It is very frustrating that Muse Behavior is basically just a AI (non-ML) behaviour designer like a behaviour tree or GOAP. Which Unity has been lacking and asked for, for YEARS. With just minimal ML integration. But it is part of Muse and locked behind the paywall for no reason. Maybe a bit of an aside from the main topic of AI. But it really bugs me how all of the editors are inconsistent with the styling of the rest of the Unity editor. Going so far as to have custom styling for sliders and toggles. It just confuses me why they would do that.
Since Unity is a publicly traded company, it only makes sense to charge customers for using their integrated services. They have to offer integrated services such as AI to remain somewhat competitive in the game dev market. Personally, since Brackeyes is coming back to do Godot, I'm going to learn Godot with every Brackeyes' Godot tutorials. Remember back when Blender just started out against paid versions of modeling programs? Yeah Blender wasn't that good starting out, but with every new update/version, it became a "go-to" product for models. It remained free. I'm thinking that Godot will be same, especially since it's FOSS.
Godot has to get a better UI to take off; Blender didn't start to be a serious competitive concern in the professional industry until it had an interface revamp to make it more like other software.
Sadly I agree, I've stopped developing in unity for now because I'm watching where the dice land. After years learning unity its a massive disappointment for a wannabe game developer. I really don't want to learn a new engine, unity was just making sense😂. But we shall see, I'll watch brackeys as well to find out. Just saying I'm not exactly happy about it
@--NLS-- I'm fluent in c#, for me it's time. I still haven't achieved my life long dream of being a game creator, and thus only get a couple hours in a day to do what I enjoy if I'm lucky.
It's interesting to see Unity integrate these into the engine in some way, but in each case there's a better option out there that takes just a little more understanding to use (and is most likely cheaper/free).
I think this is a very good summary of it. It's a bunch of tools available elsewhere, often cheaper or free. Unity will have to win on Unity integration, ease of use and value for this to succeed.
For us Newbie Unity users and coders who don't know what these other options are, could you list a few examples? Because for a newbie, these integrated AI systems look awesome!!! But after reading your comments, it makes me wonder if my excitement is misplaced.
@@cwebb7178 Nah... Chat Bot? You can run locally and it won't help you like most people think it will.. it has limitations. Image Generator? For textures is cool but for Sprites it sucks cuz u can't control it and can also be used locally, Animations? They just threw a garbage model and put a price on it.. also can be ran locally. AI is a tool that should be manipulated by the end user not a random company... is what it makes useful. These features on the video is just Unity making low effort stuff to increase the price.
Animate, specifically, is an in-house tool that was covered in Unite '22 (Using Machine Learning & AI to create natural character poses). IIRC from the talk, the advantages were that it produced animations usable for real-time (no mocap jank or 1000-keyframe nonsense) and that it was built for use in Unity, but could also easily export for editing in a modeling/animation program like Maya/Blender. I'm a bit disappointed that it's getting bundled with the genAI stuff, since I have no interest in a $30 sub for that (already have access to texture libraries and the API docs, and Unity already has tree-graph visualizers).
Heya! Lovely video. I'm the lead developer on Muse Behavior and i wanted to point out a small correction: It is in fact free! Only the AI generation stuff is behind a paywall :)
Oh that is intereating to hear. I would highly suggest making that prominently known in future materials, as I do not Beleive it was mentioned anywhere in the recent blog.
Great video! I agree with your review. The muse texture, chat, animates, and behaviour are impressive. Great potential! I look forward to using them one day.
These AI tools seem very lackluster compared to alternatives, I’m honestly glad epic isn’t trying to shoehorn AI into unreal and is spending their efforts on actual features and tools devs can use in a commercial product.
@@walterpalladino1965 Nah. None of this is the least-bit usable in a game. Even if the demo was rock-solid, anyone can already buy a bunch of beautiful, handcrafted assets, hire a team of experienced staff, and still make a shitty game. You can't teach what you don't know.
