Jump to content

Louhikarme's Modeling Madness


From Concepts to Game Asset

Posted by Louhikarme , 09 March 2017 · 2,039 views
blender, substance painter2 and 4 more...
From Concepts to Game Asset

On the video below, I talk about how we are creating evil bots for single-player mode. It outlines the main principles and techniques and has some tidbits on certain obstacles we face while bringing these things to life.


I start with discussing the specs we have in the Game Design Document (GDD), followed by an initial meeting on what the bot should look like. Then I do quick drafts using just primitive shapes to get the shape what we are going for. This process usually takes a few to several days, depending on how many modifications are needed.


Once we are happy with the basic shape of bot, I move into "high-poly" (high polygon count) modeling- adding details and the actual look to the bot. Also at this stage, I color code the materials onto the high-poly model for the texturing phase.


After that, I do the actual game asset- this means reducing the number of polys (polygons) as much as possible while still keeping the shape of the bot. I forgot to mention at this point in the video that I'll copy this mesh (3D computer model) to become a "cage mesh" for the texture "bakes". A cage file basically covers the entire high-poly model altogether. The actual game mesh, on the other hand, will have some places going slightly through the high-poly model.


Once I have that finished I "bake" (pre-calculate) the textures in xnormal, colorid, normal, ambient occulsion and curvature for use in the Substance Painter software. In Substance Painter, this colorid approach helps me to set up initial materials for the model quickly and I can start the actual texture detailing faster.
Another stage in the process is "rigging," which means placing "bones" in the mesh to allow it to move. I have two sets of bones, the actual bones that will deform the mesh and bones that I use for animating. The animating bones have constraints on them so they only move the way they should, no matter what I try to do them.


Finally, in Unreal Engine 4 (UE4) everything starts to come to life. I make an animation blueprint that lays the groundwork for Tyler. He can then add the code to call (trigger) the animations when needed in-game.







Road to SimPit - Behind the Scenes

Posted by Louhikarme , 07 September 2016 · 2,525 views
Descent Underground, 3d modeling and 2 more...
Road to SimPit - Behind the Scenes

In its current form, Descent: Underground uses unmanned drone ships (for now, more on that in Single player). For VR purposes though, it is best practices to have a cockpit.
From the idea side, things were easy, just use the existing hud and make a cockpit/Simpit around it. However, our existing HUD is a single plane that’s attached to the player's camera view. Having a 3D model just slapped on the front doesn't not look good when you would turn your head in VR mode. Without VR you could have pulled this off, though.

It comes down to perspective, and when you have single flat plane that houses all the information, you need to get sneaky and cheat. That means, you need 3D objects that don’t skew when you turn around and yet look normal to the eye. That is why the first iteration of the cockpit looks rather wonky from outside. The back wall also was something that was way too big in retrospect (more on that later).
Initial planning included lots of reference pictures from existing cockpits, bomber turrets, etc. The process in itself was simple- first create a few different mockups of the would-be cockpit layout. Then, I'd show them to the team and get feedback of what worked and what needed changing... maybe.


So, once there were a few iterations and variations, we agreed upon a very simplistic approach that wouldn't obscure the player's view. The key was to give screen space as much as possible. Even then, it took a couple more variations to come up something that would still look good. On the texture side there were also a few alterations. The first one was way too bland and didn't have depth!


Even once this one was finished, we knew this was always going to be a placeholder cockpit. We are still going to break the HUD down into separate pieces and have a much better one...


Behold the Simpit v2:
Attached Image: cockpit_v2_9.png
Now to start off with, this was that the HUD elements themselves wouldn’t be changed, or at least very little. So this time around I was able to approach this like screen panels. I made a 3D model around the HUD elements matching the size of them. I had placeholders for those and started moving the panels in their places in real 3D and this time I had the help of the Miner as a reference for the size and placement.
As we are pushing towards a working radar, its base now needed to be more prominent in the HUD. Once again, I made a couple versions and the back wall got more attention to be more like a cockpit. The screen stayed though, something for people to snap their neck trying to look at.


Once that was done, I moved to create a rig for it so it would work in the world properly. In this case the rig is pretty simple, however there were a few for programming things to do. I had to make sure the panel placement’s was set up so it would be easier for Tyler to position everything, including the radar and reticle points.


Attached Image: cockpit_v2_8.png


In both versions, the little details were painted using a normal map, since adding bolts and stuff that way is faster than actually making the geometry for them. I just needed to plan ahead on what will be geometry and what I would paint. In this case, since the player’s head is the same distance from everything, it works better with a normal map.


Once the actual model was ready, textured, and pushed into the editor, it was then time to hand it to Tyler. He's going to do his programming magic and actually add the HUD elements and make it all work.


But wait, things don't end there!


Small things matter, things that make you think "something is off but I can't really put my finger on it". If they are there, you know things are right. So here is a really early prototype version of having the glass fracture when you get hit. I can't stress enough that this is very early, and might not even make it in the game if we see it hurting framerates too badly. So, the very first view on the prototype:


Thanks for being a part of Descent: Underground with us!



April 2019

14151617 18 1920

Recent Entries

Recent Comments