A Writer's journey into game development

Lessons about game dev

2025-12-09T11:05:00.001-05:00

Meet Bobby (not Kotick)! He is going to taught you a lesson about how NOT to make games.

Bobby is a very sharp bussinessman, that sees everybody is playing games. More important, everybody is spending money in games. What would any smart bussinessman do? Well, get into the games bussiness, and he knows he has an advantage. See, Bobby is not going to spend hundredths of thousands dollars to make a game, when he can make it much cheaper.

And here comes the first mistake. Bobby goes to his birth city, in a shitty third class communist country, where it is easier to find a honest politician than a game developer. There, he assembles the Avengers... I mean, a team of talented guys that... never made a game in their lives. He invests in hardware, with good workstations equipped with fairly modern GPUs, pairde with the worst, cheapest 18 inches display you can get. Also, a server with a 24 VRAM GPU, because any game studio needs a server for... nobody knows exactly for what, but they have a server and no one can use it to work (because servers are not for working).

The second mistake: Bobby is not a gamer himself. Never played, even in his phone. Neither is a developer (probably you figured out that when I mentioned the useless server), or an artist. Yet, he insists in design the game, organize the development process, design levels, have the last word about colors and art... he decides about everything. Third mistake: he chooses Unreal as game engine. Don´t underestand me wrong, Unreal is a great engine, when you have a couple of veterans on your team.

Through the next two years the team will go from one project to another, struggling with ever changing members, irrational scopes (multiplayer games are not the best way to start learning), and poor process management. Design document? We don´t do that here, we improvise and we communicate the ideas to the team through constant phone calls, two ro three every day, stealing an hour of work or more. Technical limits? No way! Bobby doesn´t only wants to make a game, he wants to prove a point: that he can make games in Cuba, for a fraction of the price, and make it AAA quality. Levels quickly get out of hand: larger and full of stuff. The more polygons the better: when I joined, the project was a... platform (sort of) for mobiles. One single shoe of the character had 170k polygons. When I told him that it had to be remade, he answered that it wasn´t possible because visual quality was strictly required. It was not only a platform game, it has to be the damn best looking platform game on Google Play, the Crysis of the mobile platform games. The team was neither convinced that retopology was required. "Look here", they showed me the shader complexity viewin Unreal, "the character model is green, so, it doesnt requires optimization".

The project died under its own weight like many others, yet, the lessons were not learned. From the beginning, same mistakes were made. Need a shiny surface? Use lights! Emissive materials? No way, they are more expensive than using lights and they don´t shine enough (perhaps because the value was 1.0, instead of trying more to see what happens?). Look here and see, lights are dark green in the shader complexity view, so they are optimized! The lights had shadows, but dont worry, the light areas don´t overlap each other (at least, most of them), so, no problem if we have 20-30 lights in scene at same time!

We wanted to make the game in one month. Me and another guy, who had experience with Godot, ensured that we could make a endless runner in a month, yet, we were not authorized to change engine. After 8 months, the project is... hard to say. Features keep coming, and more crazy features has been discarded, becasue of course, the project can not be just a runner, it has to be a runner with original mechanics, with bosses you had to fight, and lots of things falling onto you from everywhere all the time. Actual content for the players to buy is still missing (but a lot of effort has been invested in cinematics, filler NPCs and things that will not generate direct sales, or any profit at all). And when finnally the game was tested in a phone... it rana t 15 FPS. Yeah, that´s another one: only one PC could build and the user was too busy to build. So, we, the developers, got one build to test every 2-3 months.

Well, would you say, it is impossible that after three years they haven´t learned anything. Yes, a couple of things were learned the hard way. The main artist had to discard scene lights. I discarded lights in place of emissive surfaces, without asking permission or telling anyone (nobody noticed it). We couldn´t cut the massive amount of assets used on each scene type, but at least the trees were replaced by a billboard. And no, it was not enough, we barely get 30-32 FPS in a mid range phone.