Thanks for the overview of the Muse tools. I've been curious but haven't given them any attention as yet. The sprite and texture generators pale terribly compared to what I can already do myself with ComfyUI / Stable Diffusion. The animation AI is one I'll be keeping my eyes on as I can see it improving massively in the next year or so. I was also curious about the behavior tools, but they look like a scaled back copy of Behavior Designer, and I've already got that and several other similar packages. Looks like I can save my money, at least for now, haha! 😅
Heads up! For $30 /month you get 3000 credits. It's not an unlimited service. Chat prompt = 25credits (max 120 / month) Sprite generation = 4/image (max 750 / month) So in my opinion this is a very unattractive offering. Being able to make only 120 prompts in unity chat for $30 (they even charged credits when the prompt failed and gave nothing is ridiculous given the state of the market.
The external Ai tools im already using are like 10 times better and still need tons of manual refinement, only barely speeding up my workflow than making it from scratch. Ai is still mostly disappointing for me
The idea with these AI art tools right now is really for fast concepting. So instead of you having to make a bunch of placeholder art you can use AI to do it. That would be one advantage of having it in engine. In the end if you were going to use AI for the finished game you would be much better off using something external like Stable Diffusion combined with image editors and other art apps that are going to give you far more control and options. A lot of people have the idea that AI is some kind of easy mode button but it's really not, if you want something good that is based on a concept and not just close enough random art you actually have to put in some time and effort. AI right now isn't really going to benefit anyone with no art skills outside of just placeholder art. It can get you 75% there very quickly but it still needs time and skill to turn it into something finished that doesn't look like low effort AI art. At the moment most of these tools are not worth the price especially when there's free and better options. It's impressive they have training in there though.
I think too, if you're someone in charge of putting art into a project, placeholder or otherwise, and you don't know how to use an image editor, you're at a huge disadvantage too, so for beginners it's also a bad practices trap that we really ought to advise against in case people end up missing important skills.
i will do you one better. is UNITY any good. Edit: in all seriousness brackeys being back just really solidified me never having to consider unity as even being an option anymore.
Yeah. For me Godot for messing around and prototyping games and 2d projects and Unreal for more demanding 3d stuff as Godot is not the best with 3d but slowly improving.
Yep it's quite good it has quite a lot of very good games made with it, it's easier than Unreal yet more flexible, and it's way more capable than Godot plus it's viable for production because it won't introduce breaking changes every now and then. The reason why a lot of people, including Brackeys, are switching, is ideological not practitcal, this kind of reasons can seduce and satisfy very simple surface tutorial makers like Brackeys and very unexperienced unambitious hobbyist, but it won't satisfy the need of an actual team (of any size) trying to actually make an actual game and a living out of it.
@@pacco7641 idk mank I have tried all engines so far and godot just feels better overall, why is suddenly every single thing on the itnernet about "ideology"
I switched to Godot before the whole unity runtime fee and glad that I did. I mean 2D is as easy as it gets used to use raycast from a camera get the intersection with a plane just to get the mouse position in a 2D game and convert the vector math into 2d. in godot it is just a one line of code. Ai tools won't change my mind
Me personally, as a dev I don't care how long it takes for development, how long financial problems have to last during development, AI ruins it for me.
I mean all of this seems like best use for prototyping. Unfortunately I know a lot of people who think they can just make a sellable game with just these tools and wonder why no one buys it. But still, won't get me to get the newer build of unity.
It would be great if your viewers don't need to use a magnifying lense to actually see your unity interface. please make it larger when you record lol!
If you have an RTX it doesn't make sense to pay for an AI service or an AI Tool that is paying for an AI service.. you can simply run locally and if you get the knowledge, fine tune it based on your work and needs.
