Game Dev Journal

Blender to Unity and Playmaker and even to VR.

[westingtyler]

This dev log is a casual discussion of how and why I've evolved my #indiedev game focus from tRS (2014-2018) to #NotSSGame (2018-202X).

When I started developing my video games in earnest, with the advent of my combining Blender with Playmaker and Unity, my main project for experimentation was tRS. About 3 years of casual development went into the tRS experience, starting in October 2014, before I shifted to NotSS development in about June 2018. tRS game and NotSS take place in the same narrative universe (almost all of my stories do), and production focus has shifted to NotSS for now for a couple of reasons. One cool note is that I've already incorporated the environments and story of tRS into an arc of NotSS, which makes for a more compelling narrative angle for tRS overall, than the previous less-focused iteration of tRS. While developing tRS, I found myself more interested in a few of its levels, and those levels, along with an RPG Maker game I made in 2009, is what finally cohered into NotSS around June 2018. There are even more considerations as well that have led to the shift toward #NotSSgame.

As far as dev challenges for the original VR iteration of tRS, the friction of developing using a VR headset every day meant I tended to avoid work more often. This is exacerbated by the way Oculus spams your desktop with their store page every time you turn on the headset, and it HAS to stay in your taskbar or you can't use the headset. Which pisses me off every time and puts me in a bad mood unsuitable for creative stamina. Like I said; friction. The other*r big aspect is that my computer is not technically good enough to run VR, which means my frame rates can get lower than is ideal for development, bumming me out and making me nauseated. I DO, however, still sculpt the landcapes for my games first in Oculus Medium because there's nothing else quite like having such a child-like lego sandbox available in terms of creativity with instant feedback. Alternatively I use Blender's newer Dyntopo mode since I can also wander around in 1st person in Blender itself using SHIFT+~.

In addition to the extra friction of developing in VR, the visceral power of VR can put your focus in specific areas and make you think an experience is more substantive than it really is, which bogs down development. Simply, because I'm so visual, being in a visually-impressive environment entertains me way more than it should, even without gameplay. tRS is largely environment porn at this point with little gameplay "drive" to back it up, and I could release it TOMORROW as a "walk around these cool environments" VR experience. That immersion is very important but not enough on its own to remain compelling for long. And because it's in VR, 3D modeling needs to be more precise and detailed than when working on regular 2D 3D games, and shaders need to be more tailored. Nowadays I model all my assets in Blender in a "scaled" way, getting good reference images and good dimensions numbers from google, since this makes my assets "vr-proof", though I model them at a lower-poly level than i used to, since now I'm working on a huge open world that ideally could run on a mobile VR headset, which means I need to be more economical than when I was developing for desktop VR.

So these are some of the reasons developing in 2D (I mean, NON-VR 3D.. uh, flat screen?) is better than developing in VR, at least at this stage. I am planning for VR by modeling things to scale to begin with, and it's always something I consider - trying to ensure that adding VR support will be easy - and that creates interesting development challenges. For example: how do you create cinematic cut scenes in VR? You could do the "stay in first-person and have Half Life 2-style cut scenes" thing, but what if I want crane shots and sweeping camera movements? How do you do that in VR without guaranteeing sickness? In Resident Evil 7, they just slap the cut scenes onto a 2d virtual monitor in front of you. I don't like that, either. Solving these challenges adds an extra challenge layer to development when I need to be focused on getting the core interactive game experience solid, so while I'm thinking about those VR challenges, they aren't the area of primary focus at this stage. No Man's Sky is probably one of the best VR experiences out there because they focused on making a solid and full game first before adding the VR Layer, so if the first time you play that game is in VR, you will be TOTALLY blown away. (Until the limitations of the game's design itself become apparent, but that's a separate issue, and they keep adding awesome content to that game, so who knows.) And Resident Evil 4 VR is another great example - it was an excellent game FIRST, then the VR layer is icing.

From a design and economics standpoint it's also just really smart to ensure that people without VR, and even without a gamepad, can comfortably play the game (without friction). Making sure it works well WITHOUT VR and is fun and satisfying will make it even more amazing when VR support is added.

