The unreal team wouldn't accept the engine taking 30 seconds to add a node on a "large" scene of 25 unique objects with 87 draw calls instanced 10k times, which is why AAA games use that engine and not godot. AAA games use modular assets heavily as well. I don't think having exponential complexity on adding a node to a scene or renaming a node is the fault of the person using the engine.
Game Developers of Kiwi Farms
- Thread starter Judge Dredd
- Start date
- Joined
- Nov 11, 2020
Zero-G Spelling Bee : arrange the letters to solve the puzzle with the correct word (crossword style) and do so while the other letters bounce around off the walls and shit and try to fuck up your word. spend abilities to shield you words for a short time or pin specific letters in place.
NerdShamer
kiwifarms.net
- Joined
- Aug 21, 2018
So, guys. I've been wondering if anyone else had ran into people defending the use of AI because it allegedly speeds up the process or makes things more accessible for the players?
For instance, I ran into someone who made an list on how something like ChatGPT can improve Fortnite (or any other online game) and I can't stop laughing at how retarded it is
For instance, I ran into someone who made an list on how something like ChatGPT can improve Fortnite (or any other online game) and I can't stop laughing at how retarded it is
- Joined
- May 10, 2020
I think Thomas Brush made a video about it a while ago, and its main point was basically that everyone will have to use AI cause if you don't, others will, and they'll beat you in this competitive market.
There it is:
Homocull
kiwifarms.net
- Joined
- Jun 21, 2018
That outro gave me a stroke. Like clockwork, it's a Youtube grifter with breadtube lights who only shows interest in how fast a product can be pushed to market. Clearly the only metric that matters.
I've used nearly every open-source AI tool to come out and looked at (but not used) all the SaaS cloud shit as well.
My conclusion is that the only parts of a game it can accomplish decently with minimal effort on your part is:
- Writing boilerplate functions
- Writing random dialogue (Ubisoft)
- Voicing characters (THE FINALS)
- Generating PBR textures (RTX Remix)
- Coming up with a lot of misc dialogue yourself in one sitting is challenging. I'm talking Bad Company type chatter, not pivotal story moments.
- It's a task impossible to accomplish solo. GPT-SoVITS has the best quality and voice cloning for a local TTS, plus handles emotions including laughs very well. From the same guy that made RVC and not very well known outside of the Chinese space.
- This isn't even a new thing, it was possible before the existence of AI tools.
Every other form of generation either comes out as being evidently AI, or requiring too much work to make it not look like so, thus being counter to an efficient/beneficial workflow.
For non-artists, AI art can be salvaged if you are both using custom models and working with jagged/aliased drawings. You will have to redo the coloring and shading yourself however, plus fix incoherent linework (which there are many). It's such a specific workflow that I'd recommend learning to draw instead.
But maybe you have no sense of quality control. Maybe all you want to do is push out a fat fucking turd out the door. Maybe you're retarded, or you have a feeling that the majority won't be able to tell anyway.
This is "Mekkablood". It's probably the first demonstration I've seen of someone trying to make somewhat of an actual game whilst using AI for a huge portion of the assets.
Do I even need to point out anything? The crusty as fuck Dall-E 3 images. The washed out SunoAI music trying to mimic rock. The gameplay that was likely done entirely with an AI's help, lacking any unique touches you'd normally see from a hobbyist dev.
When I say hobbyist dev I am making a clear distinction between a person with a concrete vision or at least a passion, and a person who is only interested in making bread. The only compliment I can give this game is the pile of trash rustling about inside the cockpit. It's the only human thing in the whole presentation, dude couldn't even be fucked to find a nice typeface and just had Dall-E generate the "Features" header in the description.
My point is that any self-respecting, self-aware dev would feel embarrassment at the idea of showing a game that is so blatantly being held together with free, AI web tools. Let alone entertain the concept.
Indie devs are often not the brightest bunch, so when you hear of a "competition that uses AI", this is what you should expect. It's a common denominator averse to learning, and when they hear that AI can do all this work for them (from whichever YT faggot they're currently watching), they jump excitedly at the prospect.
And then they make a Mekkablood.
How else are you going to fucking move?
Last edited:
- Joined
- Jan 1, 2023
- Joined
- Mar 1, 2021
This is great! Even if you didn't want to optimise further you can do a lot with the budget.
- Joined
- Aug 2, 2021
I hate this facet of the gaming culture. Most popular games are not 'games', they are 'products' that must appeal to everyone, and the 'experience' they provide must be the same for all users so that we don't 'alienate' some of our customers and make them spend less. Shoot me.
- Joined
- May 10, 2020
It appears that there are more and more games being released every year, so if you want to make money, you'd better use every shortcut you have to publish your game quicker until the market has become too competitive.
