New default character discussion

Level design, models, animations, physics, etc.
34 posts Page 1 of 4
Duion
Posts: 1626
Joined: Sun Feb 08, 2015 1:51 am
 
by Duion » Sun Dec 15, 2019 10:56 am
I started a new thread about making a new default character, it has been obvious for many years now that we need a replacement for the default soldier and make a new better usable more generic default character people can use.

This topic was already worked on like 5 years ago, when the steering committee still existed, but the person working on it bailed out and after that the steering committee was abandoned as well as most of the plans that were worked on till then.

I already started the progress of making a new default character, but then I figured out it may not be good enough and may not be the best workflow.

If you just make a dummy character as a placeholder, people will use that and will have a hard time replacing it with something usable later, since you likely cannot re-use anything of it.

Then I figured out making a new default character is not so much about the character itself, but about the implementation, so the new default character should not be something fancy, just a placeholder, but what will be important is that the implementation is made well to show people how it is done and that it can be extended upon, so people can use the prototype to turn into a fully usable character later.

So here was my idea so far:

1. Use Makehuman for the character generation entirely, the advantage with that will be that everyone will be able to create a new character to their liking in Makehuman, so anyone without any modeling skills will be able to replace the character in an instant.

2. Use good settings in Makehuman, in Makehuman there are many types of mesh and rig you can use, but I figured it is best to use the high polygon model as well as the high bone rig. Those will be not good for performance, but it is far easier to remove polygons and bones as it is to add them. I think the model will be around 15 000 polygons, which is roughly the same as the new default soldier anyway, the default soldier has a lot of junk on him, like armor platings, weapons and such which increase the polygon count. So even using the high polygon mesh Makehuman will not increase polygon compared to what we have now. From then on everyone needs to use the same settings when making their characters in Makehuman.

3. Do the implementation, give people a Blender file with everything already set up, like LOD levels, animation editor and all the other stuff you may need. The problem with Makehuman is, it has all kinds of export options, except one that fits Torque3D. What Torque3D needs is metric system and character facing the positive Y-axis. I already made a feature request on Makehuman, but no idea if they implemented it already, until then we have to rotate it manually, which can be tricky, since rotating a skeleton can go wonky if you do it wrong and turn your character into a giant blob.

4. Now we should have a character in Torque3D, without animations, so now we need to add a default set of animations using the default rig, so everyone having made their individual character in Makehuman should be able to link the pre-made animations to his character. The animations should be exported individually, like it is done in Torque3D now, so we can build a library of them and everyone can use them.

5. We need to figure out a way to optimize the models, meaning a way to put the textures on one atlas texture, since Makehuman exports them on many individual textures. What I did was to bake all textures on one atlas texture, but what that does is, it makes it unusable for other people, since everything is already fixed and slight changes on the character would mean to redo all the textures. Therefore I suggest keeping the character naked and overlaying the clothes, this adds polygons, but makes things much easier to use.


So I think that is some kind of ideal workflow and it does not having to do any actual modeling work, which is probably the hardest part. The problem with characters however is, you need to get the initial system right, since if you have to change it later, you have to change everything. Imagine having a rig where bones are missing and you need to add them later and then you need to do the weightpainting again and need to do all animations again. Or is just use rename one bone, you may need to do it in every animation as well and/or retarted the animations.

The most important part of the new default character is not the character itself, but to decide on a standard on how to do it, so it can be used modular, which is not possible with the current one. I already had some supposed dummy default characters opened in Blender that others made, but even having many years of experience in Blender and modeling I had absolutely no idea how to use it and in most cases probably even the creators themselves did not know how to use it, since they did not even had it implemented into a game where it was in fully working state with scripts animations LOD etc. I downloaded almost every free player model there is already and almost everything was completely useless garbage to me, since it was not how I do modeling and it was not how it has to be done for games. Therefore having a default way of doing things ist most important, as well as having it modular, so not everyone who wants to do a change in the character has to know every skill there is to make a fully functioning character.

So what do you think? If possible keep this discussion within people who actually know how to make fully working game characters, otherwise we will not get anywhere like in previous attempts.
Bishop
Posts: 98
Joined: Wed Dec 09, 2015 2:30 am
by Bishop » Sun Dec 15, 2019 5:57 pm
Yes,...this would be great ... are you already working on this?...any pictures from the development?
...there will be no problem with character base mesh? - faces flows with Makehuman model?.... i know it is not ideal for use for a realtime....many useless Ngons.
TorqueFan
Posts: 130
Joined: Thu Apr 30, 2015 5:35 am
by TorqueFan » Sun Dec 15, 2019 6:32 pm
Alright! Now we're onto something. As I eluded to in the other thread where this discussion started, I'm already knee deep in a tutorial about all of this stuff so I will be putting that out there real soon. More on that in a minute. For now, I'd like to thank you for taking an interest in this topic and going so far as to provide an outline of ideas. I can already see that you will be a valuable ally in the struggle to document all of this for our beautiful new peoples!

