Games Ambush -- Guerilla Warfare

This is the development log for a personal project I'm working on with several other people (open for recruitment). Ambush is a third person shooter that puts players into the roles of humble peasants from a remote jungle village, and pits them against an overpowering, professional army that has encroached upon their land and threatens their survival.

Key gameplay concepts are a high degree of realism applied to gun play, player movement, enemy senses and behavior, with the goal of creating seemingly impossible situations that will force the player to develop cunning guerilla strategies like hit-and-run attacks, ambushes, and nighttime raids in order to survive and drive the invaders from your territory completely.

Players will be able to hot swap between any allied fighter to have full control of multi-pronged attacks, and will be able to coordinate their strikes with up to three friends. That's right, multiplayer coop -- and over a dedicated server!

In the jungle, there is little that can stop a bullet. For this reason, players will need to move in a unique way that will be an exciting departure from the typical cover based shooter. Sprint, dive, roll, and shoot -- sit still too long and your a goner, and you can bet that the last place the enemy saw you is about to become swiss cheese.

Your enemies aren't only more numerous than you, but they have superior firepower and resources. To make matters worse, they know that they have the advantage and will aggressively seek to utterly crush the players meager forces. This ain't Duck Hunt!

The game world will be as large as we can make it, and consist of many Zones of Control which will initially be held by the enemy army. The players task is to wipe the map clean of the enemy presence, and this will be no small task. The enemy army will persistently pursue the player, react to player aggression in one zone by sending reinforcements from another, and attack any player held Zones of Control. To cope with all of this, the player will have to learn to ambush enemy supply convoys and small roving patrols, as well as use the cover of night and thunderstorms to sneak into enemy territory to steal supplies/free captive guerillas.

_______________________________________________________________

I am the 3d artist and primary designer behind the game mechanics. I will regularly update this thread with art assets as I complete them. The programmer will be joining in to update this thread with his progress as well.

Programmers Twitch stream (he got his sleeping done for this year already so this stream runs almost 24/7): https://www.twitch.tv/brokenbaconstudio

Artist Twitch stream (he has poor bastard internet so doesn't stream as much): https://www.twitch.tv/bigtimemaster


Programs and 3rd party assets we use are:

Game Engine:
Unity 2017

Art:

Maya LT
Zbrush 4R8
Substance Painter 2
Substance Designer
UVlayout
xNormal
Marvelous Designer
Gaia
Complete Terrain Shader
Vegetation Studio
EnviroSkyandWeather
SpeedTree
Tropical Forest Pack
GameTextures.com
Turbosquid.com

Organization:
Slack
Trello
GitHub
GoogleDrive

 

Machine gun in development. I started texturing the other day, but was not getting the proper amount of resolution and had some ugly seams and other anomalies. I've had these same issues before, and because of the same problem. I suck at UV's. I have no confidence in doing them, and tend to rush through them. So I decided to take a step back, and for this model I am going to take the time necessary to get the UV's absolutely perfect so I can 1. have some confidence for UV mapping in the future, and 2. get this model looking as crisp and nice as possible.

 

Can you explain the reasoning behind the third-person perspective choice as opposed to first?

Will there be AI for the other playable characters, or will it be co-op only?

How systemic will enemy attacks/movements be? From your description it sounds like pretty much all enemies will be starting from perhaps a single, impenetrable location (something like an airport with extremely heavy defenses, that a player can't defeat, and they'll be moving to these locations in the game world where they set up camp/attack the player-held zone of control. Is that the case, or are they popping into existence near the zone of control?

Does anyone else in the village want to fight? Will there be allies that attempt to hold the zones of control? Or is the player (the 4 player team (not sure why I thought it was 4)) expected to do everything?

Edit: how many allied characters can the player "control" or lead at once? How large is their team? I assumed it was 4 up there, but what will it actually be?

What kind of machine gun?

 

@EternalAmbiguity :

1. We have been back and forth on this quite a bit, but finally put down in stone that it will be a very close third person over the shoulder camera that goes into an ADS for aiming mode. Reason is because of a handfull of factors, not all of them strictly because of gameplay. Part of the reason is because of many of the time-saver assets we have right now are geared towards third person, and some gameplay reasons include giving the player a better understanding of core systems that might be hard to understand from a first person perspective. I don't really want to be exhaustive with this explanation because it would require 50k words probably, but also some of the mechanics, especially pertaining to how our AI will behave, we don't want to fully disclose because that might spoil the game for anybody who ends up playing it.

