Vapourware Codexian Game Development Thread

[ERYFKRAD]

I am trying to make a 4X AI that is not cheating with same capabilities as the human player. But if I make a map with much ocean and the AI is done discovering their starter continent it can happen that it is traveling many many turns through the ocean trying to find a new continent.

I guess it is realistic but it can make the AI take forever to find other players.

I know that looks like a dumb path to take, but in fairness, that kind of pattern can happen in a 4X even as a human player. You go one way, you don't find anything. Maybe you could keep going... or maybe you turn around and start searching in another direction. I think the only problem here is that it sometimes travels too close to known areas - arrow 4 passes pretty close by the end of arrow 1, arrow 5 passes by land that's presumably already known, and arrow 9 returns to an area that's been visited already. I would suggest that when it's on "exploration" mode (as opposed to just trying to traverse the ocean to a known location) it should avoid setting locations that are within a given distance from already-visited locations as its next area to check.

Basically, I think doubling back should be ok, it should just spread its targets out more.

I wonder if exploring in a spiral is more effective than zigzagging about.

 

[Hag]

I wonder if exploring in a spiral is more effective than zigzagging about.

W...Well of course it is.

 

[Justinian]

Hopefully there will be something playable by the end of next week.

I could probably post something on itch.io, but honestly there just isn't that much playable content. There isn't even a leveling system in place yet.

Spoiler

My hope is to produce a full playable demo within 2 weeks. I'll probably wait to post something playable online until then.

I will say, the gamedev bug is back. After months of avoiding it I'm finally back in full production mode.

 

[what am i doing]

Finally got my lazy arse around to getting the deposit editor working (soil and terrain editors as well but they're basically just copies of the climate editor). Deposits are roughly equivalent to mine-type RGOs in Victoria 2, in that they allow an operation which produces the deposit's product type to be built there. They have two traits, richness and inaccessibility, the former of which indicates how much of the material there is overall (it's not a fixed stockpile; in practice they are unlimited, just a modifier for how much of the stuff can be gotten at a time for a given production scale) and the latter (intended to be gradually overcome by technology) how difficult it is to get to. Currently the visual representation is that the more accessible it is, the more area of a tile will be covered by it, and the richer it is, the deeper the colour (white indicates that it's not very rich). I had thought about it being the other way around, but I think this way is more intuitive, visually.

 

[Nathaniel3W]

Zeusington gave me some good feedback about my game's description not really matching the enemies on the screen. I say my game is about a battle against an enemy threatening all of existence, and then the trailer and screenshots show a bunch of kids whacking plant monsters and wild animals. I really need to get started on some enemies that convey what's at stake. So here is the first batch:

 

[beardalaxy]

Hopefully there will be something playable by the end of next week.

I could probably post something on itch.io, but honestly there just isn't that much playable content. There isn't even a leveling system in place yet.

Spoiler

My hope is to produce a full playable demo within 2 weeks. I'll probably wait to post something playable online until then.

I will say, the gamedev bug is back. After months of avoiding it I'm finally back in full production mode.

You've got some really cool art there. Hopefully you keep on that gamedev bug! I will offer a word of caution and say that even though you've got the bug, it's good to still limit yourself. Working for 8 hours a day for a week on it will still lead to burn out eventually, and then you'll spend another several months doing nothing. Working for a few hours a day max will leave you wanting to work on it the next day again, and you'll be able to keep up the consistency much more until it becomes a habit, at which point burnout isn't something I really worry about anymore. If I skip a few days to do other things, I know I'm not going to lose my drive to pick things back up again whereas before, if I worked on the game for 50 hours in a week and then had a day where I couldn't do that, it was game over for months.
At least, that's how it works for me. Take it or leave it lol xD

Zeusington gave me some good feedback about my game's description not really matching the enemies on the screen. I say my game is about a battle against an enemy threatening all of existence, and then the trailer and screenshots show a bunch of kids whacking plant monsters and wild animals. I really need to get started on some enemies that convey what's at stake. So here is the first batch:

Those are some truly terrifying abominations! Like fleshy, legally distinct beholders. Cool art too.

I haven't been around here in a while, but I have been working a LOT on my game. Long story short, I started doing a playthrough of the game after finally finishing everything, just to see if there were any bugs and such before I let other people play it. Finished that playthrough just a few days ago, so now I'm one step closer to getting the beta out there! If you want to read more, this devlog has some more information about that:
https://after-midnight-games.itch.io/godsdisdain/devlog/742030/19-road-to-beta

 

[Justinian]

Hopefully there will be something playable by the end of next week.

