T2d mount-point based skeletons (for character art)
by Jason Swearingen · in Torque Game Builder · 01/27/2006 (3:56 pm) · 10 replies
I am thinking of how i will do artwork for my comming games.
For user customization (and developer flexibility) I am thinking of using skeletons (based on mount-points)
and then mounting sprites to it. To use a game-based analogy: Head, Sword, Shield, Boots, Legs, etc would all be seperate sprites that would be munged together in real-time to create the character that the user would see.
So my questions:
1) Are there many games that use this approach to art content?
2) What are the potential drawbacks to using art this way? (Animation, performance, etc)
3) What benifits are there to using art this way? (flexibility, etc)
I'm not an artist, but I do consider myself a kickass dev, so if there is a better (but more complex) way of tackling this problem i'm all ears.
-Jason
For user customization (and developer flexibility) I am thinking of using skeletons (based on mount-points)
and then mounting sprites to it. To use a game-based analogy: Head, Sword, Shield, Boots, Legs, etc would all be seperate sprites that would be munged together in real-time to create the character that the user would see.
So my questions:
1) Are there many games that use this approach to art content?
2) What are the potential drawbacks to using art this way? (Animation, performance, etc)
3) What benifits are there to using art this way? (flexibility, etc)
I'm not an artist, but I do consider myself a kickass dev, so if there is a better (but more complex) way of tackling this problem i'm all ears.
-Jason
#2
i guess other people really havent thought about the pros and cons of the system.
any t2d users here, who have worked with a large number of mount points? are there any technical issues with using, say 50 mount points on one object? (and have maybe 50 such objects on the screen at once)
01/29/2006 (3:53 pm)
Yeah, i wasnt really thinking about this for animation purposes, at least not in the 3d sense. (skeleton for poses would be ok tho)i guess other people really havent thought about the pros and cons of the system.
any t2d users here, who have worked with a large number of mount points? are there any technical issues with using, say 50 mount points on one object? (and have maybe 50 such objects on the screen at once)
#3
I have a basic mount system defined as so:
Having said that ofcourse, the game is relatively simple and there isn't a huge amount of other stuff going on at the same time (but I do have some schedules running at 20ms in order to guide the invaders around the screen when they're diving on the player or returning to their mount point - and that doesn't affect the frame rate either).
01/30/2006 (12:20 am)
My Galaxy Defender game is done with mount/link points and that runs pretty welll.I have a basic mount system defined as so:
-----|----- -----|----- -----|----- -----|----- -----|-----The vertical line is a dummy object, and each horizontal line is a dummy object mounted to the vertical one. At each '-' I create another mount point and attach my invaders, so that's 50 mounts. I'm also expanding and contracting the dummy objects so it looks like the invaders are pulsing and everything runs as smooth as silk.
Having said that ofcourse, the game is relatively simple and there isn't a huge amount of other stuff going on at the same time (but I do have some schedules running at 20ms in order to guide the invaders around the screen when they're diving on the player or returning to their mount point - and that doesn't affect the frame rate either).
#4
03/27/2006 (11:35 pm)
Hi, I just happened to stumble in on this. I think I understand what you are saying here. instead of drawing out a whole character as one sprite, you're saying to use sprites for each individual character part, and mounting it all to one master character piece. besides the potential hit on performance (more characters, more sprites; more sprites, more processing) the only thing I could forsee would be a problem with animating it. I'm not really sure how the mounting works and what all a parent node can pass down to a child node, so I might be off on this next part. But it seems to me unless a parent can actually let a child know what frame of animation it needs to play, this wouldn't work. Feel free to tell me otherwise, (I'd really love it if you did) as this would streamline my art asset creation process immensely not to mention open up huge amounts of character personalization options.
#5
For me, my big concern is I dont know any games that use such a methodology. My imagination tells me that this shouldnt be teribly hard to produce reasonable graphics with, but as there arnt any examples for me to work with (same with my other idea of a perspective based tile system) it will unfortunatly stay in the realm of speculation for the moment...
once I finish my current project (writing some tgb tools) and I am ready to start writing a game, this is definatly one of my first priorities to experiment with.