2. Friendly NPC AI will not be implemented until Prototype 2 at the earliest, but yes the game is meant to be a SP experience first and foremost. The methods for friendly AI control is planned to be as unobtrusive as possible. Our main source of inspiration regarding this aspect is the Brothers In Arms series.

3. The enemies will populate bases/outpost. One base/outpost per zone of control (think of these as territories). Each zone of control has a maximum cap of enemies allotted to it, and will respawn enemies to keep the cap every so often. So if the player wants to whittle down a zone to weaken it's base for an attempt to overrun and capture it, they will have a limited window to do so before the zone is "resupplied" with fresh troops.

But the enemies do not strictly stay in the bases. The will be very active during the daytime with convoys and search-and-destroy patrols. This is not a fully modernized army with night vision or thermal scopes, so at nighttime or during thunderstorms they will be less active, and thus this will be a time for the player to consider using stealth tactics that might otherwise be impossible to pull off.

Furthermore, zones communicate with each other and send reinforcements when necessary -- which could be a double-edged sword if the player learns to exploiit this.

4. The player will be joined by friendly AI guerillas. In the beginning of the game, this will be a small number. We have not decided how many total friendly AI the player will be able to control/have follow them.

The player will be able to acquire more friendly fighters by capturing zones of control where these fighters are being held captive. Alternatively, if the player can sneak into the base and free them undetected, they can do so that way as well. There is a finite number of allied fighters in the world, so when they die, they are gone forever.

5. We aren't sure how many AI NPC's the player will have control of yet. It may be dependent on the zone or many other factors. Performance will of course be a factor, as we don't want to detract from environment budget to have, say 30 npc's versus 15. Most likely, my hunch right now is that the players guerilla groups would be limited to squads of 10 or less, and each squad can be commanded as a single unit. So, in the late game when the player has many resources, they could stage a raid on a large enemy base with multiple squads attacking from different angles.

6. The machine gun is based on an FN Minimi, but is not an exact replica, just a close approximation. Most of the games weapons will be either familiar NATO or Russian military weapons, but all of these military weapons must be acquired from the enemy. In the beginning, the villagers will only have a default WW1 era bolt action rifle.

 

1. Understood. I was simply curious because third-person can be easier in some ways than first-person. I actually vastly prefer third-person myself.

3. How do they respawn? Do they simply respawn at the base, or is it through some realistic mechanism like a plane or train bringing troops?

4. Along similar lines to this - can the player interact at all with the village they're supposed to be protecting?

 

3. That is something I will have to discuss with the programmer in depth when it is time, and much will depend on the performance measures we've already implemented. The idea now is to only spawn gameobjects per zone and per camera visibility, so off the top of my head I imagine a set of rules that must be met to respawn enemies in a base. For instance, if x amount of time has passed and if the player is absent from the zone, the base will be resupplied to max cap.
Having actual trucks come in from outside the map with troops to resupply a base would be awesome, but that would be an icing on the cake type of thing we aren't focusing at this point.

4. If there even is a physical village... the story in this game will be condensed to a brief text introduction, most likely. Unless we come up with a lot of money or some serious talent joins us, I doubt we will have time for any kind of cinematic or dialogue. Of course I have beautiful, heart tearing introduction cinematics all choreographed in my head, but this is way beyond the scope of the project right now.

However, we have discussed having one default zone of control that would persistently supply a bare minimum of resources for the player, and this could be modeled as "the village".

 

Good to know. Thanks for the info.