Ok, then, first lesson. If you don´t know about something, get somebody who knows and pay him to do the job. Specially that: get somebody who actually knows what he is doing (second lesson). Making games is not about placing assets in the editor, is about knowing what are the best assets to place in the editor for the kind of game I am doing and my target platform. What a nice looking 3D game? not gonna happen in mobile platforms, unless you have some really great 3d artists that can texturize like real geniuses. Third lesson: listen to the people who know. Not sure if this is the place, but here is fourth lesson: principles are fine, and wanting to prove yourself is great, but ultimately, game industry is about making money. Wanna impress? Make a game that makes money. Simple is not bad. Fancy graphics does not make a game good by themselves. Focus on making something fun. That´s more like several lessons in one, but right now I lost the count, and I have to go back to work. Without consulting me, Bobby decided that in some specific level, the player skates instead of running, which means the mode made for the rest of the game and the careful work I made to support multiple playable characters with equippable items is gonna go throught the drain. If it is even remotely possible to implement such thing.

We are in beta! Or pre-alpha?

2025-04-07T23:32:00.002-04:00

Well, what I mean is that we entered a second test phase, this time with a few more people. And yes, I have been away for much longer now, but Cuba is a total disaster right now and every minute with electricity has to be spent on something that counts.

Currently, the plan is to polish current prototype and see what happens. I really have no hopes that with current look it will be accepted by a publisher, but I have no budget for a whole overhaul of art. I am thinking to use itch.io and try to get some donations, so, stay tuned if you are interested in helping a game developer in a shitty country.

My O3DE tutorial, first a series of short videos

2024-12-27T16:12:00.003-05:00

Welcome back!

2024-12-07T09:50:00.002-05:00

Good news! And bad news.Project is coming up nicely, with a small private demo released to some friends. Got great insight from it, and even fixed some issues. Big problem is that some other issues depend on a full team, or at least, on some artists working regularly.

The game requires a total rework of the art. That's inevitable. We took some bad decisions to speed up work that has proven to be a burden in advanced stages. For example: using MakeHuman to generate characters. Per se, it is not that bad, if you take some time to retouch the result. Also, animators hate MH's rig. Don't use it. I had an excellent animator, and the game anims are bad. Not totally his fault, he had to work with a subpar rig that limited what he could do.

Also, another advice. Godot really sucks with kitbashes. Now we have a nice kitbash that allows me to create all sort of fancy buildings, even with interiors, but the game suffered a severe perfomance penalty. Godot can't properly instance repeated geometries.

So, I got two proposals. First, an old friend suggested to redo all the art and go stylized. That's the look the game should have, from the beginning, and as side effect, the character modeling speeds up a lot. Second, switch to Unreal. Sounds good, but no, that is impossible. The project really doesn't benefits too much from Unreal, and if I had to switch engines, I would choose Unity. I am more confident I can replicate the current prototype in a few months using Unity. Doing the same with Unreal would take a a year.

The last bad news: my country is definitely deteriorating at light speed. Currently I'm having 6-7 hours of electricity every day, and some of those hours are while I am at office, or very late night. So, can't use them. But anyway, I am doing my best to overcome that problem.

Hope I can post more regulary! I will be back next week with some updates.

Testing O3DE: new video

2024-10-23T15:11:00.002-04:00

O3DE is no longer the newcomer. From "why do we need another engine if we have Unreal, Unity and Godot?" we went to "Maybe I should consider this as a valid candidate". Version 24.09 offers a certain level of maturity, enough to make it worth the risk of trying it. In fact, there are people already working in serious projects, like Carbonated Inc. which is using it for a multiplayer mobile game.

So, I am spending more time with O3DE. It is something completely different and requires more effort than other engines, specially because I have other projects to take care, and also because I have to learn Unreal at same time (it is one of my jobs now). But, despite lack of documentation and tutorials, I have advanced a bit, mostly thanks to the help of the community. Lets see what can I do with it in the future.