03/28/2006 (4:21 am)
@Jason Turner: In my minds eye (so be warned that's very subjective) your concerns are all pretty insignificant hurdles to overcome. Just having a sibling data structure that contains the actual skeleton information (including passing animation instructions to children) would be pretty easy to implement.For me, my big concern is I dont know any games that use such a methodology. My imagination tells me that this shouldnt be teribly hard to produce reasonable graphics with, but as there arnt any examples for me to work with (same with my other idea of a perspective based tile system) it will unfortunatly stay in the realm of speculation for the moment...
once I finish my current project (writing some tgb tools) and I am ready to start writing a game, this is definatly one of my first priorities to experiment with.
#6
03/28/2006 (3:08 pm)
See, that to me would be the easiest hurdle to overcome. I'd rather try to tread new waters, than swim in the overcrowded kiddie pool (not to call pros kiddies, but it seemed an appropriate metaphor) I'd really like to work with you on this when you get the time; as I'm not the most adept programmer (now) I'd think it would really help me out to see what someone more knowledgable than I would do, to overcome such hurdles. I'd gladly contribute whatever art assets I can, as well as my musical talents (which I'd say are my strongest suits) when they're needed. All I'd ask is for plentiful comments, and an occasional explanation as to the thinking behind certain methodologies. 
#7
Any luck with this? I was thinking of trying the same thing.
Not so much for animation, but in the hopes that I could vary the look of the characters as they find armor/items. Example: characters are made of 5 parts: head, torso, arms, & lower body. Finding armor could change the look of the torso, weapons could be mounted to the arms, and boots would change the lower body.
I'm just getting started with TGB (and awaiting that 1.1.1 update for the Mac), so I doubt I'll be the first to solve this problem. If I actually do make any progress with this, of course I'll let the community know.
-John
08/31/2006 (7:13 am)
Jason,Any luck with this? I was thinking of trying the same thing.
Not so much for animation, but in the hopes that I could vary the look of the characters as they find armor/items. Example: characters are made of 5 parts: head, torso, arms, & lower body. Finding armor could change the look of the torso, weapons could be mounted to the arms, and boots would change the lower body.
I'm just getting started with TGB (and awaiting that 1.1.1 update for the Mac), so I doubt I'll be the first to solve this problem. If I actually do make any progress with this, of course I'll let the community know.
-John
#8
Steve
08/31/2006 (7:18 am)
I also toyed around with doing the same thing - but it's probably hard to perfect. Although, the guy who made www.ragdollkungfu.com probably took advantage of the issue...Steve
#9
09/11/2006 (6:41 pm)
What do you think of using DTS objects over static sprites??
#10
EDIT:
I am not looking for the easy out of "of course, 3d objects are more complex than a .pgn" We want to hear from someone with real data from running a TGB level full of 3d objects as opposed to sprites..
11/17/2006 (12:15 am)
I have the same question as Phillip and Kevin.. are their advantage/disadvantages to using sprites over 3d models in TGB? I guess no one has fully played with the 3d models in TGB enough to judge this yet.. but rest assured I am working on doing so now :)EDIT:
I am not looking for the easy out of "of course, 3d objects are more complex than a .pgn" We want to hear from someone with real data from running a TGB level full of 3d objects as opposed to sprites..
Torque Owner Philip Mansfield
Default Studio Name
As T2D doesn't have pixel based collision, this type of system would be the next best thing for any number of games that need that type of collision (such as fighters).
I guess my main question would be what advantages would it offer over creating 3D models and using the DTS objects in T2D?
It would also depend on what level of granularity your skeleton would have. Would the arm be seperated into hand/lower arm/upper arm segments? Would there be some kind of IK so pulling the hand forward causes the rest of the arm to move as well?
This could be a great resource, but I'd wary of being quite complicated to get working properly and looking natural.