(*) Credit values are approximate examples only, actual consumption varies significantly; Unity may change rates at any time. Check the Unity Dashboard for accurate info.
Request Categories & Credit Prices (*):
AI Assistant Message: ~27
Generate an image: 1-6
Generate a texture: 1-9
Generate a 3D model: 8-13
Generate SFX: 1-5
Generate music: 3-5
Text to speech: ~3
Animation=txt>motion: ~5
Generate a skybox: 2-5
Generate video: 5-15
Attention Unity Credits reset automatically at the start of each billing cycle, and any unused credits don’t roll over.
BUT since Unity AI is a subscription (purchased separately for the free Unity Personal plan or included in other Unity plans) with a monthly allocation of Unity Credits,
I found that this “refreshing” affects only the allocated credits and not one-time purchased bundles (as shown in my illustration), which expire 12 months from the date of purchase and do not renew.
Thanks for the info, though. I was curious about it to see if I could save time during the early prototyping phase where I just need to move fast and quality doesn’t matter. With numbers like that however, you saved me the bother of installing and setting up.
That’s actually not a good use case for AI in my experience. If you dump a couple random assets of the kinds you need into your project and use them as you see fit you can prototype 100x faster than with AI.
The AI workflows are hampered by long generation times, and on top you have to create an asset, enter the prompt, configure any settings and when you get the results, pick one of the solutions and apply it. This will easily cost you 30-90 seconds of your time for each individual asset! And that’s besides the point of the asset likely being very unfit even for prototyping, ie the size (or playtime, or visibility, etc) just doesn’t match so you’ll try another prompt …
Whereas if you just “browse and select” what’s already in the project that’ll be done in 5 seconds per asset. Especially if you use one of the prototyping assets packs (ie Kenney’s or Synty) you will move 10-100x faster without AI generated assets and the results will truly look like a prototype, not a hodgepodge of random assets.
With code generation this turns around entirely, where using AI will get you highly workable if not production-ready results 10x to 100x faster.
Prototyping, mate. Prototyping. Capsules, cubes, and spheres the whole way.
It’s the scene infastructure and code generation I’d be after. While most devs knows that a one-button ‘make me a game’ isn’t going to deliver what your average person would want, that’s exactly the kind of tool that would be useful for rapid prototyping.
Unfortunately, that’s also the part with the highest cost so I guess for now I’ll just do it the old-fashioned way.
I made a playable prototype in 2 weeks with just the free Gemini to generate the code ie character controller, ai movement, attacking & damage, spawn system, loot magnet, day/night cycle, camera tilting based on north/south movement, HUD, etc etc etc.
For prototyping most of these are individual scripts with few interdependencies, not whole systems. So you can generate those for free right now. This has saved me an extra 2 weeks easily!
I went with Gemini primarily due to that reason and the fact it’s giving me an answer within 1-3 seconds, as opposed to 10+ seconds common with more elaborate AIs. The latter has their use once you need to refactor and turn them into actual systems.
That’s pretty neat looking and surprisingly impressive if you didn’t have to touch any code yourself!
Though, the time-scale that would get me excited is more like, minutes or even just a couple of hours. Anything longer than that and it moves into the territory of me being able to do it entirely by hand using all the tricks and libraries of code I’ve made over the years. And again, I’m not necessarily looking for a result that would be presentable to even a seasoned game tester. Just something that works close enough to what I ask for that I can proof-of-concept ideas, tweak things, maybe fix a few borked issues generated to ensure the idea is solid enough to be worth pursuing, etc.
I realize even THAT is a pretty big ask but really, if it can’t do that much, then I have almost no use for it, especially if it costs me money.
@makaka-org
Yeah, unfortunately, there are a lot of unsophisticated people investing in this tech right now and it puts pressure on companies like Unity to spend time on these kinds of things. It might pay off down the road someday but anyone that knows anything about gamedev could tell that a functioning and powerful game engine is still very much a vital tool even if chatbots were doing all the dev work. The existence of ‘make-me-a-game-bots’ still wouldn’t preclude the usefulness of a well-made engine. I’m not sure how to explain that to people that just throw money at things to make line go up, though.
The shoe company Allbirds, once valued at $4 billion, has lost 99% of its value and is now worth around $39 million. But wait, they just announced that they’re pivoting to AI Instead of being a shoe company, they now want to become an AI company that offers its own data centers.
Somehow, the people who turned a $4 billion company into a $39 million company announced this, and investors thought it was a good idea. They managed to raise $50 million for the pivot, which is more than the company itself is currently worth, and gave them to the same people who lost nearly $4 billion.
If that’s not a bubble, and investors genuinely believe they’ll make money from this, then I don’t know what is. Some people hear “AI” and immediately start throwing money at whoever says it.
I grabbed the initial version from AI, and then modified it. Sometimes adding features by just updating one function. There’s pretty much no script left unchanged, but the 80% completion barrier that would have taken me an extra hour or two for every instance is just gone.
Burned through 1,200 credits trying to get unity AI to help with lightmapping issues in my scene. All fixes didn’t work, then I ran out of credits. Not impressed, I will not continue to pay for this. Waste of time.
I burned through the 1000 credits in a couple of hours. I’m not as interested in image and model generation and tested it mainly for C# scripts. With that pricing model I’m not going to be able to use it. I ran out of credits and switched over to Grok to finish things off. Getting better results with Unity AI is likely, but you can get SuperGrok and spend less for 50x the amount of use. It’s just not practical as far as I can see.
Unity Credits reset automatically at the start of each billing cycle, and any unused credits don’t roll over.
Instead of being a shoe company, they now want to become an AI company that offers its own data centers.