Video: dialog system

2024-07-22T10:58:00.000-04:00

Project progress, and some thoughts about our workflow

2023-07-19T19:09:00.002-04:00

Well, considering most of the team is working only a few hours every week, we are not too behind schedule, but I don't foresee we can release a demo for november. Some time has been spent trying to find a workflow that help an small and unexperienced team to produce a sizeable amount of complex assets. We are still experimenting, but a few things are already in place.

Ok, lets see what have we done.

After several tests, we have defined player models structure. We are using a body separated in six parts: head, torso, arms, hands, legs and feet. To save time, the character is generated via MakeHuman. The pro is that you get a complete and anatomically correct model in minutes. The cons, you have to work a bit to lower the poly count, because MakeHuman offers a high poly model and a very low poly one that looks ugly. Also, all the models looks pretty much the same (you will notice that in the character creation scene), but as we use an isometric camera, during gameplay the racial differences are almost imperceptible. My advice: don't do that if your artist have enough time to sculpt and retopologize, or can go directly to modeling.

The characters are nothing without items, and we found a 3d artists that produced several of them. Now we have more than 30 weapons. The armors required more tests, until I could convince the team that the correct workflow involves producing equipment pieces that can replace body parts (which requires the same skeleton as the base body) or can be simply put on top of the body. For each race/sex an additional step was required: shape keys. Shape keys provide different shapes for a single model, instead of having one model for each race/sex. Looks difficult? Well, not that much. Needs effort and polishing, but, what in game development doesn´t needs it?

Currently we are generating body variations to create a variety of NPCs from a single base model. It is an old method, used by CD Projekt RED in The Witcher 2/3.
The combat system works, but it will be put to test with some volunteers soon. All the systems required are mostly there: weapons do damage, armors block it, combat related skills affect it, or decide in dodge rolls, same as attributes. We have spells/abilities, and even some NPCs can use their unique abilities, like corpse eaters who can cast Fetid Spit on you.

Quest system has suffered our lack of programmers. So, tw developers have worked on it, each one proposing his own solution, and leaving it incomplete. Im working on it right now. The dialog system works, but it is a huge monster that nobody but me underestands, and it is under continue work.

One of the things I needed since the beginning was a save system. Took some days to have one and now you can save and load the game. It is still far from finished, but helps to save time when testing quests, combat, etc.

I knew that the UI would be a problem. Somebody volunteered to do the UX design, but after a couple of proposals (one of them I had to modify it a lot) didn't produced anything more. Now I have found very close to me a friend that accepted to work on that, and also provide the graphics, not only the UX. Work should be delivered soon... or at least, I expect so.

I will prepare another video soon, showing the scene in a more complete state. We lack proper terrain support, that is hurting the scene and makes it look unfinished and plain, but I decided to wait for Godot's own terrain system. After trying third party addons, I considered that it wasnt worth the effort of adding two addons (one for terrain, other for scatter), each one with bugs/problems, just to remove them later. Check the video in previous post if ou havent, and see in the description how to support us.

Prototype video

2023-07-17T14:38:00.002-04:00

A bit late, but here is it!

At last, testing O3DE 23.05

2023-06-14T16:52:00.002-04:00

Took some time, but at last I managed to install O3De 23.05. I have no time to extensively test it, because there is a project requiring all my effort, but I took a couple of nights to work with it.

I cant find any dramatic changes, this is not a big leap like 22.10 was. Im not sure if we will see another release like that again. Im not going to get into a proper analysis (a few hours are not enough), but here are some things that O3DE needs to improve:

Project creation process now involves a build, that requires Visual Studio. That is annoying for artists.
Project build is complex. And I mean complex. It takes 12 steps, and manually copying stuff. Lucky us, it is being fixed and I hope we will have the new building script in next major release.
Animation workflow is annoying. As a developer currently working in a project that requires several animations and retargeting, I dont see feasible switching to a workflow that produces separate files for each animation. By the way, animation graphs are fucking awesome, check the video!
Documentation.O3DE severely lacks documentation and tutorials.