I used to be an idealist, and scoffed at marketing and market research cause, you know, video games are an art, and I'm an artist and not some sellout fag who only cares about profits. Besides, if the game is excellent, it can stand on its own merits without any marketing, right?
But as I grew older, I also got more cynical. I realized that I'm actually not that talented and I won't become the new Edmund McMillen or Eric Barone. But I still want to be a full-time game developer instead of a web monkey working on yet another generic CRUD app that I would never use personally. And to achieve that, I need my games to make money, and since I'm just a mediocre dev, I need to conduct market research. I was going to make a roguelite, so I decided to check how many roguelites were presented on last Steam next fest, which was about 300. Sure, you might say that 90% of everything is crap, but that still leaves us with 30 games that are pretty good. Which is a lot, given that roguelites are intended to be replayed over and over.
And I know that my approach of wasting time building my own engine doesn't look appropriate to this market situation, but I'm also not pretending to be a completely rational and reasonable person and if I wasn't such a control freak, I'd just use Unity or Godot and whatever AI tools are available.
But as I grew older, I also got more cynical. I realized that I'm actually not that talented and I won't become the new Edmund McMillen or Eric Barone. But I still want to be a full-time game developer instead of a web monkey working on yet another generic CRUD app that I would never use personally. And to achieve that, I need my games to make money, and since I'm just a mediocre dev, I need to conduct market research. I was going to make a roguelite, so I decided to check how many roguelites were presented on last Steam next fest, which was about 300. Sure, you might say that 90% of everything is crap, but that still leaves us with 30 games that are pretty good. Which is a lot, given that roguelites are intended to be replayed over and over.
And I know that my approach of wasting time building my own engine doesn't look appropriate to this market situation, but I'm also not pretending to be a completely rational and reasonable person and if I wasn't such a control freak, I'd just use Unity or Godot and whatever AI tools are available.
- Joined
- Apr 21, 2024
- Joined
- Aug 2, 2021
Pretty sure Edmund got a boost/visibility from having his games featured on NG. So imagine out of those 300 games, people saw yours first.
So it seems like knowing people is how you get ahead. You'd be surprised how much engagement and popularity is fabricated.
90% of publishing and marketing is paying for engagement.
- Joined
- Aug 23, 2018
This is always something that baffles me. I remember some e3 like where the featured dev had no interest in game development, they had just taken a
I saw a picture recently of some AI image used to advertise some local party. That's the future of AI to me. A more advanced version of clip art and comic sans.
It's a bit strange seeing this. I assume you're younger, because I remember games like FUSE, which was focus tested from this
If you want to make a roguelike, make one. Just don't expect it to sell. It's up there with pixel art platformer and card game as an overstaurated genre.
I don't know but I know it when I see it, it's a mix of visual and audio feedback. If you compare Fallout New Vegas to Apex Legends you can tell FNV has shit gunplay but it's hard to pinpoint why. With most things that come down to feel you just keep iterating until it feels right.
- Joined
- Mar 1, 2021
- Joined
- Aug 23, 2018
I didn't want to mention it because I don't feel I'm qualified. I will give my thoughts though since "visual and audio feedback" is a bit vague.
First, reaction time. When you pull the trigger the gun should react. New Vegas is bad at this. FO4 is way better. Doom has a slight delay on most guns (though mods can fix this, including a Doom mod I can longer find that had a name like "quick doom" or something like that).
Guns should have power and the world should react to them. Sparks, smoke, muzzle flash, shell casings. All add to the effect. No one element is needed, but the more you have the better. Likewise, the target should react. Blood splatter, bullet holes on walls, small objects should shatter. Etc. Not all are needed, but the more the better.
Damage values should also reflect the weapon. Having a gun that sounds like an atom bomb but does chip damage feels bad.
Related to this is how much realism you want. In Timesplitter 3, shell casings come out the wrong side of the gun, because having them fly across the screen looks cool.
Last edited:
- Joined
- Jul 4, 2024
I have been working on developing board games, as its something that I can do in my spare time. I find the inherit limitations to be liberating, as there is a much lower expectation of complexity and graphical fidelity. One does not have to spend inordinate time on jiggle physics or weapon animations to make a successful product, so greater emphasis can be placed on actually designing something that is fun to play. Much easier to make radical changes to the whole concept during production, and there is a much lower sunken cost to shelving a game if you aren't pleased with the product.
I'm working on my third game, the first two being shelved, and I think I have something fun. It has done well in playlisting, all that is left to do is put some ideas to paper and do some graphic design. There is a wonderful program called cardmaker that makes iterating playing cards a breeze. I encourage anyone who wants to make a game but doesn't feel like they can choose to try their hand at making traditional games
I'm working on my third game, the first two being shelved, and I think I have something fun. It has done well in playlisting, all that is left to do is put some ideas to paper and do some graphic design. There is a wonderful program called cardmaker that makes iterating playing cards a breeze. I encourage anyone who wants to make a game but doesn't feel like they can choose to try their hand at making traditional games
- Joined
- Sep 28, 2022
good sound effects
good visuals
good 'feel' (responsive controls, recoil mechanics, and aim that all feel good and appropriate for the enemies you fight)
the 'feel' part depends a lot on what kind of shooter you are going for. a gun that feels great in a cod style cover shooter would feel terrible in a doom eternal style arena shooter, and vice versa.