 

This is probably my most favorite point of view when playing a game. This allows the character 'art' to be seen, all the hard work put into making a compelling, visually appealing character, and is a great perspective to give for the player to 'play as' the character rather than 'being' the character.
One of the very best (imo) games to deliver this perspective is Gears of War. For a very specific type of game I want to create in about 5-10 years from now - Gears of War(s) are the definitive references that do this point of view exceptionally well. Unfortunately I have never played a gears game but I have researched this character perspective extensively and Gears has a great set of up close and camera/character control to deliver both melee and ranged weapon combat in a smooth/controlled/cinematic method.
Off the top of my head at the moment here are some other notable games that are great references (that I've played);
Shadow of Mordor
The Last of Us
God of War 3
Horizon Zero Dawn
Uncharted
Witcher 2

One element of design to strongly pay attention to is animation - for 3rd person perspective. The more frames in an animation the less responsive the control (might) feel, but the more believable/accurate/smooth the animations will be. The less frames in the animation the more responsive the controls will feel, but accuracy will suffer especially in transitions. If melee is part of the game (which at this point doesn't sound like it is) moving from shooting to hand to hand combat takes careful and iterative attention to make the final delivery look good, control well, and be believable and enjoyable for the player.

 

As far as this part is concerned a valid example to look at would be Ghost Recon Wildlands, which is a "military themed" third person shooter so it does a decent job with balancing animation and player responsiveness.

 

GR: Wildlands is one of those games that inspired me with this game idea. Not because I liked Wildlands, but because I liked a few parts of it but was overall sorely disappointed.

I agree, Wildlands looked amazing and the controller/animator was on point, but the shooting and aiming was, IMO, just god awful. That was the main reason I dropped it at the beta and never looked back.

But, @theANMATOR2b , thanks for the advice about animation. I am just getting together my first rig for the player character -- should be done in a couple days after some experimentation with how I want to handle the modular equipment -- and then I'll be delving into animation. For this games purposes, and considering that I am learning the entire art side (modeling characters and environment, rigging, animating) from the beginning, I doubt it's realistic for me to expect super smooth, super cool looking animations, and that is also kind of beyond the scope and point of the game. I want the controller to feel really fluid and responsive, and the animations should help convey the guerilla's character -- not necessarily be a showcase of experience and modern techniques like Horizon Zero Dawn for example.

 

Woo hoo!

I haven't done any manual fix ups to the skin weighting yet, but.... he's alive!

Once the deformations are cleaned up, I'll be starting to learn animation because these guerilla's are going to have a very distinctive way of moving -- after all, they aren't trained soldiers like their enemies.
https://i.gyazo.com/0ae1403bdeb6ca9aa1faa66a103c56f3.gif

 

Quick render of our machine gun model with Iray.

In Ambush, the machine gun is an absolute game changer. Firefights revolve around the "gun" -- if it is used well, it can multiply the killing power of a small team of fighters exponentially. But, for the players to get their hands on a machine gun, first they'll have to defeat a machine gun team.

Note that this will not be the final model, but rather a "good enough" model for the first prototype iteration. Assets like this and our first character model work as placeholders for now, but the reason for building them from scratch rather than just using cubes is for my own learning by trial and error.

The next step for this model is to composite the multiple texture maps produced in Substance Painter into one file for optimization purposes, and then get it set up and looking good in engine.

 

Some fancy rendered images of the new machine gun with a single material....

This model was made low-poly only, without a high resolution version used for normal map creation. I did it this way for speed, mostly, and also because I think given the typical distance from the camera we will have in game, most of the details wouldn't even read.

 

But more importantly, here she is in engine.

Note: not a first person game, but I don't have the character rigged to carry a gun yet.

 

A revised version of the first player character. Here is a poor farmer. He don't even got shoes.

Texturing is just beginning. The goal for this revised character was to make a simpler model so that I could focus on improving my UV layout and texture map baking skills. Well, I learned a lot of new mistakes that I can make and sprouted a few new gray hairs, but overall the UV mapping and normal/AO maps are much cleaner for this guy than for the original.

One thing I've learned through lots of trial and error with map baking process is this: You can't just click buttons and eventually get things right. Because each iteration takes time to bake, you need to spend a lot of time reading and trying to understand exactly what is going on and what all the options available actually do. Don't expect to understand normal map baking and cages and ray cast with one model, or two or three. Do expect the process to be frustrating, but just keep trying and eventually you'll start to figure things out, even if its only enough to kind of get things to work for you even if you don't fully understand what you are doing.

Finished game model soon to come.

 

Emmanuel, a farmer:
"When the soldiers first came to our village, they stole from us, and said insulting things. I was angry, but my father insisted I not say a word. For three days I was so angry that I could hardly work. Then, the soldiers returned... and they did evil things.

Anger filled me completely. I forgot everything. I only wanted to kill those men. If not for my father restraining me, I would not be alive today.

