A friend mine suggested me another way: let the simulation run on client-side only and send via network the forces applyed to an object only. (but this, I think, may result in different calculations for every client)
02/27/2005 (8:26 am)
Drew, the last comment i made I was talking with you.Anyway, I think I can share this with the people (in fact is not "so confidential")...
May you send me the source of your integration? If so, I'll work on it with my team for a TSE implementation.
Because we are using TSE for our project and we need a working ODE integration.
What are the known bugs/todo?
If you agree to send me something, you can find my email address in my profile or contact me on igpgames@tiscali.it
Thanks in advance.
Hi, Ivix.
P.S.: suggestion for the network side: Isn't better to share only the world position (xyz) and the local rotation (xyz) values of all objects? A friend mine suggested me another way: let the simulation run on client-side only and send via network the forces applyed to an object only. (but this, I think, may result in different calculations for every client)
Torque Owner Josiah Wang
nice!