I could probably post something on itch.io, but honestly there just isn't that much playable content. There isn't even a leveling system in place yet.

Spoiler

My hope is to produce a full playable demo within 2 weeks. I'll probably wait to post something playable online until then.

I will say, the gamedev bug is back. After months of avoiding it I'm finally back in full production mode.

You've got some really cool art there.

Yeah, I'd love to take credit for it but it's just AI art (it was a pleasant surprise to find it can now produce logos and proper text). I hope it won't clash too much with the low res pixel art of the actual game. But honestly, I'll take some graphic style clashes as long as I have a better interface than my last game.

Hopefully you keep on that gamedev bug! I will offer a word of caution and say that even though you've got the bug, it's good to still limit yourself. Working for 8 hours a day for a week on it will still lead to burn out eventually, and then you'll spend another several months doing nothing. Working for a few hours a day max will leave you wanting to work on it the next day again, and you'll be able to keep up the consistency much more until it becomes a habit, at which point burnout isn't something I really worry about anymore. If I skip a few days to do other things, I know I'm not going to lose my drive to pick things back up again whereas before, if I worked on the game for 50 hours in a week and then had a day where I couldn't do that, it was game over for months.
At least, that's how it works for me. Take it or leave it lol xD

I took the last 2 days off, for me though I can't take too much time off because I want to ingrain the habit of working on it regularly. I can't imagine taking a lot of time off at this point though, once I put a certain amount of work into something I want to see it through.

 

[Nathaniel3W]

Justinian bro, that's AI? That looks great. I wouldn't have guessed. Man, AI really is going to take over video game asset production.

 

[Justinian]

Justinian bro, that's AI? That looks great. I wouldn't have guessed. Man, AI really is going to take over video game asset production.

Yeah, and I've only scratched the surface myself. You still need to do a bunch of editing and I don't see that changing anytime soon, but it will speed up production by quite a bit.

 

[The_Sloth_Sleeps]

Justinian bro, that's AI? That looks great. I wouldn't have guessed. Man, AI really is going to take over video game asset production.

Yeah, and I've only scratched the surface myself. You still need to do a bunch of editing and I don't see that changing anytime soon, but it will speed up production by quite a bit.

I was going to say, the first thing I noticed is that it looks like AI art. Its got that lack of consistent style.

You are right about only scratching the surface. You will be able to churn out endless variations just like everyone else.

 

[Justinian]

Justinian bro, that's AI? That looks great. I wouldn't have guessed. Man, AI really is going to take over video game asset production.

Yeah, and I've only scratched the surface myself. You still need to do a bunch of editing and I don't see that changing anytime soon, but it will speed up production by quite a bit.

I was going to say, the first thing I noticed is that it looks like AI art. Its got that lack of consistent style.

You are right about only scratching the surface. You will be able to churn out endless variations just like everyone else.

Unfortunately, "predictions" AFTER the fact don't mean much of anything. Anyone can claim "I knew that was AI art" after it's revealed it was AI art.
I also disagree that the style isn't consistent. Where I intend it (and am able) to make it consistent IE: portraits, it is the same style.

 

[The_Sloth_Sleeps]

Justinian bro, that's AI? That looks great. I wouldn't have guessed. Man, AI really is going to take over video game asset production.

Yeah, and I've only scratched the surface myself. You still need to do a bunch of editing and I don't see that changing anytime soon, but it will speed up production by quite a bit.

I was going to say, the first thing I noticed is that it looks like AI art. Its got that lack of consistent style.

You are right about only scratching the surface. You will be able to churn out endless variations just like everyone else.

Unfortunately, "predictions" AFTER the fact don't mean much of anything. Anyone can claim "I knew that was AI art" after it's revealed it was AI art.
I also disagree that the style isn't consistent. Where I intend it (and am able) to make it consistent IE: portraits, it is the same style.

Its got nothing to do with predictions. If I see an AI generated main menu screen the dev is too lazy to be bothered to learn how to make art then that tells me the game cant be too hot either.

 

[Justinian]

Justinian bro, that's AI? That looks great. I wouldn't have guessed. Man, AI really is going to take over video game asset production.

Yeah, and I've only scratched the surface myself. You still need to do a bunch of editing and I don't see that changing anytime soon, but it will speed up production by quite a bit.