My anger -- now I have learned to save it. When I am weak, it gives me new strength.

Every day, I say a prayer: let every drop of anger in my blood allow me to drive those murderous dogs from my people's land, forever.

 

Next character is getting blocked out now.

The first thing I do is take the base model, which comes from the previous character, and clothe him in Marvelous Designer. So these clothes are hi-poly, while the rest is still low poly. So it's a bit of a mixed workflow -- typically I like to go low-to-high as that saves me the tedium of retopologizing, which I'm not very good at because I don't have the patience to make sure its all perfect, but I have to get the clothes on first as everything else fits on top of them.

This will be the standard enemy Team Leader, which is, as the name suggest, the leader of our base enemy element, the fire team. The team leader will be one of the more dangerous enemies, as he carries a radio that can be used to call for reinforcements. Additionally, he uses smoke grenades that can disorient the player, allowing the enemy fire team to get close and throw frag grenades.

This guy is much slower to blockout than my farmer with overalls and nothing else. His gear will include a vietnam style US radio (right now the square on his back), an entrenching tool, a canteen and pouch, accessory pouch with smoke grenades, compass pouch, medical kit bag, a load bearing vest with magazine and accessory pouches, and I'll probably put something on the left side to so he isn't lopsided. Not sure what yet though.


I bought the boots, which already are textured. I hate doing shoelaces, so I let somebody else do that for me.

 

Further blocking out. Once the pack frames shoulder straps are finished, it will be time to start sculpting in Zbrush.

 

Now the fun part of character creation -- Zbrush.

Though this part can also become a serious time sink. I find it easy to get carried away with detailing, much more than what would even be necessary for our perspective or scope of game.

What helps me is the fact that I have completed a couple characters before, so now I have an idea of what level of details read in engine, which details can be accomplished more easily during the texturing process, and what kind of details need to be accentuated in order to be noticed from the typical camera distance.




I've deviated from my reference images quite a bit and made this guy look almost like a caricature for a career military hardass, but he's the team leader and I want each of the enemy types to be distinctive in order to let the player know that they aren't just separate skins for varieties sake, but that each enemy fills a certain role and holds strategic importance.

Because I am making realistic and complex characters, which takes a good amount of time, there will only be one model for each type of enemy -- for awhile but ideally I would eventually have a handful of different faces so that the player isn't fighting the same six guys over and over, of course.
Most clothing and equipment that the soldiers carry will be available to the player to equip to their character for practical and visual purposes.

 

between the other thread and my own thoughts, I was thinking about how you're going to implement damage.

Standard hitpoints for the whole enemy? Specific locations?

 

Well, we are a long way from implementing this, but the current thought is one hit body or head k/o, two hits anywhere else k/o or wounded.

That's what the most bare bones method would be. Given how difficult it would be to get an accurate shot on the enemy, I think it woud be borderline frustrating if you had to shoot them 3 or more times to get a kill.

But I'd love to explore somehting a little mroe robust if we had time. Perhaps only a shot to vital organs gives a kill, but it only happens after 10-30 seconds. But a spine or headshot drops them into an instant ragdoll. Limb shots put the enmy into wounded state, which immobilizes them and will call on another enemy to drag/carry them out of the fight if possible.

That would be the kind of stuff that doesn't make or break the game though, and would take a lot of work to implement, so tons of planning time hasn't been spent on that so far.

 

The game res mesh is ready to be UV'd. I can't wait to texture this, I think he'll turn out pretty nice.

But here is a good opportunity to note not exactly an obstacle but a potential problem we need to plan for. For optimization purposes, the way I set up the materials for these enemy characters and their equipment needs to involve some planning and the ability to modify later.

So, much of the clothing and equipment this enemy soldier is wearing will be able to be taken and used by the guerilla's. In fact, the only thing I think that wouldn't be lootable would be the radio, since guerilla's wouldn't have the means to use one securely anyway.

For right now, I will make each standalone item with its own map. 1k or lower, just whatever it takes to get the appropriate level of quality. This will make five different materials for a single enemy character, which just seems excessive to me. But I've seen that there is methods for combining materials at runtime, or I could just manually make a multiple variations of maps based on equipment loadouts, htough that would be extremely time consuming I expect.

