Where new test build will become available?
- Balthazar
- Moderator
- Posts: 2055
- Joined: Wed Nov 01, 2006 4:31 pm
- Location: Russian Federation
- Contact:
Where new test build will become available?
Well, new test build seems to be obvious
Re: Where new test build will become available?
You can got a playable [1809] build from here:
http://translate.google.com/translate?u ... N&ie=UTF-8
http://translate.google.com/translate?u ... N&ie=UTF-8
- zuzuf
- Administrateur - Site Admin
- Posts: 3281
- Joined: Mon Oct 30, 2006 8:49 pm
- Location: Toulouse, France
- Contact:
Re: Where new test build will become available?
it'll be available when we'll have fixed several obvious things broken by recent changes.
Most likely after 0.5.4 release (so not before next week).
Most likely after 0.5.4 release (so not before next week).
=>;-D Penguin Powered
- Balthazar
- Moderator
- Posts: 2055
- Joined: Wed Nov 01, 2006 4:31 pm
- Location: Russian Federation
- Contact:
Re: Where new test build will become available?
There is no way I can download anything from there. It asks to install some nano-robot stuff, all in chineese... Hell, nothing I can do about it. Is there a more easy way to download only TA3D build (10-15 mb). Rapidshare maybe?xpoy wrote:You can got a playable [1809] build from here:
http://translate.google.com/translate?u ... N&ie=UTF-8
Re: Where new test build will become available?
you should copy TA files for playable
This is the playable files At least
http://cid-dd761cec1c1d3498.skydrive.li ... D/beta.rar
This is the playable files At least
http://cid-dd761cec1c1d3498.skydrive.li ... D/beta.rar
Re: Where new test build will become available?
Rev 1809 : http://ta3d-download.shikami.org/binari ... -r1809.exe (it is the ta3d binary only)
Damien Gerard
Ta3d & Yuni Developer
Ta3d & Yuni Developer
- Balthazar
- Moderator
- Posts: 2055
- Joined: Wed Nov 01, 2006 4:31 pm
- Location: Russian Federation
- Contact:
Re: Where new test build will become available?
Hurray Thanks man!milipili wrote:Rev 1809 : http://ta3d-download.shikami.org/binari ... -r1809.exe (it is the ta3d binary only)
Re: Where new test build will become available?
Well....
Milipili should reply more quickly,blablabla
- Balthazar
- Moderator
- Posts: 2055
- Joined: Wed Nov 01, 2006 4:31 pm
- Location: Russian Federation
- Contact:
Re: Where new test build will become available?
Somehow you are prefer quite annoying file hostingsxpoy wrote:
Well....
Milipili should reply more quickly,blablabla
Re: Where new test build will become available?
Daily snapshots are planned.
Damien Gerard
Ta3d & Yuni Developer
Ta3d & Yuni Developer
Re: Where new test build will become available?
Program error?
I just think that DLL may useful.
I just think that DLL may useful.
- zuzuf
- Administrateur - Site Admin
- Posts: 3281
- Joined: Mon Oct 30, 2006 8:49 pm
- Location: Toulouse, France
- Contact:
Re: Where new test build will become available?
no, sorry, when I try to download it, the server sends me an error
=>;-D Penguin Powered
Re: Where new test build will become available?
I upload to another site.
This time may work.
http://bbs.taclub.net/taclub/newTA/atta ... gBQlQiGqt4
Re: Where new test build will become available?
What is Maze Engine ? Sorry but I have more affinity with Japanese rather than Chinese
Damien Gerard
Ta3d & Yuni Developer
Ta3d & Yuni Developer
- zuzuf
- Administrateur - Site Admin
- Posts: 3281
- Joined: Mon Oct 30, 2006 8:49 pm
- Location: Toulouse, France
- Contact:
Re: Where new test build will become available?
Thanks, it worked
I tested the program and all the AI I found in the package. The problem is not finding a path, TA3D does that well (but it doesn't try to optimize it, but it seems those AIs doesn't either), the real problem is the blocky representation used to compute the path doesn't match the real stuffs once you look at how things move .