I was going to say, the first thing I noticed is that it looks like AI art. Its got that lack of consistent style.

You are right about only scratching the surface. You will be able to churn out endless variations just like everyone else.

Unfortunately, "predictions" AFTER the fact don't mean much of anything. Anyone can claim "I knew that was AI art" after it's revealed it was AI art.
I also disagree that the style isn't consistent. Where I intend it (and am able) to make it consistent IE: portraits, it is the same style.

Its got nothing to do with predictions. If I see an AI generated main menu screen the dev is too lazy to be bothered to learn how to make art then that tells me the game cant be too hot either.

Sure let me add a year+ of development time to learn how to draw portraits people will see for a couple of seconds while selecting their character.

 

[Zeusington]

The funny thing about AI asset creation is that it's still more efficient to use it and simply work it until it doesn't "look like AI" anymore than it is to learn the skill set, make something worse, or wageslave / sell out for the capital to hire an artist. Most of what I've seen people replace with AI is extremely banal, stupid shit like backgrounds, side characters, etc. Teenagers and demographics with lesser means are the only ones who try to pass off mostly unaltered AI stuff as a centerpiece.

There's always the landmine that is using images as training data without consent, or outright unintentional AI-spat plagiarism, but you can always just generate a mannequin with the pose, shading, and proportions you want (extremely easy) and detail on top of it in your style. Hell, artists will use this as a base with nary a sliver of reserve half the time. Borrowing imagery and transforming it is literally how humans already do it, and positively fucking nobody is paying the artists for looking at an image for inspiration, AI or not.

tl;dr don't be retarded, put in some work of your own, and AI is fine.

 

[beardalaxy]

I think AI is great for concept art, placeholders, and inspiration, but I would never ship a product with it. I'd heavily advise against using it unless your game is going to be free or proof-of-concept or something. People will catch on to it real quick when one quirk puts the rest of the art into question. AI art, and just AI in general, is also on the verge of being regulated so I wouldn't lean too heavily on it. Something as simple as a platform not allowing it will limit your reach. It's a ticking time bomb in many ways.

If you can't afford to pay artists a lot of money for the style that you want, you may have to adjust your style and change it to a simpler one. It can also be helpful if you know anyone personally who does artwork, they might help you out for cheaper than some guy on the internet would. I paid $500 for 15 actual paintings because I did it in a bulk deal with an old acquaintance. Just tell them what you would need, what your budget is, and see where it goes.

 

[The_Sloth_Sleeps]

Anyone can learn to draw. I don't believe in hiring artists either. Its not economically feasible. AI is not the solution either.
The solution is to work within your (or your friends) ability and focus on the game.

The difficulty with teaching yourself art is, that as you improve you will become unhappy with with the asset you made at first and then want to remake them.

And so you remake them, and improve more, and the cycle might never end.

 

[The_Sloth_Sleeps]

The funny thing about AI asset creation is that it's still more efficient to use it and simply work it until it doesn't "look like AI" anymore than it is to learn the skill set, make something worse, or wageslave / sell out for the capital to hire an artist. Most of what I've seen people replace with AI is extremely banal, stupid shit like backgrounds, side characters, etc.

Backgrounds, portraits, UIs, has "that" sort of funky look. I'm not convinced its easy to work the AI feel out.
I have tried it and have always given up and just used methods I used prior, such as photos and 3D modelling.

 

[beardalaxy]

I don't know if this has been brought up in this thread yet, but while we're on the topic of making your own art, there is a pretty cool game coming soon that actually teaches you how to make pixel art. There's a demo available atm too.
Pixel Art Academy: Learn Mode

And here is a YouTube video of a guy talking about it.


My own pixel art improved quite a bit when I was editing tiles, when I was trying to do that tileset swap. I spent hours and hours looking at how the tiles were constructed and was able to make some of my own. In the final game, there are actually a couple icons I made from scratch! So looking at other art and replicating it or tweaking it can help a lot.

 

[The_Sloth_Sleeps]

I don't know if this has been brought up in this thread yet, but while we're on the topic of making your own art, there is a pretty cool game coming soon that actually teaches you how to make pixel art. There's a demo available atm too.
Pixel Art Academy: Learn Mode

And here is a YouTube video of a guy talking about it.


My own pixel art improved quite a bit when I was editing tiles, when I was trying to do that tileset swap. I spent hours and hours looking at how the tiles were constructed and was able to make some of my own. In the final game, there are actually a couple icons I made from scratch! So looking at other art and replicating it or tweaking it can help a lot.