Once I realized I can develop most gameplay systems without using VR and can integrate the VR later, NotSS became a perfect candidate for this since the goal is to have a gameplay scripting system that serves most of my needs for any further interactive software experiences (ISE?). ISEs are traditionally called "video games", but the Star Trek Holodeck is far more my guiding light than, say, Pac-Man is, and based on TNG and Voyager Holodeck episodes, I'd not call those games but "interactive software experiences". I wish there was some better word than 'games', and I'm stil waiting for someone to come up with it, so for now I just say ISE or game. I'm really brain-zapped from how game dev has stagnated in a lot of ways and how the true power of this stuff is still untapped. AI will unlock convincing people within a decade or so, but even apart from that there is still free design ground to improve, and I want to do it.

Another big reason for shifting development to NotSS is that NotSS is a more coherent experience that feels more "driven" than the more ambient, experimental tRS. Whereas NotSS is a singular experience, tRS is actually about 45 unique short game scenarios and is a larger undertaking because its needs, characters, and stories are more varied. tRS's varied experiences are a result of my needing to jump around to avoid getting bored with any particular scenario and are a way for me to explore and uncover my ideal game genres and mechanics. NotSS is the current iteration of that experimentation, and I can tell I'm on the right trick because I almost NEVER get burnt out on creating NotSS, and it's designed such that when I get sick of 3d modeling, I can work on Playmaker scripting, or I can develop characters or dialogue. NotSS is EXACTLY the kind of game I wish existed but doesn't, so now it's going to, but it will take time. There's always SOMETHING I'm excited to work on, some problem I'm excited to solve.

Overall the idea of playing a finished NotSS is really exciting and more compelling than the idea of playing a finished tRS, and as I've said, tRS has now been incorporated narratively and physically into NotSS, so we still get all the benefits of its development.

The big difference between NotSS and tRS is that tRS on its own would be an amazing virtual reality game. NotSS, if I can just complete it, will be a revolutionary video game unlike anything I've ever played that has the potential to shift the way many game mechanics are developed for games, while simultaneously hitting the sweet spot in several under-saturated game genres. I really want another Earthbound and another Majora's Mask, so why not combine them? If I can pull it off, that is.

Legit my biggest worry is that NotSS will take so long to finish that by the time it's done there will be AI tools for players to hit a "make game" button to generate something more compelling than what I'll create. I hope there will still be room for hand-crafted experiences in that crazy AI-driven world, and taking advantage of AI-driven tools as they develop so far has given my storytelling a great boost. For now I just work hard each day, and it's exciting to see the project #NotSSGame as it comes together.

[westingtyler]

I've been working on my game #NotSSGame about 3.5 years now, and I just got a world map draft done that I'm happy with, so I'm gathering ref images to make hundreds of props now in Blender. I'm studying Bethesda's Creation Kit as reference for how to name and organize all the 3d models (props PRP, entities ENT, setpieces STP, structures STR, landscapes LND, vehicles VEH, etc. is my own naming scheme I'm using), and I'm prepping to make hundreds of character models using AI generated portraits I created on Artbreeder. I've done tests, and I can drop them into Character Creator 3 with the Headshot plugin (CC3+HS totals about $600), and it uses AI to automatically generate rigged game ready character models with like 56 blend shape facial expressions, at a really high AAA fidelity - these character models even have a tearline in the eyes, which gives a lot of believability. To pay a professional to do 1 character at this quality would take like 3 weeks and hundreds of dollars, if not thousands, I've read. It's a new age now for low-budget storytellers.

So here's a screenshot from Blender of my world map and a picture of a character face I generated in CC3 (the entire body is generated, with good topology, too, and rigged, not just the face). Before snapping this image I easily posed her face in a way to give the impression of a subtle emotion, and it works really well. Character Creator 3 is expensive but very worth it as a game dev tool. I won't be using this AAA quality of characters probably since my style will be a bit more low poly and painterly, but being able to auto generate entire character bodies and faces in a few clicks, is an awesome start that I can use as a base or that I can use as reference for creating my own models from scratch later on. Also here's an image of the portraits folder of my characters. That "yearbook" image is a sample of the 2000 or so character faces I've generated and tweaked using Artbreeder. And also here are pictures of setpieces I made for my game (tv on stand, and porch swing).

In a few weeks the AI Dungeon people say they are adding a pixel art generator into their app/site, so if you generate a text adventure, AI generated, it will also generate pixel art images of what's happening, so that's freakin exciting. AI is getting NUTS. I've asked the devs on twitter if the images will be commercially useable, because it would be cool to have story books in my games with that cool pixel art images.