- Joined
- May 10, 2020
Good point, I should've considered that Edmund had been making games for about a decade before hitting it very big with Binding of Isaac. But then again, he was doing that when web games were still profitable, nowadays barely anyone plays them and monetizing them is harder. You'll have to switch to mobile games or PC games.
No, I remember that game. They were correct to do market research, but their logic is opposite to mine: they tried following a trend, even though there were already plenty of games catering to that audience, while I'm deliberately trying to avoid trends hoping that there's a free niche.
Also, if Overstrike was released today the way it was unveiled back in 2011, it'd probably be hated due to using "millennial writing" or something.
I spent 2 hours banging my head against the wall because the changes I was making to the Godot engine were not being reflected in the output and the debugger was showing the wrong line numbers when stepping through. Turns out I had changed the build config and the binary file name changed, so I was running the wrong binary the whole time 
but at least I finally got navmesh baking under 20ms on the main thread
but at least I finally got navmesh baking under 20ms on the main thread
- Joined
- Oct 31, 2023
Since this is the only thread for gamedev concepts (Ill probably make one in the future for Game pitches and concept discussions) I wanted to ask something
Which is the open world/sandbox game with the highest NPC routine score? The NPC routine test is a test I devised to test game AI in sandbox games. There are different ways to test AIs in different games depending on what the goal is (I have couple of them for different game genres). In a sandbox game the goal of the AI should be to enhance immersion, broadly speaking. To do that, the most prominent goal of the NPC (Human or animal) is to perform its own routine, not just action but sequence of actions which logically make sense in the game context (guards surveil the area, animals traverse hunt/escape, Trucks deliver things, businessmen go to work etc etc). Reactivity is different from this as that refers to actions the NPCs perform in relation to the player. Generally its a bit too much to expect NPCs to behave autonomously but the less they follow a logical sequence of actions autonomously, the worse the NPC AI is, generally. Take GTA, a good way to perform the test is to follow a random car on the street or an NPC to track their trajectory, track their actions, whether they go in and out of places, whether their attire matches their actions logically, things like that (GTA is just an example, this can be done in Far Cry Witcher Saints Row RDR anything. Im just used to doing this in GTA). From my tests in GTA San Andreas, I can judge the AI to be 3/10 in quality since the NPCs have circular trajectories which doesnt match vehicle type or attire, NPCs spawn only in the vicinity of the player, NPC attire/vehicle changes based on area even though within that bubble they dont act logically consistent, things like that (this is not shitting on SA, its a product of its time. Its also possible to go so fast that theres a time delay in NPC spawning, in theory anyways). The game with the highest score in this test Ive seen personally is MGSV which comes close to a 9 and Ive rarely seen a game which comes that close (MGS3 and BOTW come somewhere around 6 or a 7).
Which is the open world/sandbox game with the highest NPC routine score? The NPC routine test is a test I devised to test game AI in sandbox games. There are different ways to test AIs in different games depending on what the goal is (I have couple of them for different game genres). In a sandbox game the goal of the AI should be to enhance immersion, broadly speaking. To do that, the most prominent goal of the NPC (Human or animal) is to perform its own routine, not just action but sequence of actions which logically make sense in the game context (guards surveil the area, animals traverse hunt/escape, Trucks deliver things, businessmen go to work etc etc). Reactivity is different from this as that refers to actions the NPCs perform in relation to the player. Generally its a bit too much to expect NPCs to behave autonomously but the less they follow a logical sequence of actions autonomously, the worse the NPC AI is, generally. Take GTA, a good way to perform the test is to follow a random car on the street or an NPC to track their trajectory, track their actions, whether they go in and out of places, whether their attire matches their actions logically, things like that (GTA is just an example, this can be done in Far Cry Witcher Saints Row RDR anything. Im just used to doing this in GTA). From my tests in GTA San Andreas, I can judge the AI to be 3/10 in quality since the NPCs have circular trajectories which doesnt match vehicle type or attire, NPCs spawn only in the vicinity of the player, NPC attire/vehicle changes based on area even though within that bubble they dont act logically consistent, things like that (this is not shitting on SA, its a product of its time. Its also possible to go so fast that theres a time delay in NPC spawning, in theory anyways). The game with the highest score in this test Ive seen personally is MGSV which comes close to a 9 and Ive rarely seen a game which comes that close (MGS3 and BOTW come somewhere around 6 or a 7).