I’m working on a prototype of some systems for a small team project where we will be ultimately needing one Muy Grande (Epicly) sized monster (about 60m x 30m x 100m). There will only be one of these, but there is a potential that it will be fully visible during gameplay. I’m wondering what the recommended approach for such a creature would be. Should we use separate meshes for performance? If so how do we animate it – animate it in sections? I’m just looking for some pointers and fundamental or commonly occuring problems with doing something like this. Any information or tips on this would be greatly appreciated. Thanks in advance.
The challenge here is collision; something of that size is going to require a whole lot more than simplified sphyl collision to work properly.
I do something similar in my game, but the advantage on my end is that it’s a giant robot, meaning it can afford to be segmented. I use a placeholder skeletal mesh to animate, but then I make this mesh invisible and instead attach static meshes directly to the bone sockets of the skeleton in-game. So the skeleton itself just drives around a lot of collision-capable geometry. For something that needs to vertex-deform IDK how you’d do it.
I would tend to keep the creature one mesh. If you are doing something this large, you are essentially the defacto test team on this but unless it’s armor or clothing or weapons I think the creature may be better in one piece. The standard practice is to animate a low poly count creature though when you do, and switch out to the “high” resolution creature.
In other words I don’t see that you would approach character animation differently for a huge creature than for a small creature. Create your mesh, have different level of detail models that you can swap out to. Animate it in one place creating your basic list of moves, walk, rage, run, jump, attack etc.
Have another team work on testing visibility in Unreal and what happens when a creature is this large. I would recommend creating an AI creature that does something simple, like find the player character. Tesla has some good tutorials on this that are very short on youtube.
However I would take the AI creature and scale it up to the size you anticipate needing. It will be a default blue guy character but you can swap out for your creature later.
The next step as you finish your model is to take your model and turn off the blue guy, then turn on your creature in the parenting of the AI.
Try different versions. See what happens to the UVs and the textures and the lighting. Don’t worry at this stage about animation, allow your animator to keep working while someone tests the size and the basic AI.
When the animation is ready then do your tests with animation to see how it goes, but you will gain valuable insights and solve problems with world scales, how large something is and what clipping planes you might run into with the stand in model and the blue guy animation.
If you have clothing or armour you can always add those things additionally in Unreal later, however I always use clothes and armor to reduce the size of a model. Get rid of any polygons that you will definitely not see and keep it one mesh if this doesn’t make the creature super heavy.
A semi-decent way to approach this is to just make a small character, animate, rig, and all that goodness in just a basic, player-sized state. It’ll make observing problems a lot easier than trying to zoom around a building sized monster to check for problems. When you feel like you’re good, then just scale it up inside your blueprint/C++ and Unreal do most all the hard work for you. You can then fine tune it, scale back down do some more work, then re-scale up, rinse and repeat.
That is how I’d approach it.