Now I do have a small difference in opinion about the method by which we should present this information. I really feel that we need to present the information in a smaller, bite-sized fashion(which is exactly what I'm typing up now) so that it is easier to digest. I do think you are all over this, Duion, and what you are suggesting here all fits hand in glove with what I'm proposing. While we may have a difference in opinion and workflows(and honestly you have some great ideas around workflow), the point here is that the relevant information is reaching newcomers in an easy to read fashion. Let's not allow small things like this create a barrier between us, because when it's all said and done we both want the same thing for this Community: an easy way to approach the existing Player object system in Torque.

Now I know when I proposed this idea viewtopic.php?f=10&t=1737&start=10#p12631 I suggested a new Default Soldier model. Honestly I still DO think we need this, but now that I've gotten deeper into the documentation creation process I have realized that not only do we need that replacement Soldier but we also need a kickass series of tutorials that build up to the level of expertise that you have to offer. This means that while I'm here pegging away at this series you can be preparing all of what you are suggesting at the same time and allow me to lay a little bit of groundwork for you that leads into that new Default Soldier.

It's not really a proposition at this point, since I'm already hard at work this weekend on this new tutorial series(Part I), but I do want to emphasize here that readability is key. Rather than taking all of these advanced things into consideration when first starting out with Torque, I feel like the information should be super fun for the newcomer to start out. Advancing through the tutorials, the information gets more and more complex, until ultimately the new user is ready to tackle this here what you are talking about. See there how we can work together and make this something bigger than we would have each by ourselves? I am confident that this will fit together very well, leading the newcomer through a series and then ultimately engaging them full on with the Blender and new Soldier process. This way the user is having fun and seeing right away how easy it can be to get animated Players moving in Torque and THEN when they are full well and ready they will be prepared to tackle the more advanced things you are talking about. Make sense?

Please let me reiterate that I want to work together with you on this. I also believe that you have a wealth of knowledge just bleeding to be shared so please let's try and see eye to eye on this. When it's all over with, we'll have all of my ideas AND all of yours in a readable format for our Community!

1- I'm starting out with mixamo. I'm going to show people how to straight up use the entire library there as an introduction to the Player animation system. What we'll end up with here is a system for newcomers to follow that is super easy and fast and fun to get rolling with.

2- As the user's level of experience improves(because we are actively teaching, as it were), we will slowly introduce topics such as movement triggers, weapon states, nodes, and so on.

3 - Eventually, once the user has been prepared by this process, they will be able to digest all of the stuff you have above. Which I agree with all 100%.

Anyways, as stated, this tutorial is already in motion. I'm going back to it now ;)
Duion
Posts: 1626
Joined: Sun Feb 08, 2015 1:51 am
 
by Duion » Sun Dec 15, 2019 6:40 pm
@ Bishop

There is not so much anything to work on, but to decide on how to do it, I already explained that, everyone will be able to create their own character in Makehuman or use a default one someone else made with generic features.

I may add that those people who did not already make characters, they should try getting one from Makehuman into Torque and see where the problems are. It actually is not that hard, anyone should be able to do it, but the bigger problems arise from the implementation. We are not going to do any modeling for now.

You can use my character as a tutorial if you want: https://duion.com/temp-stuff
But as you can see I removed the mesh and textures except maybe the hands and feet and made my own, you can use it, but it is a paintballer and not a generic character. The floating bones on the outside are helper bones for inverse kinematics, it helps you animate, but they are not necessary I also forgot how exactly I used them or intented to use them.
Duion
Posts: 1626
Joined: Sun Feb 08, 2015 1:51 am
 
by Duion » Sun Dec 15, 2019 6:51 pm
@
User avatar
TorqueFan

I strongly advice against using any proprietary third party that has to be paid for or registered for or anything like that. Even if it is "easier" and has more animations etc already, that is not going to be what we need. If we use Makehuman we have a free and open source solution that anyone can use and based on that we can interchange out progress.
Animations are not that big of a deal anyway, animations for FPS games are not what most people think, they often only consist of a few frames that get interpolated. My most complex animation is the run cycle, which is 11 frames, some animations are even as simple as 2 frames and they often need to be exactly symmetric and looping, meaning that the 11 frames animation is in reality 12 frames, but the first and the last frame are identical. Most animations from libraries are useless, since they are not perfectly symmetric and not perfectly looping, so you spend more time fixing that as it would take to make a new one from scratch. Of course motion captured are more realistic, but it depends on your end goal and for the most part hand made animations using a few frames are just fine.