In any case, it will slow me down too much to try and fully understand advanced topics like this right now. I'd rather keep creating, and when it comes time to tackle those challenges we'll do so then.

Update: I'm lucky enough to have a guy on the team who has a good bit of experience with the texturing side of things. After discussing all the possibilities we knew, we've decided to stick with plan A as it will be the most simple method. If we really need to trim off all the fat later, we can explore more involved methods for combining multiple textures per material, etc.

But why so much fuss over performance anyway? Is it really important? Well, I haven't made a game yet so I can't answer that, but I do know htat we want to get as much foliage into our game jungle as possible, and use models that look as good as I can possibly make them. Aside from that, we want good lighting and weather effects, and during gameplay there could be as many as 30 enemies and potentially thousands of rounds being fired at a time.

So with all that in mind, I think we will really being pushing things to the limit, and thus trying to keep up with the latest performance measures will be a good thing for us.

 

Hi @BIGTIMEMASTER ,

Awesome looking project! Like the idea of it not being Call of Duty like gameplay but instead a more difficult process of out-smarting the enemy. Look forward to seeing updates!

I see you are using TFP here in Deferred Rendering but not with the Custom Deferred shaders applied. If you need assistance with this please let me know. The leaves will look right when using Deferred.

Cheers!

 

Hey, thanks for the reply.

Yeah, about the lighting and all of that -- I did have some questions to send you but before I do that I need to do a little bit of reading so that I'll actually be able to understand any answers beyond "click this, click that."

Right now I just don't know anything about lighting not just in Unity but in general, so a lot to learn before I get started there. So far I have only tossed a bunch of things together as far as the environment goes to get a general idea of things -- no actual work has really been done so far on the environemnt.

Awesome assets by the way. I have a bad habit of never writing reviews, but I'll try to improve on that. There is one model included in your asset I particularly like -- the big semi-rotten stump. For our project, I want to include lots and ltos of assets like these, because in a real forest or jungle there is usually as many dead or deformed trees as there is straight, upright living ones. Something you don't see too often in games.

Do you have any more assets like that? If you do, or could perhaps create a package with a handful of dead, fallen, rotten, or even living but perhaps bent over or otherwise deformed trees, I'd happily purchase that.

 

Yeah understand but here is a quick video that party goes over making the Deferred shading looking right when you are ready-

Thanks man! No worries. Not at the moment no. But I have already started scanning a bunch for my next pack. And yes a good chuck of the assets in the new pack will mostly be dead cover and trunks since it really does make a forest feel real.

 

What I learned today is that Maya LT has a polygon cap on exports which threw a wrench into my map baking process. So I had to do my baking in Maya LT, which is unfamiliar to me, but I have achieved decent results -- well, except for a few certain areas. I'm sure you can see them right away!

But baking over and over is slow and frustrating, so I've looked for a way I can fix these errors by editing the normal map directly, rather than trying over and over to get a perfect bake. The problem is, I don't know photoshop -- but I do know substance painter and luckily I can edit the normal map directly, and in 3d. Same with the AO map. So despite some folds clearly interfering with my raycast, I believe it may be easiest and most efficient to simply fix errors like this inside of Substance Painter.
No excuse for not trying to get better with my map baking, but if 90% of the model is good I think its a better use of my time to to clean things up this way rather than keep banging my head against walls.

Also, given this entire mesh having one 2k map, the details I put onto the vest with micromesh didn't even show up well on the normal map -- so i just dropped that entirely. I should be able to add some kind of texture pattern in substance to give the look or a coarse woven nylon material.




Also, here is a motivational note:

You never see screenshots like this on artstation. People only show you their best renders. But I want to share this crappy looking mess with you so that, should you see a decent looking finished game model later, you'll understand that to go from cube to game character is a long and many times frustrating process, and a beginner like myself really has no business trying to make a complicated character with lots of pieces like this.

But nobody can stop me, can they?

and here is with revisions to the normal and AO map made inside substance. Not ideal, but better than before obviously.

 

Getting there. Another day or two. I want to make a few notes about things I am learning about the game asset workflow:


Less Zbrush, more Substance. The more characters I complete, and the more time I spend dinking around in Substance Painter, the more I learn that so many things can more easily be done with texturing than with fancy sculpting, micromeshing, etc.