I keep my projections: maybe in three more releases it will be usable.

New project

2023-06-06T17:33:00.003-04:00

Time to talk about new project! We started about seven months ago, right after finishing (or abandoning) a combat prototype. It is a single player, 3d, isometric RPG, nothing original or innovative. Quite a big task, and I will explain why right now.

First of all, we are not veterans. Many of the solutions which might be obvious to some experienced developer are not so for us. For example, I asked several times how to implement equiping items in player: clothes, armors, etc. As far as I remember, the solutions are a few:

1- The WoW solution: armors are a texture, you simply change the material in the body part. You can add extra decorations by attaching geometries. Everybody wants to do this, specially the artists.

2- The simple solution: you put the armor, or whatever, on top of charatcer, attaching it to a bone, or something. Useless for animated things like leggins or gloves.

3- The replacement solution: you switch the body part by the piece of equipment. The equiped part must have the same skeleton used in the character body. It is the best solution, and the code isnt that hard.

I tested the last one in Unity. The 3d artist produced a player model and an armor set, and everything worked. this time, for Godot, I implemented the code in a couple of hours, two years ago (Godot 4 was not even alpha, I compiled my own copy). Yet, when I told the new artist, the first thing he told me was: "thats impossible, you cant make edges fit, nobody does that, Wasteland 2 doesnt does it that way. The correct way is to put everything on top of player.". After a couple of weeks I managed to convince him that having one skeleton playing animations, and adding 1/2 whatever more skeletons (one per item) and making them follow the same animation was a recipe for disaster (two months later I had to explain him again that it was a big mistake).

The other problem: nobody in the team reads the design docs. Or the script. Technically, I am the design doc and the script. They ask me when they want to know something.

The bigger problem: we are all volunteers. You think a team of four artists and three programmers is big? Well, I have written 90% of the code. Most of the time, the programmers are busy. Or the game doesnt works and they cant test it and continue working. That wasnt the idea. I am a mediocre programmer, I was supposed to write 10% of the code, design and write.

Godot 4 is quite incomplete. I have serious issues with navigation and obstacle avoidance, but cant find if it is my fault or engine's fault. We cant use Unity, by the way, or Unreal.

We have power cuts. The two main artists have a contract with another team and sometimes they work one hour in the project, per week. They think the project is too big.

And despite that, we have achievements! The guys are good (check my twitter and LinkedIn profile for some images of the artists work), they just cant believe they can do this project if they spend more time on it. All the time they complaint that we are not enough.

Anyway, take a look at our work. We will be releasing videos at some point.

A week in a cuban game programmer's life

2023-02-20T09:46:00.003-05:00

Saturday: bus from Havana is late, so I arrive home at 10 AM instead of 8 AM. A minute after opening the door, lights goes off. In the end, I can only work 2-3 hours in the game, instead of the whole day as I expected.

Sunday: a day for resting. More exactly, an afternoon for the only nap in the week, morning is for my daughter. Also, it is super bowl night, so no work.

Monday: day work, and work in the game during night.

Tuesday: some weird fire in a sugar cain filed produces an space-time singul;arity that leaves half of the country without electricity during 6 hours. My half, of course.

Wednesday: day work, and work in the game during night.

Thursday: a day of work is too much. When I come home, power cut for an hour. When power is back, cell phone network is down.

Friday: day work, and work in the game during night.

Saturday: A human error causes a massive failure in the electrical network, 90% of the country without power. Of course, my 90%. Blackout lasts until 8 PM, so I can only use a couple of hours to work on the game.

Disclaimer: everything before yesterday is blurry in my memory, so the dates might be incorrect. But in essence, that was my week.