If you use Makehuman you already should be able to use my animations and the more people use it, the more of a library we have to share and eventually we can contribute back to the Makehuman community and in return get some more support from them.
TorqueFan
Posts: 130
Joined: Thu Apr 30, 2015 5:35 am
by TorqueFan » Sun Dec 15, 2019 7:01 pm
Mixamo is free, with optional paid services. Additionally all of that is used as an introductory process to Torque animation, which is exactly what mixamo was designed for: learning.

Again, I find you speaking to me as if I am unaware of the material involved with the topic of discussion. It's unfortunate that you really do make it so very hard to work with you. You throw around terms like 'useless' and always introduce things to folks as if they had never heard of it before and you are shedding light on something that was hidden.

I'm not attempting to throw a new user headfirst into a character modeling and Torque hook-in process. The mixamo step is for everyone to get moving right away(again as a beginner) and get to the point that they can do those things. THIS is the underlying issue with the existing documentation that is scattered about, there is no mediator between the experienced user and the new one.
Duion
Posts: 1626
Joined: Sun Feb 08, 2015 1:51 am
 
by Duion » Sun Dec 15, 2019 7:12 pm
"Free" means "not free", we have an open source engine and additions to the engine also need to be open source, which is what proprietary software cannot provide. The main point is, that anything done with Mixamo can not be integrated into the open source engine, since their models and animations are proprietary. Stuff exported from Makehuman however is public domain, means you can do anything with it.

I'm not trying to sabotage anything, I just did my homework. Sure you can use Mixamo to show people, but you cannot actually use anything from it, because of their copyright. So if you make a character and animations with Mixamo, it can never be added to the main engine repo. The licenses for tools are different and some actually allow to use their export, but the last time I looked there was no proprietary software that allowed that.
TorqueFan
Posts: 130
Joined: Thu Apr 30, 2015 5:35 am
by TorqueFan » Sun Dec 15, 2019 7:18 pm
I believe you might be misinterpreting my intentions with using mixamo here. The mixamo workflow I wanted to introduce was for exactly what I am doing, making a tutorial and introducing folks to Torque in a fun and engaging way. I've repeated myself on this point many times above and really don't think you've grasped it yet. The mixamo process is an INTRODUCTION. Not meant to be kept or cherished or integrated into the engine. Used exactly as it was designed to be used, as a learning tool. I went over all of this above, how this learning process could ultimately lead up to creating your own model, which then yes you wouldn't want to include mixamo at all for.

If you can have fun doing something, chances are much more likely you are going to stick with it. But if you can't get to the point that you are able to model/animate/script a character just jumping headfirst into that process can be a huge deterrent.

I also wanted to add, once again, that I am not trying to say let's use one over the other. Read carefully my wall of text above. I want both!
Duion
Posts: 1626
Joined: Sun Feb 08, 2015 1:51 am
 
by Duion » Sun Dec 15, 2019 7:29 pm
We are talking here about something people will be actually able to use, not about fun or making a tutorial and also not about creating your own model.

We will be using Blender and Makehuman, that's it, that is what most people will use since it is free and open source and there competitive open source tools out there other than those.

I already made a fully working character in a fully working game that is released on the market and I gave you the source files, so we already have something to show people how it is done. The debate now is about deciding on a standard so we can exchange our progress and Mixamo is definitively not an option for that.
TorqueFan
Posts: 130
Joined: Thu Apr 30, 2015 5:35 am
by TorqueFan » Sun Dec 15, 2019 7:32 pm
Also don't spread misinformation. Mixamo really is free, unless you want to use Fuse, which in our case we won't.

Additionally, yes this IS 100% full well about learning and a tutorial.

I find it hillarious how I took the initiative to begin a new tutorial series geared towards helping newcomers, posted on the boards I was being proactive on this front, and you ran off to try and 'head me off' and put up a post about it prior to me releasing the tutorial. Now that has evolved to you pasting up a link of some old model you created and calling that a tutorial. That is underhanded to say the least. I've tried to be civil with you and invite you to join me on this path so that we can share your ideas and mine with the Community but you have made it clear that won't be possible. The evidence is in this thread. Your reputation precedes you, Duion, so I'm just gonna let that be that.

I'll release my tutorial, if not but for the simple fact that you are attempting to discourage it.
Last edited by TorqueFan on Sun Dec 15, 2019 7:59 pm, edited 1 time in total.
34 posts Page 1 of 4

Who is online

Users browsing this forum: No registered users and 1 guest