Hey Exanimates, We're working as hard as we can to get everything ready for the 0.9.5 update, but one of the new creatures we're introducing has proven a lot more problematic than we expected. This creature is larger and more oddly proportioned than anything we've done before, and this has caused a lot of problems with our physics driven character motion. Character physics have been our biggest development challenge, it took us literally years of experimenting, iterating and refining them before we got them to a point where most players were happy with how characters move and respond to inputs. This is immeasurably more complex than just an animation problem. Our character's movements are driven exclusively by physical forces and virtual muscles, they physically interact with everything, they must promptly react to inputs and move in any direction while also turning, performing actions, negotiating uneven ground, reacting to collisions and forces and trying to stay balanced, all at the same time. And that's just the basics, with many more specific and difficult problems we had to tackle. It's not just an animation or physics problem, the meat of it is how characters reactively decide when and how to move each part of their body, how much force to exert with each muscle in every moment to produce slick, smooth and plausibly human-like movement while also feeling very responsive to control. Humans are able to do this thanks to millions of years of evolution and a lifetime of practice in the real world. A program that can replicate much of this in a fast real time sim is no small feat. We started with something very clumsy, and while we hoped to improve it, we believed that players would be more forgiving of some of the clumsiness for everything that such a dynamic system has to offer. We know now that's not how things go. There's always room to improve many things, but after all this work the core physics that are the foundation of gameplay are undeniably in a good place and far better than we ever imagined they could be. Unfortunately, this new creature threw a huge spanner in the works of all this fine tuning we've done over the years. We've been focusing on teaching humans to move, and so far we've gotten away with just some broad adjustments for other humanoid characters. The odd proportions and even just the distribution of mass on this new creature really did not work well at all, and this was kind of terrifying as it's a very delicately balanced system that at this point feels more like magic than science. Any small change could throw everything out of whack, and we couldn't even be sure if getting this creature working is possible. Humans after all are real, bio-mechanically sound creatures, this is something less anatomically plausible. The new creature could barely even stand, it wasn't able to move its limbs properly and dashing and running around was far beyond its capabilities. We weren't just going to give up and cut it though, we had to try. We still want to add many more creatures, our plans include that creature types are very local and vary as we expand the world. We're even considering adding procedural elements to creatures as well, building on our already very modular and morphable character system. So we began the difficult and scary process of trying to understand all the reasons why it failed to move properly and do something about it. Our first problem was with basic muscle function, and this at least was relatively simple to solve. Instead of scaling all muscles while assuming human proportions we now measure parts of the body and scale each muscle accordingly. This worked well and it's fully automatic, adapting to any change in body structure for any individual character, which is great. Our next problem was definitely on the scary side: footwork. All the fast and completely dynamic work of movement and staying balanced in combat while performing actions. This was the magic sauce that we were so scared of spoiling, something we arrived at through years of trial and error, sometimes working on a promising idea for months only to discard it when we just couldn't get it to work, and sometimes discovering improvements more by accident than intent. Our new creature wasn't impressed, and so we warily tried to unravel the magic and figure out exactly why some things that worked so well didn't here. Eventually we transformed enough of the magic into exact math that took the specific proportions of the body into account and got things working well without breaking anything. Again we had something that should just work better than ever with any creature. Big sigh of relief. Our final problem was with walking and running, which has proven much more difficult than we anticipated. We'd recently made some improvements for other creatures and thought this would be easy, but no matter what we do the new creature keeps falling, stumbling, dragging their feet and so on. Walking and running is the most artificial part of our character physics, unlike the entirely physics driven and dynamic combat movement where motions are generated on the fly, this is based on motions that are designed by hand to look natural, with some invisible forces and essentially hacks to help the character stay upright and moving. This seems like a simpler system, but it must still interact with physics and have dynamic behaviours, and the simplified part is where things went wrong. It's just too rigidly coded to our current animations, so the new creature with its weird long legs and animations just can't move effectively or stay balanced. We tried to just make adjustments, but nothing worked and it became clear the only solution was a more complete simulation, and this means a significant overhaul that we're still in the process of. We're now analysing the specific hand made animations as they apply to the character, comparing animation inputs to what's actually happening in the physics to make dynamic adjustments, predicting exactly where and when the feet should land and how high above the ground they should be as they move, how much each leg should push the character off the ground and propel them forward at any moment among other things. The creature now moves well on flat ground at least, and things are generally improved for everyone, the motions look more grounded and physical. We're still working on balancing on uneven ground, but we've done a lot of the big difficult stuff so hopefully we can fix the last few issues quickly. We do expect to see more general improvements, and as we keep adding more verticality to the world and more creatures these are important. Some of the things we developed here are also interesting for features we hope to add in future, such as more contextual movement and traversal animations and things like quick recoveries for losing balance or falling in different ways. Rewinding a little, before we could even test the new creature in game, we've been using our new in-house tools for rigging and skinning. You never know what's missing, needs fixing or improving until you use something, so there was a lot of that. While the tools are fairly specific for now, they're part of new tool systems we plan to use for many things yet and there was quite a lot of engine level support to getting the entire workflow in a good solid state. We also realised that there was one important step still missing from our character production process, which is modelling additional collision mesh for parts of the character such as the joints. Our tools aren't quite yet equipped for this kind of modelling, but we did one better and can now just generate this entirely automatically. At this point we can take our character artist's raw 3D models and create textures and materials, rig them, skin them, animate them, give them properties and behaviours and add them as fully functional in game characters without ever touching a 3rd party program or a line of code. We don't expect you to understand just how much of a difficult and confusing process this previously was, but this is a pretty big deal. We have also been working on some improvements to our procedural item and material systems, some of these were urgent, but we've also got other things still in progress and can probably wait until after the update. We're working on some improvements to how materials and parts of objects blend together, adding more advanced effects to character materials and also expanding the procedural materials to work not only as part of our procedural item system, but as a general engine feature. This means we can give any object dynamic procedural materials, we can use them while authoring items which is great for our new tools and we can "break" the procedural item rules as needed. If for example an armour had a lot of additional elements that don't fit in our character layouts, we could add "modules" with custom layouts. We already made our weapon layouts more versatile and added some new features for compound elements to support some new weapons. Speaking of which, we're adding a first batch of the new ornate procedural weapons that we've teased a few times on our discord. These will be rare loot, and once we have more types we will probably also add them as alternative tournament rewards. Here's a few examples of some in game generated outputs:
Have a great new year, Bare Mettle