Global Game Jam 23

2023-02-13T10:35:00.005-05:00

Testing animation retargeting in Godot 4

2022-12-12T09:52:00.001-05:00

We have finished the project and it is time tostart a new one. This time, I have convinced the team to work on a combat prototype and try Godot 4. After a month, Im quite pleased with the progress. Specially, we have applied animation retargeting, which is a new feature in Godot 4.

In previous project, our main artist devised a clever way to reuse animations, but now we have a native solution. Our first attemp didnt worked, but I found that remapping the armature starting by the feet solved the problem. Weird, isnt it? By the way, you can see the prototype here. WE are keeping it public to let people learn from our mistakes.

And, if you need some animation retargeting tutorial, this is the guide we used:

O3DE test: implementing an isometric camera

2022-11-22T10:08:00.000-05:00

This has been a quite complicated video to make. During process my PC died, a couple of weeks later I had to continue in another PC, and then edited in my new PC. The result: a total mess.

But at least you can see how does O3DE works. Do not consider this as a tutorial! It is just a test I made to learn the engine.

O3DCon videos already available!

2022-11-07T20:02:00.000-05:00

After a long wait they are here! Check the official Youtube Channel.I was preparing a little video about the engine, but my board died.

O3DE successfully installed!

2022-10-28T14:06:00.001-04:00

Yesterday at last I could bring my PC to the work and install O3DE, a long pending task. I have to say Im quite happy with that. Maybe it is a trivial task for you, but for me it required finding something to carry the PC to a place with internet (and save the hours), dodge the power cuts, and sort a lot of difficulties. When I finished, the editor crashed opening the test project. Im fucked! was my first thought. Then, my mouse broke, and I knew I was definitely and totally fucked.

But at home I found that O3DE idoes work! I started playing with it, but you know, it lacks documentation even for the most basic tasks. Amazon should take a look at that. I cant imagine an engine that doesnt documents how to program a game with its scripting language, only with visual scripting. Im sure New World wasnt made only with visual scripting. And no, my mouse didnt work.

I dont know how many hours can I spend with this engine, but Iw ill try to bring my experiences to you in a couple of weeks.

New video: equip or attach items to your player in 3D

2022-10-11T12:07:00.001-04:00

Took me a couple of hours to recreate this little proof of concept. This is not a full tutorial, but can help you to start.

It is here! Testing Godot 4 beta 1

2022-09-16T12:38:00.005-04:00

New video: Battle Gear postmortem

2022-09-15T09:21:00.001-04:00

The first of a serie of videos about mistakes made during development (in spanish).

Remember that now you can support me on Patreon!

I broke my table, but still it is a good month

2022-09-06T15:19:00.004-04:00

Pretty stupid, but I did it. Punched the mouse a bit too hard and broke the table. It comes to join the broken keyboard (not punched, it is just too old) when I need it the most, to complete a novel for a literary contest. Just a couple of hours to finish it!

But I still can say it is a good month. I have had power most nights since last saturday. I am scared. Power cuts are a constant since a year ago.

Anyway, after finishing, Im taking a long vacations from writing, to start working on game development. Soon we will bereleasing our Android game and then we say bye to cuban market. Sorry, it is not the time.

Developing games in Cuba, my experience

2022-08-05T12:15:00.003-04:00

It has been a long time since my last post. I have no time to write, specially not in english, and to make things worse, I found some sort of bug in Blogger. But you are reading this to know about my successful career as game developer, not to hear me complain about bugs.

In 2021 I released my first game. It was a remake of an old cuban game: The Labyrinth of Knowledge. You can get it free on itch.io, but I would appreciate if you leave a tip for a new keyboard and mouse. The future looked bright: there were changes and finally we could create software startups, including the ones dedicated exclusively to games. Two of those startups released successful titles for Android and made money of it. The first dev group, ConWiro, developed a payment gateway that used SMS, but was the second studio, Nameless, who made much more profit of it, with the first cuban made multiplayer game. Even the government showed some interest in pushing game development.

