eromancer
patreon
Inner Circle Update for November 28 2017!
🕑 Added 2017-11-28 17:44:53 +0000 UTC Hey guys!I’m devoting this entire post to man jelly! How exciting! >_>
To start things off, I made a short video showing our fluid simulation working in Houdini (actual simulation time was about 5-10 minutes… this is playing back a cached version). We’re finally to the point with our simulation and processing workflow where we can begin integrating it into the new art style. Before I get into the details though, check out the end result (check here for a full resolution version)!
After pulling the trigger on scene-specific lighting for battle, we had decided we needed to put maximum priority on getting 3D liquid working in order to have it for battle portraits for v0.06. So, in our last post, we left off with Ubercharge starting his journey into Houdini. His goal has been to create a procedural system and workflow that will allow us to, as simply as possible, pop a scene from Daz or Maya into Houdini and start spraying buckets o' jizz errywhere.
Houdini is a node-based program for procedural generation of 3D geometry and effects. Fluids, smoke, and explosions are some popular uses, but it's a very extensive and flexible program. I've also spent a lot of time this week learning as much as I can, not only because I'll be the one using Uber's system, but because I intend to create battle FX animations (skill / damage FX, etc.) for the new art style using it. TK, Uber, and I have a huge video tutorial course we're plowing through, but the fluid stuff has proven to be our trial by fire.
There have been a number of challenges we had to overcome to get to this stage, and a few more on the horizon still to get it really good looking. The first was the jizz material, which after literally days of experimenting in different lighting situations and sending it back and forth Uber and I have it looking pretty good. The second challenge was collision for the fluid simulation. Uber discovered that for proper behavior, the collision model needs to have actual volume (most clothes for characters are just planes, meaning they are infinitely thin). Currently we're fixing it on a case by case basis by extrusion, but an ideal solution would be to have a process for creating closed models similar to what's required for 3D printing. Daz has to rear it's ugly head somewhere in everything we do, and here's no exception. We found a lovely little bug with the rendering engine where when a refractive object (like, pretty much every liquid) clips with another object you get nasty shadow artifacts. This is, of course, bad when you want to cover your characters in goop. Uber's solution was to add a boolean component to his liquid system that carves out the liquid geometry in the shape of the collision object, meaning liquid perfectly fits on top of whatever it's covering (and thus, no more clipping :D).
Here's the culmination of Uber's efforts so far, complete with a procedural bubble generation system! Please mind that there are still a bunch of experimental nodes and dead-ends in that diagram.
A couple of things that you aren't seeing in the example scene that we plan on improving upon are general wetness and stringy/drippy/elastic liquid. For the former, we are considering using an alternate set of "wet" materials for characters/clothes and using the render manager I made awhile back to simply pump out a wet and dry render when necessary, then blend between them in Photoshop (this is basically what I did with the old art style). For the latter, we decided that using actual elastic liquid simulation to get the drippy effects we want in some cases is simply way too time consuming. Not only does simulation time explode, but it's really tough to get right on the first (or first dozen) tries. Instead, Uber plans on modeling a set of shapes that we can simply merge into our simulated liquid meshes and easily edit to perfection. He created this test model in just a few minutes, which could easily be edited for use over and over when we need an effect like this.
That's about all I have to say about splooge for now :0. TK has finished Malise's suit morphs and rigging, and has made progress on armor damage as well, so hopefully in the next post I'll have some images to show!
Comments
Jamie C.
I agree with Arnold the story and environment hooked me! I like the other bits but definitely story and your guys amazing world spaces. :D its a fun game and should never be a one time deal it should be like bacon or cake you never get enough.
Only fan base wishful thinking at this point. It would be very nice though. Maybe we should be making up names for MATM 2. MATM 2: A New Stretch.
malise and the machine 2 in unreal engine confirmed?
Thank you for that extremely hot picture of Malise, and for the hard work that you guys are putting in every day. It's great to see these kinds of progress being made with each update. I share the same level of eagerness for v0.06 but I know it will take quite a while more for you guys to get everything pieced together. So keep up the great work, and remember don't burn out yourself... keep things at a steady, manageable pace and I look forward to a great game, even if it means another few months in the works. I only have one wish though and that is to not only have awesome graphics but also a compelling storyline that keeps players glued to it. I know it's not an easy task having to do graphics, animations, coding and storytelling when you guys are working with a very lean team. Let me know if you need any form of help, I'll be glad to chip in in anyway I can!
DCLXVI
I can't get over it. A game like this needs animations. i would not mind waiting 10 more years if it meant animations would be included...
Not in this detail, and by particle simulation standards this is a pretty simple example. Granted, we haven't been doing it long enough to know all the optimization tricks, but the one in the video took between 5 and 10 minutes to simulate, and then another minute to mesh a single frame (I actually cut it short before all particles came to rest -- the one used in the actual image with Malise probably only took two minutes before I cut it). It could however be pre-baked into an animation to be used in a 3D game engine.
Still not sure yet, but it'll be easier to gauge once the actual remaining content is the only thing we have left to knock out. The stuff in this post is a good example of something that had to be done earlier than expected, and until battle and the interface is working in the new art style I can't say there won't be more things that could throw my estimate off. I'd like to have an inner circle test release once we get battle put together though.
I'm surprised Youtube hasn't busted that clip yet for the pose. The simulation "rain" is probably okay by their standards though.
Would it be possible to use this stuff realtime in the Unreal Engine?
Not your bussines
Pretty cool stuff, from a technical standpoint the project grows evergoing more intense. I know this gets kinda old but could you give us a approximate date for the release of v0.06? Seeing all these really cool advances just keeps growing my hunger further and further for finally seeing all these things playable and used. Also that new look for Malise is just amazing, can't get enough of her!