[westingtyler]

The neighborhood where my 2d adventure game, MESL, starts. You get candy from neighbors, then the adventure begins.

[westingtyler]

People. Only a few steps away from having crafted characters auto populate their worlds. The goal is for me or anyone else to be able to change and create characters the same way a person would write characters in a novel or screenplay -- by writing them.  Each character needs a CAN file, an AskResponse file, and to be added to the character_inits file. When a realm loads, the system checks the character_inits file to see which characters should exist there. Then it takes each one and creates an Entity Null with their respective names, at the place designated by the file. Then the system takes each Null, uses its name to find a Can file for that character, and loads and parses all those cans (can fight, can sleep, can run, etc.) to populate the Entity null’s can list. Once that can list is at its end, the Entity null checks if it has a can exist tag and what its settings are (model, rig, animator, texture, voice.) and adds those things to the Null, thus creating a visible, interactable person you can walk up to and ask questions of. The questions call responses in that character’s AskResponses file. Otherwise the character will go about their daily schedule, as specified in their Can Schedule tag.

Again, the cool thing is that players can edit those files to determine who spawns where looking like what on what schedule with what abilities and with what knowledge and responses to what questions you can ask. It would be trivial to edit an AskResponse file and a character_init file to recreate yourself and your opinions and personal history, for example, or that of any character you’d want to create.

And the cool thing about the can exist tag and the manifestation tags, is that an Entity doesn’t have to be a human. it doesn’t have to be alive at all. It could be a talking door, or a houseplant that has guard duty and then drinks at the pub each night. As the level tasks take shape, I’ll build a prop array, and those props will be useable as models/rigs, too, as well as the unique models for created characters, which will hopefully have variety and can be used in new characters as well.

Right now it’s just about merging a few systems that each set overlapping details. Figuring out which take priority over others. And it’s about figuring out how to create completely new characters at runtime, which could be useful with magic to, say, enchant your kitchen sink and then marry it or teach it to garden.

[westingtyler]

Claybrooke, a Clockwork Village, is now a space that exists. As with all these levels, the point is “you understand what I’m TRYING to do, don’t you? That’s a huge step up from staring at a blank screen, which is what this level was 3 days ago. It’s an iterative process, and the next step will be to walk around and adjust scale, to replace the box assets like the dock, with things like a real doc, and then to place people, create their schedules and personalities, and watch it play itself out. As far as NPCs go, I’m currently trying to figure out how to store response lists in excel spreadsheets that players can edit -- which would mean any player could create their own people, essentially. There are lots of people who live and go about their daily lives in this town, and the goal first of all, before any “game” garbage, is to create a space that feels like a real, lived-in place. You can sit on a bench all day and just watch people come and go, or you could follow them everywhere (but if you want into their homes and aren’t their friends, you better find a window or an invisibility spell or a lockpick or something.) The game layer is pretty high level, and so many devs skip over those two lower levels that I think should ALWAYS exist in the game. 1) ambient environmental actions (sitting, lying down, eating, cooking, using the bathroom, etc.) (very uncommon) 2) people who live their lives. (I very rarely see this in games.) 3) goals and stories and plights. (where common game things are like quests and stuff.) Don’t forget the other two pieces if you want to create a real-feeling world a player can become immersed in! At least, in my opinion. I grew up on Ultima VI and Maniac Mansion, and it’s tough to settle for anything less. In Ultima VI you can churn butter. You can also pickpocket a cow to get its meat, and it doesn’t notice, but that’s a whole different story.

[westingtyler]

The Hotel Realm. I don’t think I’ve shown it before because it’s still hideous. Less hideous now, but still hideous, but here it is. Here you can be a bellhop and ruin the guests’ meals and charge movies to their rooms. And you can blame co-workers. You can also do nice things and even give credit to your co-workers and get them promoted/fired. And you can date guests behind their spouses’ backs. Totally inspired by the Merrymore Hotel scam you can do in Super Mario RPG, but cranked up to 11. I’ll probably call this the Merrymoon Hotel. Or whatever won’t get me sued. And the waterfall idea is inspired by Becky’s apartment from Talespin but way better because why not. The goal is for this is to look a lot like Theed from Star Wars: Episode 1, but more commercial and down to Earth. Today after I see Han Solo followed immediately by Infinity War, I’ll post pictures of Claybrooke, the clockwork village I’m working on, and an actually new realm.