In march of this year I got my papers: I was the creator of a new bussiness oriented to game development. The process also required to create a fiscal bank account, that took a nice bite of my savings: 3000 pesos, frozen there forever and Wakanda. I had my own game group! I could make myself a card with "CEO" on it! Our second project was in late stage, two moths away from the first beta, after a very rough year of work (that year of work... is another story and will be told other time). Then things started to look not so bright.

Remember the SMS based payment gateway? Well, the ISP takes 30% of the sales. Another company, Desoft, which has the only IP in the country allowed to access the SMS server, took another 30% (and guess what, they dont even developed the code to handle the payments via SMS). ConWiro needed a little percent too, so lets say that the game developer ends up receiving just a 30$ of the money paid by the player. From your cut, you have to pay a 15% monthly tax, and another 50% at the end of the year.

So, most of the money was lost in the process, but microtransactions were the way (my first game barely sold 60 copies, even when the demo reached 3000 downloads). I had to personally chase somebody to develop a Godot plugin to access the SMS gateway during a month. There was only month left for the first closed beta and I wanted to release it with microtransactions to test the process. And I got the plugin... just to find that I had no use for it.

ConWiro had to close the SMS gateway, because the operational costs were too high. Im not going to talk about the hidden, weird fees charged by our ISP (ETECSA) and Desoft. Maybe you would ask why are SMS payments so important for us. Nobody uses that in the modern world! Are no other choices, like using a back acount, like civilized people?

Ok, first let me remind you that it is Cuba we are talking about. Bank accounts and debit cards are something new. Payment gateways using bank accounts are... not in early stage, they are in primitive stage. ETECSA has an app (widely adopted and used, I was one of the first user, I bet) that lets you manage and pay for services using USSD codes, but its API is undocumented and they simply take months to answer your request to access it. Enzona is a more modern, web oriented service, a real payment gateway, but its API is poorly designed: you can only use it from a web page, not directly from a mobile/desktop application. And it fails, recently they were 20 days out of service. Not their fault, of course, they depend on ETECSA datacenter and it failed.

We released the closed beta, the public beta, and the final version in a couple of months, but still we dont foresee any significant profit. Im sorry, but I lied to you. Im not having a very successful career as CEO of a game studio. Actually, the other two serious groups have announced that they will not work anymore for the cuban market and neither will I. Seems that we had an very bad case of idealism, but we are cured now.

In a future post I will talk about how difficult will be for us to work for the global market.

The Labyrinth of Knowledge for PC

2021-08-04T15:46:00.005-04:00

At last, we have a port of Labyrinth of Knowledge for PC. You can play on Windows and Linux, using keyboard like in the early 90s, or using a controller (still needs polishing).

You can get it free, or leave a donation, in case you want to help me rebuild my broken PC.

Big problem

2021-07-26T15:53:00.002-04:00

Well, shit happens. My motherboard died a week ago, leaving me in the middle of two projects. Never again will buy Gigabyte, the plans are to get an MSI Tomahawk or some ASUS, probably with a Ryzen 3600. But... I have to wait a couple of months. Or more. Travels from USA to Cuba are rather limited now. Also, I still have to save part of the money.

I had plans to add a few things to Labyrinth of Knowledge and release a PC port, but the few thing end up requiring a refactoring. There is a little bug pending, and then I will release a first version in itch.io or something.

Old Unity project, with recen engine version

2021-07-06T09:21:00.005-04:00

Started in 2015. Came back to life in 2018 and was abandoned again in the beginning of 2019. One and a half year later, resurrected with latest Unity.

Dungeons and Honor, multiplayer made in Cuba!

2021-04-30T11:38:00.000-04:00

Not my work, but Im proud to present you this game being developed in Cuba. Against the odd, we fight and we build our future videogames industry!

If you are interested in the tech details, it uses Unity and Mirror for networking. The VPS is rented in our ISP, but the lowest ping I have experienced is 90ms, the average is 115-120, with peaks of 200.