I think if this was a AAA hero character with multiple large texture maps for different parts of the body -- in that case some of the detail you might make in Zbrush could look better than what you can do with textures alone. That's just my speculation. But this guy is a throw away enemy, so I've kept him all in one 2k map. So if you get the camera much closer than it is here, things start looking muddy anyway.

Some of these items I didn't even bake the high poly version for. The grenades, the pouches. I can easily paint in folds and noise directly onto their normal channels and AO channels, so unless I want to show off a badass looking untextured model, I see no reason to do that stuff in Zbrush.

TLR -- for learning, complete assets full assets in a timely manner - forget about making masterpieces - that way you will learn the most efficient workflow and which programs do best for different aspects of the asset creation.


Some game related notes: some thought has gone into exactly what kind of uniform the enemy army should wear. We've settled on this altered version of the classic BDU woodland pattern, which looks a little modernized, but hopefully not too specialized. We want the regular army troops to appear well equipped and reasonably modern, but this is a developing nation army and not outfitted with the most current military tech.

Also, we want to draw a distinction between the regular army troops, because later we want to have a secondary type of enemy -- special forces. These will be small hunter/killer teams that will move into area's where the player has been most active. The special forces teams may have dangerous technology like night vision, and will practice stealth and ambush tactics to turn the predators into prey.

Should the player take down a special forces team though, that same technology could be theirs.

 

And... another one!

33K tris, 1x2k pbr map. In the future, I can probably trim the tri count down a bit, and definitely do a lot better with the laying out of the UV. Much has been learned, but overall I htink this guy looks pretty good for our first enemy. I can't wait to kill him!

His boot textures still need to be merged with the rest, but other than that this guy is finished. Note some minor changes from the pretty render model to the in game version. I tweaked some spots and removed one pouch on his vest.

In game shot is taken inside of @Baldinoboy 's Tropical Forest Pack demo scene. I used the video he posted above to get the lighting a little better, and man does it make a difference along with some of the standard post process effects. I can't wait to learn more about lighting and rendering down the road.

One problem I see is that my ambient occlusion doesn't seem to be showing up very well or at all in engine. For instance, the little rings on the web belt should be blackened in the inside. At least, I filled them completely black on the ambient channel, so I don't think the ambient light in the scene should overpower that? I will have to research that more in the future.

 

Haha, Can not wait to kill your own creations. Love it. When you are shooting him you can think how annoying UV mapping is and it is his fault

 

"You yellow belly son of a.... DRAW!"

A-pose showdown. Just checking out the updated Tropical Forest Pack, and really enjoying just how good a Unity scene can look with five minutes of work with the lighting (and really nice environment models).

Figured out issues with ambient occlusion map. I was exporting it wrong, nothing to do with Unity.



Next step is to move into animation full time until I have all the basics we need for P1 finished, but first....

I don't want to make machine gun reload animations to begin with, so our battle rifle has to be built. This will be based on the ubiquitous G3 type of rifle. As this will be the first model I've made entirely in programs I own -- not student versions -- I'll put it on the asset store for a nominal price when finished.

Here's progress so far:

 

I'll call this the UBR-91 -- Ubiquitous Battle Rifle. Modeled after the G3 family of weapons, but like the machine gun I have taken bits and pieces from several different variants of the rifle to create something unique. I also didn't strictly adhere to real world sizes and proportions. Overall, I just tried to make it look cool while still looking believably realistic.

The UBR-91 will be the primary weapon of the enemy army. A select fire weapon capable of full or semi automatic fire, but the solid kick from the .30 caliber round will make single, well placed shots the most effective means for engaging targets. Effective range: 500 meters.

Features rotating rear sight post with elevation adjustments for 100-400 meters, a telescoping stock for ease of carry in thick vegetation, and a detachable 20 round box magazine.

Pull the trigger to make it go boom.

Still to go -- topology cleanup, unwrap, and texture. Look for this on the asset store soon.

 

Here's a short video showing how I merged different texture maps together inside of Substance Painter -- no Photoshop required. I haven't seen or heard of anybody doing this, so I thought I should share.

Unfortunately, the sound isn't working, and I don't have the time right now to troubleshoot why. But it's pretty self explanatory, assuming you know how to get around in Substance Painter already. Just 2x the playback speed as I spend some time explaining stuff that you won't hear.

