ManaTech Dev Blog @manatech - Tumblr Blog

Hiatus

This project will be on hold while we pursue other outlets here at IDEK Studios.

Pictures of the new implemented character model, as well as examples of some of its issues.

#Manapunk #Manatech #Dev #unity3d #mecanim #Pictures yay #animation #character

Status Update 015

The new character model has been implemented as the player, and the animation controller for basic movement has been set up. Unfortunately, it seems apparent that I will have to redo the player model and its animations before attempting any sort of release, as there are several problems with the current model that I did not foresee when I made it last year.

The main issues are as follows

The model does not use a true "t-pose". Those of you who know what that is can clearly see this in some of my older posts. For those of you who don't know what a "t-pose" is, Google is your friend. I had not really bothered to create a t-pose when modeling it as it wasn't really an issue with the animation project it was originally used it, but apparently Unity's Mecanim system is very particular about that, and its efforts to do so chafed against mine, and resulted in a very confusing mess of animations that don't match up on what are suppose dot be the same models but technically aren't.

While the model contains IK handles and BlendShapes that worked in well in Maya, I don't think they were properly configured for usage in Unity, as they have exhibited very strange behaviors within the Mecanim system.

The model itself was built to be as low-res as possible while still looking decent with the clothes the animated character was wearing. It's proving to not be enough, and some bits about it are still bothering me, particularly his head.

I'm not as bummed-out about this turn of events as I had thought I would be. Though it was sort of a failure, it went much better than I expected it to, and I will have to create a lot more models in the future for NPCs, particularly the male/female models, and either different preset body types or using BlendShapes for a character customization system. I'm not yet sure which would work best.

In other news, I managed to fix the "spiderman slide" glitch! I added a raycast that checked if there was something underneath the player in addition to the collision. While doing this I learned two things, one, there is another glitch that actually relates more to spiderman, where the player is stuck on the wall if he holds down the jump button (specifically, he cannot fall, but can slide along the length of the surface), and two, there was another bug that has apparently been around since I originally set up the PCC in which the character could not jump if he idled for too long. Apparently it was because the rigidbody was falling asleep -_-. Of course, it was very easy to fix once I realized that, but that didn't happen for a while.

I was fiddling with the design for the dual contouring implementation, and have concluded that the problem is in fact with the formula that I had based the entire program on, and that I had expected to work because it came directly from a paper dedicated to the subject. My best bet to find a solution is to try and find a comprehensive linear algebra class/library or whatever that is written in c# and that can perform a least squares approximation on matrices, because that's basically what needs to be done.

tl;dr

The player model has been updated and he has basic movement animations, but will need to be redone again eventually. I fixed the "spiderman slide" and learned that there is another spider-related glitch and fixed a sleeping rigidbody problem. Also, i need to find a linear algebra library to fix the terrain engine.

#Manatech #Manapunk #Dev #unity3d #mecanim #Status Update #animation #glitches

Was trying to find something else and ended up with this ancient thing. This was the initial design for the main character in Manatech (at that point called Tiny RPG) and is what led to the midget that is currently acting as the test character model. Wow, those are really short legs.

Glitches

Since I apparently never spoke about them before SU 014, there are two main glitches/bugs in ManaTech at the moment, the Crowd Seizure and the Spider-man Slide. Despite their names, they are not terrible dance moves, though I imagine that if someone where to name a dance as such, it would defientely be terrible.

Ahem.

The "Crowd Seizure" is an unfortunate phenomenon that seems to happen when a large number of rigidbodies using the PCM (Physics Character Motor) are close together. The glitch itself is that, in this situation, they are unable to move properly, and instead spazz out like they're having seizures. I don't really know what's causing this, though my guess is that it has more to do with the physics engine than the script. Hopefully I'm wrong about that.

