Lightwave DTS Exporter problems and other related thoughts...
by Jukka Kokkonen · in Torque Game Engine · 03/21/2006 (3:21 pm) · 4 replies
I think the lw2dts exporter seems like a great tool (thanks to David Wyand for that)... But, then come the few issues... ;)
After doing some tests runs with the lw2dts exporter, the following problems arised:
1) A few crashes while running the "setup for IK". - This is not really a problem. Solved it by using some slightly different configurations. And I don't even remember anymore what exactly was causing these crashes.
2) Match goal orientation didn't seem to work. - This is not really a problem, baking IK to keys solves this one.
3) Weird results with bone animations??? The transformed vertices don't seem to move quite correctly in relation to the bones... Does the exporter perhaps require that the animation's 0 frame is the rest pose? (as in my test cases it's not). Or could it have something to do with such bones that are not connected to the parent bone? (when viewed in showtool pro, it seems that the exporter has added some extra "connecting" bones to those cases).
In any case, this problem causes the test animations to simply get screwed up. (Note: all bone weights should be normalized, vertices affected by multiple weightmaps changed to have only one weightmap, IK has been baked to keyframes and then disabled, and I have tried toggling most exporter/bone options on/off with no luck)
4) Source code is not available? (other than the old version?) - This is not a problem _personally_. Should I ever get something done with my personal indie torque license, I'll probably be sticking to whatever dts features I have readily available instead of adding my own improvements...
However, this is a big issue when considering the possible use of Torque (with Lightwave) in a game development company - The fact that lw2dts is not "a rock solid" exporter officially developed by garagegames, but something a bit unofficial and not yet quite 100% functional adds a risk to any company going for the Torque+Lightwave combination. What it means, that if the exporter has any flaws (which seems more likely for the unofficial exporters), it may be very hard to get past them once they appear.
Also, the possible flaws are not the only ones adding to the risk... Another obvious risk is the inability to add any new features to the dts models (if the exporter cannot be changed to support them). Basically, in my particular case, the dts models/exporter would likely be modified. In which case, if the exporter source code is not available, then the only way to go would be to modify the current in-house exporter to output torque dts format instead of its current format. This however, would somewhat start to eat up the whole idea of changing to torque... (What would be the point to change from a familiar in-house engine to unfamiliar torque and then just having to port the tools and stuff from previous engine to the new one - it's just added risk and added costs, with no gain.)
Of course there's always the possibility to change to torque and use Max (instead of lightwave), but that would be a very costly transition - not only in the hard cold cash needed for Max licensing, but also very costly in lost artists' experience. (If it takes several months for a several artists to get the same level of productivity with the new unfamiliar tools compared to the old ones, then we're talking about several $10k costs for the transition - something that is probably unacceptable)
(Note, this issue does not concern just the dts exporters... It's the same with dif exporters/tools... I wonder when the constructor is coming out and if it's going to be better than gtkradiant... ;)
Any thoughts about these? David? GG guys?
After doing some tests runs with the lw2dts exporter, the following problems arised:
1) A few crashes while running the "setup for IK". - This is not really a problem. Solved it by using some slightly different configurations. And I don't even remember anymore what exactly was causing these crashes.
2) Match goal orientation didn't seem to work. - This is not really a problem, baking IK to keys solves this one.
3) Weird results with bone animations??? The transformed vertices don't seem to move quite correctly in relation to the bones... Does the exporter perhaps require that the animation's 0 frame is the rest pose? (as in my test cases it's not). Or could it have something to do with such bones that are not connected to the parent bone? (when viewed in showtool pro, it seems that the exporter has added some extra "connecting" bones to those cases).
In any case, this problem causes the test animations to simply get screwed up. (Note: all bone weights should be normalized, vertices affected by multiple weightmaps changed to have only one weightmap, IK has been baked to keyframes and then disabled, and I have tried toggling most exporter/bone options on/off with no luck)
4) Source code is not available? (other than the old version?) - This is not a problem _personally_. Should I ever get something done with my personal indie torque license, I'll probably be sticking to whatever dts features I have readily available instead of adding my own improvements...
However, this is a big issue when considering the possible use of Torque (with Lightwave) in a game development company - The fact that lw2dts is not "a rock solid" exporter officially developed by garagegames, but something a bit unofficial and not yet quite 100% functional adds a risk to any company going for the Torque+Lightwave combination. What it means, that if the exporter has any flaws (which seems more likely for the unofficial exporters), it may be very hard to get past them once they appear.
Also, the possible flaws are not the only ones adding to the risk... Another obvious risk is the inability to add any new features to the dts models (if the exporter cannot be changed to support them). Basically, in my particular case, the dts models/exporter would likely be modified. In which case, if the exporter source code is not available, then the only way to go would be to modify the current in-house exporter to output torque dts format instead of its current format. This however, would somewhat start to eat up the whole idea of changing to torque... (What would be the point to change from a familiar in-house engine to unfamiliar torque and then just having to port the tools and stuff from previous engine to the new one - it's just added risk and added costs, with no gain.)