[westingtyler]

The Boarded Up Gun Shop Realm, for listening to the rain while writing your novel, which saves to a text file. Handy for taking notes. The zombies can’t get in. Probably. The glass finally feels perfect, even with dripping raindrops on it. I’m using the new notes system to wander through the levels taking notes, labeled in various categories, currently focusing on audio ambiences, which really bring a level to life.

[westingtyler]

Finally figured out how to make a good water shader with vertex-displaced waves, depth blending, and edge foam. Depth blending would also make it really easy to make low-lying fog, pseudo-volumetric smoke, and thick ice, among other things like a convincing gelatinous cube monster with a skeleton inside.  It’s funny how I really want to explore this place even though I slapped it together in 20 seconds and KNOW there’s nothing interesting, no hidden mysteries or adventures, waiting around the corner or beyond the horizon.

[westingtyler]

The Rainy School Day Realm finally exists. This is where the game starts. Ya know, where your magical talking pug reminds you not to miss the bus. Standard. There will eventually be a basement and an upstairs, and there will soon be the bus stop and the rest of everything.  Now I’m gathering reference photos for the Raveled Space Between Realms. It includes 21 unthreaded ‘moments’ from various realms. Basically, if you escape a magical prison, you’ll be floating in nothingness forever unless you can build a sailboat. That’s what that’s about. In this realm I learned how to easily move the pivot of objects in blender and how to inset faces, and that makes stuff like cabinets and bookshelves completed in a few seconds, where it would take half an hour before. I watched Star Trek Voyager for two 30-hour shifts (all of season 4) to model and texture it. The toilet is gigantic and needs to be shrunk.    I placed audio loops dealing with rain in various places, even duplicates, and made them gentle and quiet and fading from one to the other depending on where you go. Having spatialized audio really sells the mood, even if you have a bunch of quiet duplicates peppered through the level. Also the splattering rain particles are sweet. Also I learned how to use unity cloth to make swaying curtains, but I didn’t implement that yet. Spike the Fish Who Can Be Petted and Zog the South American Bullfrog Who Eats Other Frogs aren’t here yet, but their aquariums exist in rudimentary form. I like this realm and am sick of it and want it to be complete to explore. Get a cool idea, then work tirelessly for 1500 days or something, then maybe you’ll have something. That’s how it seems to work. Ideas are easy. What is it? Like, do 90% of the work, then that other 10% is the other 90%? good stuff. It comes together slowly but surely, but you gotta make backups, and it has to be something you don’t lose interest in. That’s why this game is 45 games in one. I get sick of one temporarily, I can move to another for awhile. Great system.

[westingtyler]

Finally drafted the Beanstalk Realm, one of the final locations you visit in the game. I’ve also been doing rigging and IK avatar tests, so now I can start drawing out the characters you play as in each realm, then modeling, rigging, texturing, and implementing them. You play as 45 different characters, and you meet plenty, too. I’ve also made the Question Bar, which is basically a classic Fallout-style “ask a question, get a response from an npc, or respond to what they said” kind of ui element. There are up to five entries, and sometimes they are timed with a brief window to respond if desired. Lone Echo uses a similar system. Also I made a video critiquing the Fallout 4 Compass: https://www.youtube.com/watch?v=ULVl6R3Auco&feature=youtu.be

[westingtyler]

the roof is better. there’s also furniture and lights inside. and there are windows and doors that need scaling.

[westingtyler]

#screenshotsaturday .. ish. Spent like 30 hours straight (after spending like 30 hours straight on the iceberg realm yesterday) congealing the Weekend Realm into something that makes sense. The color and feeling is right, and the rest is just detail and refinement. I need to walk around in the house to see how the scales of the rooms and furniture need to change.

[westingtyler]

Refining the iceberg realm.

[westingtyler]

I’m using the progressive lightmapper to tweak and generate the lighting in each scene. If in doubt, if the scene seems flat and boring, take away all the light and add, one by one, highlights to things and areas that matter. Render, tweak, repeat. Sometimes the falloff for a light is too abrupt, so I add secret duplicates of lesser intensity selectively around the light “source” adding light only exactly where I want it, creating artificial bounces.

[westingtyler]

The Subway Car Interior. Work in progress.

[westingtyler]

Subway Safe Area before adding shading and detail to objects.

[westingtyler]

The Elevator Shaft and another Mansion Gate with the lighting not accidentally left out.