The "Spider-man Slide" is fun, but would like be a game breaker because of what it enables you to do. Basically, the section of the PCC (Physics Character Controller) that regulates jumping, is currently set to just check for collisions and then assume its on the ground. As such, the glitch allows you to "jump" indefinitely up walls, giving the appearance of a slide. Unlike the Crowd Seizure, this one is merely a result of a lack of foresight, and continues to be present because of sheer laziness.

Status Update 014

Switching gears away from the terrain system, as I seem to have run into a wall with the code. It's in the troubleshooting phase and I fear that the formula being used to convert the values extracted from the density function and its gradient into the evaluated vertices for the mesh generation is in some way flawed; the values being returned are absurd, and while the mesh technically renders, it is somewhere else and I can only seem to find it in the preview window. When the outputs are supposed to be between 0 and 1 and you're ending with values from 0 all the way to 50 and in some weird cases over 2000, something is obviously wrong. Until I feel like taking another crack at it, I will be focusing more on the player and other gameplay mechanics. While destructible terrain is awesome, its not currently necessary at all.

On the mechanics front

the old character model is still being used.

The strange "crowd seizure" glitch is still an unsolved mystery.

The "spider-man slide" still happens as well.

The new character model's avatar is configured, but only the run forward and strafe left/right animations, along with two auxilary idle animations, are done. Still need a basic idle animation, a backwards run, jumping, and landing. More will of course be needed later, but those are the only ones needed that the system can currently support

The To-Do list for the near-immediate future

Create the needed Idle, backwards run, jumping, and landing animations for the NHM (New Human Model)

Creating a new player Gameobject that uses the NHM with the necessary tweaks

Create a basic enemy and the structure required (probably starting with an Angry Cube of Death)

Create health and damage system

Create crouching, walking, and basic attack animations (unarmed ready, punch, 1H sword swing)

Implement crouching, walking, and basic attack animations

Inventory system

Equipment system

Making sure weapons damage things properly

Create a shield or two

Create blocking animations for unarmed, 1H sword, and shield

Implement said animations

Be able to equip shields

Blocking system

Begin working on a basic hostile human AI

Teach it how to pick up, equip, and attack things

Teach it how to block

Next tasks will likely be related to ranged combat, which may require some additional camera scripting to go with the aiming system. After that is probably climbing.

Oh yeah, I also got a laptop, which will probably help with things.

Status Update 013

The terrain system is essentially done conceptually, now I must face the reality that I don’t really know what I’m doing, and it’s making me hesitant. I have a rudimentary set of scripts I’m working on that should let me visualize the meshes I’m generating, though I’m not sure how well it will perform. If it is not, it would be best to write it for the GPU, parallel processing is quite good for what I need done as it is basically evaluating thousands of points through the same process. Though since it’s not quite that fun, I’ve been waffling between it and trying to work to a point where I can start messing with combat.

Before that can be done however, I’ll need to set up the new animations and port the player model to the new character model. I’ve been having a bit of an issue with its colliders though, I have some set up for the limbs that are needed for the ragdoll and to provide a more accurate collision set up and hitbox. Unfortunately, the movement system is built for rigidbodies so that they can rotate freely and be at the mercy of physics. The rigidbody has to be able to touch the ground but not collide with the other colliders within the character. On the surface seems simple, but I haven’t found a way to really do this in Unity. Perhaps I’m not looking hard enough.

Another issue is of course the AI, which right now is semi-capable of moving towards a dynamic or static object in as direct of a path as possible. I still haven’t gotten them to recognize and overcome obstacles yet as the “pieces” would just need to be recalibrated for the new model who is nearly twice the height of the old one. The next AI navigation step, while more of an issue with the terrain than in the flat testing area, is recognizing walkable surfaces. On polygons this is quite simple as I would only need to check the surface normals (I actually only need to look at the Y component) to determine if a surface was a floor, incline, wall, or ceiling.