 

Finished.

This model has been submitted to the asset store -- my first!

Package includes 4k, 2k, 1k, and 512 textures, as well as prefabs for each and a 2 tier LOD prefab.

Next, I'll be rigging our characters and this rifle, then learning animation. Once I've got satisfactory animations for this rifle, I'll update the asset store package and anybody who has already bought the asset can get the animations as a free upgrade.

 

Forgot to put the bad guy on sketchfab

 

I want to make a few notes about rigging as I've learned a few helpful things and there isn't as much information about this topic as there is things like modeling and texturing:

In the picture below, what I am testing out is that our evil bastards gear does *not* deform, but follows the torso realistically. I didn't have any idea how to accomplish this at first, so I did some searching for how people rig armor pieces. Surprisingly, there wasn't a whole lot out there, just a few very dated threads which didn't exactly answer my question or give me much in the way of useful ideas.

But luckily, Maya LT has a really friendly, intuitive rigging system -- both the human IK system and the skin weight painting tool. So just by poking around I learned that you can flood a weight value to a vertex selection, and I figured since all the spine joints follow the hip joint rotation, if I just assigned all the gear to the hip joint, that should give me good results. And it did! I love it when idea's just work.

Oh, one other thing for Maya users -- quick select sets. Learn how to use quick select sets!

So I've been messing around, taking each joint to the extremities of realistic movement, and I think there won't be any issues of gear clipping through the body. Turns out this was pretty easy to accomplish, but I won't hang my hat on this until I've retargeted and actually tested out with some of the unity mocap animations you can use for free.

Should be getting into animation tomorrow.

 

I got a little sidetracked today.

Still practicing rigging, but made some simple animations to find problem spots and then test out my weight painting.

 

About rigging --

I'm probably watched every video tutorial about rigging that google can find, and one thing is the same for all of them -- they are all geared towards total beginners, which I am, but unfortunately the character I am rigging is not a beginner appropriate model (i.e. one contiguous mesh, like a naked person of super simple cartoon character).

But this is actually not that big of a problem, because rigging it seems isn't nearly as obscure as things like map baking. It's a pretty straight forward process, and after dicking around for two days, taking my character through all sorts of bends and poses, I think I've got most of the kinks worked out now.

A few important things I've learned -- you can either keep the different parts of your model as separate objects, or just separate into quick select groups. Either way doesn't matter, but you want to be able to shuffle through all of your stuff quickly to allow rapid experimentation.

There was some small bits of my grenades that were not merged to the whole which wasted time as I kept having to chase them down. Before you rig, select each of your objects and run a merge at 0.01 to connect any lose vertices (unless you need some parts disconnected.)

I didn't build a proxy geometry during the initial rig, but now after playing around for a bit I have a good idea what my trouble spots are. For instance, I scaled the left elbow up a bit as I found it almost impossible not to lose more volume than I wanted there, and I also slid more edgeloops towards the shoulders to keep the end from coming out of the vest.

Lastly, I am going to try making a new skeleton with an extra joint solely for rotating the small exposed part of the arm. Previously, it had only had the elbow effector or the wrist joints to rotate from, and this led to it popping out of place during key animations like aiming a rifle. I am not sure if this will solve the problem, but its my best guess right now.

 

I have quite a few things I want to say about rigging - there are so many important little things that are usually only casually mentioned in tutorials. Like, in the beginning when placing your joints, people will just say, "its up to you where to put the joints", or maybe "use some discretion when placing joints." But you are new, you don't know how things will turn out.

Well, what I've found is that, for me at least, putting joints exactly where i think they would be in real life isn't the best choice, particularly at the shoulders. Overall, my best advice and lesson learned now that I've got this guy looking mostly decent (haven't touched up the legs yet) in his most deformed pose, is to use your quick rig or humanIK tool, and give yourself a day to just mess around, pulling and pushing, seeing how joints potion and different weighting effects things. Then, after that, you'll have a much better idea how to make a better skeleton from the beginning.

Also, and this is something sure to frustrate you -- don't subtract weight from joints. Only add weight from the joint you want to reach normalization. If you are going through your joints and find that say, your knee joint is influencing to high up towards the hip socket, don't just replace that weight with a zero value. What will happen is that other joints, and it could be something crazy like the thumb, will fill the void to maintain a normalized value. So if you go around subtracting, you'll end up with a crazy mess that it hard to make sense out of.