Interesting. I think pixel art is a great way to start. Ultimately you are investing in yourself by being able to draw. If you get even half decent at pixel art then you can launch into much more advanced stuff, its really not that hard.

The best investment I ever made was a computer tablet, and Krita, for free. I started out with a sketch book and pencils years ago but I found progressing that way more difficult. With pixel art, the complexity is reduced and if you learn a software package and follow some tutorials you can see the progress by the day. Yes its worth investing in yourself.

 

[Zeusington]

Anyone can learn to draw. I don't believe in hiring artists either. Its not economically feasible. AI is not the solution either.
The solution is to work within your (or your friends) ability and focus on the game.

To be fair, being a dilettante is necessary when it comes to gamedev, so getting some skill in art is necessary, but there are several ways to go and AI can help without being a crutch. If you're going to do flats and then draw muscles on, for example, you can just get AI to spit something out and color-crunch for a better place to start.

Improving pixel art may cross-pollenate with other types of imagery, but I do think learning to produce quality, communicative imagery at any resolution will be a better foundation than jumping into line ratios, (not using) pillow shading techniques, appropriate use of dithering, etc. This is coming from a guy who learned pixel art in MSPaint 23 years ago by editting sprite sheets and just noticing some patterns. It *is* very good for learning how to iterate, however, which is *extremely* important for gamedev, but if you can't abstract that concept and internalize it for other applications then it won't be the sauce.

AI isn't the "solution" nor would I ever recommend it to be, but we've been using AI as a tool for many years now in differing levels of sophistication as a part of digital and practical development. Just don't be an idiot and use un-editted AI images as a paid asset, or have AI do a pass over plagarized images like a bunch of WotC dumbasses did for MtG cards recently.

 

[Justinian]

One of the more annoying coding tasks (level-up system) is done. The level-up system will feed me all attacks until attack slots are filled, then spawn minor stat upgrades instead. There are still some tweaks that need to be made (different characters have different attack slots which the system doesn't currently take into account, I haven't implemented any passive items or item slots yet), but the core system and all its annoyances and frustrations are now behind me.

Spoiler

 

[Twiglard]

In order to implement doors and windows, I needed some generalized way of adding holes to geometry for physics and rendering. Because it's limited to axis aligned bounding boxes, we can phrase a general problem:

When cutting a hole in an AABB what are the resulting AABB's?​

Actually this can be done branchlessly using a small LUT (9*16 bytes).

visualization

In the examples below, the cut-out rectangle is red.

top left corner case

LUT value is { 2, {{H1, R1, R0, H1}, {R0, R1, H1, R1}} }

left side case

LUT value is { 1, {{H1, R1, R0, R1}} }

center case

LUT value is

Spoiler

{8, {
    {R0, H0, R0, H0},
    {H0, H1, R0, H0},
    {H1, R1, R0, H0},
    {R0, H0, H0, H1},
    {H1, R1, H0, H1},
    {R0, H0, H1, R1},
    {H0, H1, H1, R1},
    {H1, R1, H1, R1},
}}

The trick is that for each of the (xmin, xmax, ymin, ymax), coordinate is taken from one of enum class location { R0, R1, H0, H1, };

 

[Twiglard]

Tried integrating SMAA to smooth the borders between individual tlies and reduce the amount of minor pixel extrusions that would occur where small landmass features turned into one visible pixel when viewed at a distance. SMAA did not help; it produced no visible change and on particularly low settings (such as depth-based edges) would generate a lot of "sparkling" visual artifacts.

Why'd you need SMAA to draw alpha masked sprites? This could be something else like UV off-by-half or depth buffer imprecision.

 

[what am i doing]

Tried integrating SMAA to smooth the borders between individual tlies and reduce the amount of minor pixel extrusions that would occur where small landmass features turned into one visible pixel when viewed at a distance. SMAA did not help; it produced no visible change and on particularly low settings (such as depth-based edges) would generate a lot of "sparkling" visual artifacts.

Why'd you need SMAA to draw alpha masked sprites? This could be something else like UV off-by-half or depth buffer imprecision.

I don't use alpha masked sprites. I did try that as a performance measure waaaay, way back when I was using OpenGL, but the results were not very crisp and I didn't like it.

 

[Twiglard]

Holy fuck, this actually works.

Naturally you can have several holes affecting a single AABB, overlapping holes, and so on.