Of course there's always the possibility to change to torque and use Max (instead of lightwave), but that would be a very costly transition - not only in the hard cold cash needed for Max licensing, but also very costly in lost artists' experience. (If it takes several months for a several artists to get the same level of productivity with the new unfamiliar tools compared to the old ones, then we're talking about several $10k costs for the transition - something that is probably unacceptable)
(Note, this issue does not concern just the dts exporters... It's the same with dif exporters/tools... I wonder when the constructor is coming out and if it's going to be better than gtkradiant... ;)
Any thoughts about these? David? GG guys?
#2
I can't go into any details concerning the "company decisions" about possible torque licensing, etc. But let's just say that we are still very much evaluating different possibilities when it comes to the engine used by our future projects...
Personally, I think that torque is one of the most interesting candidates, mostly due to the low licensing fees. I'd guess the biggest issues are the tools (content-pipeline), TSE not being ready yet (a lot of added risk) and the cost/risk of transition to an unfamiliar engine.
Now, if it would be possible to eliminate the tool issues, then it would be 1 down, 2 more to go. ;)
03/22/2006 (3:40 am)
Thanks for the quick reply...I can't go into any details concerning the "company decisions" about possible torque licensing, etc. But let's just say that we are still very much evaluating different possibilities when it comes to the engine used by our future projects...
Personally, I think that torque is one of the most interesting candidates, mostly due to the low licensing fees. I'd guess the biggest issues are the tools (content-pipeline), TSE not being ready yet (a lot of added risk) and the cost/risk of transition to an unfamiliar engine.
Now, if it would be possible to eliminate the tool issues, then it would be 1 down, 2 more to go. ;)
#3
question, though- I see that it now supports blend animations, which is great because we're doing a lot of underwater plants that we're able to bang out real quick by setting up some blend sequences using Lightwave's Morph Mixer.
However, when exporting to DTS or DSQ, no matter what settings I'm checking in the exporter, the sequences are showing up but I see no movement when I test in ShowTool. Dave, obviously I'm being an idiot, but what's the proper way to get those blend animations out of LW and into Torque?
02/01/2007 (5:20 pm)
Also a really big fan of Dave's tool - question, though- I see that it now supports blend animations, which is great because we're doing a lot of underwater plants that we're able to bang out real quick by setting up some blend sequences using Lightwave's Morph Mixer.
However, when exporting to DTS or DSQ, no matter what settings I'm checking in the exporter, the sequences are showing up but I see no movement when I test in ShowTool. Dave, obviously I'm being an idiot, but what's the proper way to get those blend animations out of LW and into Torque?
#4
I've reconfigured the thing to work with bones. Put me down for someone interested in that as a new feature, though.
02/04/2007 (11:30 pm)
Ah-ha! Don't answer that... found deep in the cobwebbed annals of Gnometech's forums that morphmixer isn't supported.I've reconfigured the thing to work with bones. Put me down for someone interested in that as a new feature, though.
Associate David Wyand
Gnometech Inc.
I'll try to answer these in order:
1. Hmm, not sure what would be happening there. The 'setup for IK' process is quite straight forward, code wise. If you do manage to reproduce it, please pass the procedure along.
2. Unfortunately the LightWave SDK does not pass along Match Goal Orientation information. Effectively this means that a plug-in does not receive the correct item rotation when this option is enabled. I have a very trivial case workaround in the exporter right now, but it is in no way robust. I had always planned on going back and working on it, but it has been a couple of years now. Hopefully LightWave 9 will address this long-standing issue with the SDK.
3. Off hand I'm not sure what could be causing this. Something we've found is that you need to make sure you're making correct use of the Unaffected by IK of Descendants flag for the LightWave SDK to correctly pass along IK information to a plug-in. Without this flag set, everything will look fine in LightWave, but a plug-in does not receive the correct rotations.
It also never hurts to have your root pose in frame 0. Although if you have your rest positions set up correctly in the bones, and you are only making use of bone-based animation (ie: not exporting an animating NULL) then generally you're OK with not doing that.
4. You are correct in that there is no recent source code publicly available. And as of right now, there probably won't be. The intent is to make a 'pro' version for sale sometime in the future. As to when, I'm not sure. There's too much Constructing going on right now. :o) This would follow the planned path for pro versions of the Max and Maya exporters that were mentioned at last year's IGC.
Now, if there is a company looking to purchase a license to the LW exporter source code, then I'm sure something could be worked out. This would likely be true for any of the tools for sale by GG, although I cannot speak for them. I would certainly allow it for TST Pro, for example.
And if a company is looking for LightWave, Torque and DTS file format help, I could be made available for contracting. :o)
- LightWave Dave