Not a single tutorial I read mentioned this -- not once -- and yet it is such an easy trap to fall into. I realized the better way of only going around and systematically adding weight only after reading a forum thread from like 2002 or something in which somebody was complaining about this same problem.

Initially I had searched to see if there was a way to turn off auto-normalization, but I didn't find a straight answer and as I don't understand what many of the functions surrounding skinning and normalization do, it's better if I don't experiment too much now that I have satisfactory results -- and pretty quickly too. Just a few hours work today and the character is good to go.

 

Man that is looking awesome! Also cool to see hints shared. I have always been too scared to even touch rigging. Would be cool to try eventually though.

 

Thanks.

You know, I've found rigging to be like 80% fun, 20% work. So far. I am only doing very basic things here and using Maya's automated tool sets for the brunt of the work, but nonetheless it has been pretty smooth sailing compared to other things I've learned.

 

Are you doing that video series you talked about?

 

I would like to make a comprehensive series of videos detailing all the mistakes I make, how I solved them, etc., but the fact is it takes significant time and effort, and I spend all of my time and energy working on the project... so no. Not yet, I should say though, because once I get the major workflows down and things become more a matter of repetition than learning for the first time, then I may have more energy to put into that.

 

Here is an important note about problem solving for the self-taught hobbyist:

I had trouble finding a way to get the lead hand to follow along, staying glued to the rifle as you see below. I knew I wanted to have this functionality because having to move the hand after the gun for every animations would probably be tedious and time consuming, but I couldn't accomplish it.

Well, I searched the internebz up and down for methods of weapon rigging, etc., but couldn't find an solution or any ideas to build a solution from. It seemed I was doing things the right way, but it simply wasn't producing results.

Finally, after quitting for the day and thinking about the issue with a fresh mind this morning, I realized the issue had to have something to do with the fact that I wasn't using a home-made skeleton and control rig, but the humanIK system which is kind of like a self-contained system. So I read up on the humanIK documentation, and in ten minutes I had my answer. Just add an auxiliary effector, and voila. Great success.


TLR, if you have a problem and you haven't read all of the documentation for the system in which the problem exist, read the goddamn manual.



The good thing about making a tactical shooter -- I don't need "tactical consultants" to tell me how our characters should be holding their weapons.

 

I saw your post at reddit, I learnt a little unity here and there. May I know how to join ?

 

2/7/18:

I had been working from a humanIK rig which was very easy to set up and functioned well, but most of the reference I am learning from is geared towards custom rigs, and as I moved into animation I kept finding myself hung up on problems stemming from my lack of understanding of rigging and how the humanIK system differs from ordinary rigs.

So I'm going back a step and learning rigging properly. Lots of repetition with the different constraint and parenting methods should give me a solid base of understanding so that things there aren't step-by-step videos for, I'll be able to figure out on my own.

I'll build one custom rig that should carry most of our characters and animations -- certainly for P1.


Update: My custom rig sucks. HumanIK lacks a few features I wanted, but will suffice. Moving on...

 

2/21/18

Trouble keeping dependable help with this. Maybe it is harder to get excited about something that is not your own? I don't know. I can join in on others projects and get more excited about it than the owners in no time. Maybe I've just been getting a string of beginners who aren't quite set on game development yet? Who knows.

I've got a few animations done, but they are not up to the standard I want for this project. However, gittun gud at animation will require probably as much time as I've put into modeling thus far, and I don't want to spread myself so thin. Ultimately, my main priority right now has to be developing a career, and I'm ready to start applying to jobs. So I've started working on portfolio pieces and also am doing some character models for an up-and-coming indie studio.

All this means that Ambush is getting put on the way-back burner. At some point, probably some years down the road from here, I'm hoping to come back and either have some money (i.e., convince the wife to let me have the money) to hire a professional programmer and animator, or have made some reliable contacts who love tactical shooters and want to make the best damn one they've ever played, not just screw around a little, get bored, and then fade out.

But it may be sooner than a few years. If I can get settled into at least semi-reliable work with modeling, I'll spend some free time studying coding again, and then I should be able to put together a single player prototype with the core functionalities of the game myself.