Colliders may prove to be different and for more complex terrain will need a mesh collider. Considering the resolution of the visible graphics, a simplification will be needed to create a literal Navigational Mesh as opposed to boxes used for viable path spaces. Unity’s mesh colliders have a limit of 256 faces so the challenge is creating meshes as simple as they can be without sacrificing dramatic and obvious edges on the geometry, and to have all of these meshes match up. Perhaps they could be generated for the area around all entities and move with them, deleting old ones as you pass them. As long as this can be done quickly, that approach might work.

Shots of the textured and partially rigged character model. Those heads are for what Maya calls Blend Shapes, basically it will allow for different expressions.

Unfortunately, due to this being my first time with them, I screwed up and used a different base and then everything got out of whack. My attempts to get the head with the properly index vertices has ended in mysterious crashes and stuff. Not sure what's wrong with it. Shouldn't be a problem in Unity since i'm going to make a script for it anyway.

I also didn't realize his teeth were missing until after a had gotten his joint and skin weights in order. Thank god the rigid bind will work ok on the head.

Status Update 012

Been pretty busy, hasn't been much time to work on this project.

I did manage to finalize the details for the inventory and equipment stuff though. And the new character model's almost done, just need to finish rigging it right now.

I also wrote both of the Orckish Alphabets, the sophisticated Highland syllable-based alphabet one and the Common alphabet.

The more advanced AI stuff is going to be done later, i need to get some more crucial elements done before the fancy stuff.

Higher priorities:

New character model

Inventory and equipment

weapon system

health and stats

test ragdolls

Combat AI

Quest system

Lower priorities

AI object avoidance and pathfinding

fix the sheep

create algorithm for slicing a mesh along a plane

interesting enemy models

sound

One World level system

Story stuff

Spoken Sorkan language

Spoken Orck Language

other things

Here, have some screenshots

Status Update 011

Working with basic AI stuff now, and had to fix a weird error that made what should have taken half an hour to code take a over two weeks.

Currently I have put together a wander and a follow script, the latter of which was screwed up by an odd bit of code that worked for the player but not for the AI.

Next up is some object avoidance stuff; I've figured out an algorithm to use, I just need to figure out a few more details before getting to work with it.

Status Update 010

Aerobatics are finally working...for the most part. The only problem is this weird ability he seems to now have on inclined surfaces, in which he can jump practically 50 feet in the air if he ran up an incline like a ball on a ramp or something/ Its kinda cool but kinda impossible at the same time. I'll leave it for now because it's honestly kinda cool :)

Next stop is Basic AI stuff, i'll just stick with the old ones for now and worry about graphics later.

Status Update 009

I may have found a way around my dilemma with aerobatics :D The problem was that I couldn't directly set the velocity of the rigidbody without accounting for gravity. If I were to create my own gravity, it would not update If i were to change the global variable for gravity. Solution: apply the new velocity like normal, Get the global gravity variable and add it again as a new force. I will test it shortly and post the results.

Return From Hiatus (S.U. 008)

I’m working on the project again, will update as necessary

nippura

manatech

Oh hey, a flying whale!

thevertigomaster

manatech

Reblogged from my other blog:

Got a new monitor today, and noticed that there’s a discrepancy between the pixel colors, probably because new one is led or something. The other one has a reddish orange tint and its been throwing off the colors on my art.

Status Update 007: Aerobatics and 2D stuff

The game is currently unstable, I was trying to create aerobatics (moving around whilst jumping), probably overthought it and ended up with a complicated blob of code that doesn't work worth a crap. Instead of trying to directly edit the velocity values, it would be better to add a sort of air drag system to prevent zero-g floating while in the air. Once that's working I plan to test the ledge grabbing thing at some point.

On another note, I'm developing a top down game with a tileset in my class, and it may prove to be a great exercise for AI programming at a more simple level. I love XNA, and Visual Studio works so much better than Unity's MonoDevelop.

Trending Blogs

Recently Viewed Blogs

ManaTech Dev Blog