Been a minute since the last blog but I’ve been grinding on my project and been messing around with AI a lot. I have a system in play where I can voice/type my ideas and it’ll translate into my email so I can put that idea into Midjourney (because Midjourney doesn’t have an API atm) so I can get phenomenal looking concept “art”, which then goes to ChatGPT to give me brainstorm ideas and positives/negatives about that idea, then all that will go into my database with everything I need. Btw when I said voice, I got an automated system where I can use my watch or phone and say an idea and that will go to OpenAI’s whisper transcription and send that to my email. So when I’m out and about, if I think of an idea I can just quickly open up the recorder on my watch, record whatever, and when it’s done it’ll go to my phone and there is an app which it can constantly check for new files and upload that file to my Gdrive and that’s when it goes to Whisper and translate it automatically then send it to my email. I wish I can do everything automatically but I have to manually put in the idea into Midjourney for everything else to work unfortunately. Btw I’m still using pipedream to do this.
I’ve messed around with this for a while and changed it’s setup a couple times because I was using OpenAI’s image generator but it was very ass compared to Midjourney. Other AI tool’s I’ve been experimenting was adobe firefly which was ok but has it’s issues but photoshop man.. whoa it’s generative imaging is insane. I’ve yet to use it but last few days I’ve seen it everywhere on twitter. Basically you have an image and you want to have more stuff outside of your image or even have 2 images and have it fill the background for you. Shit’s fkn crazy cool.
Another thing I was messing around with is Cascadeur which is an AI solution of creating natural looking animations, I haven’t messed around with it too much mainly due to figuring out how to attach an object to the mesh because there’s like no tutorials of doing that and it’s documentation is wrong related to that. I’ve went through a few videos because there’s nothing much on the topic but I found one that finally met my needs which was a one out of like 20 videos I’ve watched that actually told me how to properly do it. Timestamped below:
https://youtu.be/wMXwmobKwb0?t=449
The version is a little old but it’s generally the same process.
Another tool I’ve been seeing around lately, actually literally in the last few days again (AI is evolving quick) is generative dialogue. NVIDA ACE and Replica Smart NPC’s UE Plugin looks very interesting and could even change AI in games dramatically. Because think of this, AI in games are very static, programable to whatever the game is set to do and mostly useless in multiplayer games. Image if you can just ask the NPC about something and it’ll respond you what it knows in it’s memory. Such as in a survival game you get it to collect wood and on it’s way back it or you can tell it if it’s seen anyone and what they are holding, wearing, where they are going etc. Or even just it’s general experience if doing specific things is worth it or even if it’s experience can be better/more efficient. Possibilities is endless.
AI is crazy good and I’m 100% on board with it. I understand a lot of people are against AI as a whole because it’s making artists somewhat irrelevant but AI still has it’s flaws especially in art. AI art is fucking amazing but one key thing that you can’t really do is doing something very specific to your needs and the fact that you can’t edit it yourself manually via .psd file etc. AI is technically art, any type of technology is art. Using it as a tool can only help you get there faster and better. If you can’t accept AI in this current era of technology you will be arrogantly behind, because this is part of society now. Why not embrace it, it’s cool fkn shit seeing all this new stuff develop which helps us in nearly every way. I get that art is supposed to be “human made/learned” but this is too. You can still be an artist as a skill. It’s like driving a car compared to walking, it makes going the destination much faster and convenient but walking gives you the exercise that you body would appreciate.
Anyway that’s enough of AI for now.. So in the last month and a half, I’ve been mainly working on reload animations for weapons. I’ve got a dynamic system that I have cut up animations and can be modular. This will be especially useful when I need to add more weapons and just select which animations I want that would fit and this would especially will be useful for modular weapons. A couple animations that have been a little pain to do such as the pump M870 shotgun, the bolt action Intervention sniper and the M249.
The M870 took me a whole week to get working properly because I had the 1 by 1 reload but I had that twice because I’ve made it so when you have less ammo than the visual ammo shown on the side of the shotgun it’ll grab it from the side instead of from your character. This took soo long mainly due to trying to sync it all up properly.
M249 didn’t take all that long but it was a weapon that you can visually see the ammo belt on the side of the weapon. So I also had to do two different sets of animations because if I was reloading I wanted to have it so if I have enough ammo in the inventory or gun depends if I need to change out the magazine/ammo-box or just put in a couple bullets from the inventory to the weapon. I found this setup on the marketplace and thought it was a very innovative way to reloading an LMG.
The Intervention I kinda just stopped doing because with the way I’ve set up weapons, the weapon animations I currently have are not really compatible with what I want so I’m currently just leaving it for now and want to work on other more important stuff then come back to it when I want. So yeah reload animations took a big chunk of development time but I wanted to make sure it’s all working and looking fairly decent.
That’s pretty much it for game dev updates, I have added Weapon and Ammo Dispensers so I don’t have to put down ammo or restarting the play session when I need more ammo, plus it’s cool. Another thing is I made a weapon rack showing all the weapon’s info like form my google sheets. Around the end of July I got plans of putting out a demo of the game and also I want to create a video of my progress within a year of working as an independent solo developer.
Honestly I haven’t really been playing much of anything. I stopped playing Rust because Dead Island 2 came out and it was fairly ok. Not much content really, very short. Hope more dlc comes out for it but it was enjoyable. Not too much of a fan of the card and fury abilities tho. Played cursed Halo which was pretty funny but annoying at times. But other than that, shit is still just soo stale that I didn’t even play or stream anything for 2 weeks after that. A couple of the OG boys I used to play with back in the day in Ark hit me up to relearn everything and check out the tek stuff to get ready for UE5 version and the sequel. The game is still a netcode mess which I’m not surprised in the slightest. There’s 188 creatures in ark now which that’s not including variants which is insane because when the game first came out it was only 25. There’s so many new different dinos with different abilities even now which I’m not too much a fan of but this game is casual so it’s whatevs.