TA3D algorithm is very simple, it goes toward the target destination until it hits some obstacle, then it follows the edges of this obstacle (which is a lot faster when the obstacle is not convex). The computed path you get with that king of algorithm is not optimized, but you can then check for direct paths between points of the original path (if you test from the end and then divide by two the path length each step it doesn't take too long), note that this could be optimized to get the optimal path at a computational cost of 2^nb_obstacles times current complexity. Once you have a path, units must follow it, but they have to obey physics laws and rounding errors , so they are never exactly where they should and sometimes they can't go where they are ordered to even if it looks they can, because of those small errors. I tried to add a small distance between objects and units without success
NB: concerning heavily recursive algorithms (like superman :p), they are too costly when there are non convex obstacles (think a map with rivers), that's why TA3D doesn't use that kind of algorithm any more (on some maps, pathfinder could use up to 95% of the CPU )
I tested the program and all the AI I found in the package. The problem is not finding a path, TA3D does that well (but it doesn't try to optimize it, but it seems those AIs doesn't either), the real problem is the blocky representation used to compute the path doesn't match the real stuffs once you look at how things move .
TA3D algorithm is very simple, it goes toward the target destination until it hits some obstacle, then it follows the edges of this obstacle (which is a lot faster when the obstacle is not convex). The computed path you get with that king of algorithm is not optimized, but you can then check for direct paths between points of the original path (if you test from the end and then divide by two the path length each step it doesn't take too long), note that this could be optimized to get the optimal path at a computational cost of 2^nb_obstacles times current complexity. Once you have a path, units must follow it, but they have to obey physics laws and rounding errors , so they are never exactly where they should and sometimes they can't go where they are ordered to even if it looks they can, because of those small errors. I tried to add a small distance between objects and units without success
NB: concerning heavily recursive algorithms (like superman :p), they are too costly when there are non convex obstacles (think a map with rivers), that's why TA3D doesn't use that kind of algorithm any more (on some maps, pathfinder could use up to 95% of the CPU )
Re: Where new test build will become available?
algorithm trouble me more than english.
I study by Memory and Reasoning more than Think
Same as you see,may my advice are very bad.
But i will show all my advice, at least they maybe work,
TA use a slope and a Climbing ability to check something,i think we can made some simple path map for common Climbing ability when load map.It just cast some load time and some memory.The simple path map only include 2 types groud,that pass or unable pass.
Or we can use a virtual,that virtual unit pass run the return of wayFind,if they arrived,it ok,or return false,a wapper.May it cast too many Performance.
I study by Memory and Reasoning more than Think
Same as you see,may my advice are very bad.
But i will show all my advice, at least they maybe work,
TA use a slope and a Climbing ability to check something,i think we can made some simple path map for common Climbing ability when load map.It just cast some load time and some memory.The simple path map only include 2 types groud,that pass or unable pass.
Or we can use a virtual,that virtual unit pass run the return of wayFind,if they arrived,it ok,or return false,a wapper.May it cast too many Performance.
- zuzuf
- Administrateur - Site Admin
- Posts: 3281
- Joined: Mon Oct 30, 2006 8:49 pm
- Location: Toulouse, France
- Contact:
Re: Where new test build will become available?
The main problem is memory requirements ... TA uses slopes and footprint sizes as two constraints to units movements, as well as a minimum and maximum map height (sea level being defined as the 0 here). Units have their own parameters, so precomputation cost would be very high and memory requirements would be worse.
Also, the yardmap (TA's units and features map in footprint units) is changing during the game and changes pathfinder's behavior, so precomputing is not the way to go I think. Current algorithm is very fast and things are computed only when it is required, I think we could just find better constraints to make it find good paths, by good I mean easy to follow. For example we prefer that the unit follows a wide road instead of crossing a dense forest.
Concerning slopes, TA define slopes, but I don't really know how those slopes are given, is it an angle, a tan, which unit system ?
Also, the yardmap (TA's units and features map in footprint units) is changing during the game and changes pathfinder's behavior, so precomputing is not the way to go I think. Current algorithm is very fast and things are computed only when it is required, I think we could just find better constraints to make it find good paths, by good I mean easy to follow. For example we prefer that the unit follows a wide road instead of crossing a dense forest.
Concerning slopes, TA define slopes, but I don't really know how those slopes are given, is it an angle, a tan, which unit system ?
=>;-D Penguin Powered
Re: Where new test build will become available?
It was a convex
The unit's fbi,or may in tdf,just that file define category.The maxslope and maxunderwater.It was a useful thing in convex
PS: i wil sleep early since today,thus i couldn't keep those immediately talk.
The unit's fbi,or may in tdf,just that file define category.The maxslope and maxunderwater.It was a useful thing in convex
PS: i wil sleep early since today,thus i couldn't keep those immediately talk.
Who is online
Users browsing this forum: No registered users and 5 guests