People should check out the history of shareholders & board members before further enriching gen AI ghouls. The EA/Unity guy did not step out of a vacuum. We get ourselves into theses messes & and make *them* so rich that we can't get out. PS #StopKillingGames
Oh trust me, I have VERY much covered Unity's missteps and the trials and tribulations of them going public and their ex-CEO on this channel. Probably moreso than anywhere else on the internet.
@@gamefromscratch yes Mike you have indeed. No need to trust you, Ive watched them. I was commenting more among we gaming & dev community. Its time we stopped sowing the seeds of our own destruction. I mean I bought games in the early 80s and companies taking the money and running isn't new. Utterly scraping/scamming/stealing and spying are newer along with denying us the ability to play games we bought. I can still load a tape into my ZX Spectrum even if the publisher became a global mega corp by the 1990s. But all this DRM, surveillance, online only, subscription, gen AI thieving and 'disruption' of society whether laying people off or burning fossil fuels and draining reservoirs to run server farms has got to stop yesterday. See you got me started! 😁
@@VertexRage From the AI Faq linked: "For producing sprites, textures, animations, and AI-enabled behaviors, we custom-built our own model trained entirely on data, images, and animations that Unity owns or has licensed." I feel it's safe to assume that this means commercial use is acceptable. There is a point about DMCA removal for your own content within the model-- I don't like this, it worries me. This is not the blanket "This will not happen" statement I wanted to see.
@@anchorlightforge I wouldn't assume anything atm. Too many lies about data sources in the tech industry recently. Especially with this DMCA addition (they wouldn't add that if they were 100% sure that their data is 100% clean, would they?)
@@VertexRage A part of me says "owned and licensed by Unity" is enough to sufficiently pass the buck. If there is a legal issue, you weren't aware of it. The reality is I quit Unity in September and YMMV. 😅
I do not undestand why Behaviour Graph tool is part of Muse, why it can't be standalone with Muse extensions. It could be nice for users who would like to code actions manualy.
So its a game dev tool that tells you what you may or may not put in your game. Do I have to "jail break" my own software to have it make what I want? No thanks.
@@rodvik It's a safeguard because there could very well be illegal content generated through the generator. What concerns me more is why said results are happening-- what's going into the dataset to trigger such results? If this is using the same dataset OpenAI and all the ethical-- understatement of the century-- PROBLEMS that entails, it's gg for this whole operation.
@@BlueSpawn I don't use that word lightly. If you're shocked by the idea that I would use that word in that context, you'll be even more shocked by some of the horrendous content OpenAI is responsible for including in their model. They may be generated images but those patterns are developed through source images that don't belong in a public dataset, or anywhere for that matter. Whatever the usual game censorship arguments are, I assure you this is a completely different and valid fear.
Unity , I think, is proud of how active their asset store market is. I wonder how well AI scripting and logic could be incorporated into projects using various 3rd party extension assets...
They should've made Muse free for Pro users, and let indies use it for free (or at least give some free tokens to prevent people from spamming). By doing so, Unity can win some of the trust they lost from their users.
I've been doing A LOT of work with AI the last few years. Your biggest issue is the prompts. You will NEVER get good results with a prompt like "knight in shining armor" there's just not enough information to go off of and depending on the model and samplers it can be wildly different and inconsistent. Something better would be like... 8k, masterpiece, raw photo of an adult male knight without helmet, winter landscape with mountains in background, standing by a lake, short brown hair, medium length beard, muted colors, dynamic lighting, trending on ArtStation, trending on CGSociety, Intricate, High Detail, Sharp focus, dramatic, photorealistic painting art by midjourney and greg rutkowski, VSCO, film grain, vignette, volumetric lighting Stuff like that. Where you give a little bit more information on not only the subject, the scene, but also what it might need to look like (The overall picture, lighting, whether it's a painting, photorealistic, concept art, etc). Also Negative prompts imo are MORE important than the normal prompt. You can have a decent prompt, but without a negative prompt it can still look like some cursed horror.
1- If you have to provide detailed prompts, with clear in/out-group signifiers & delimiters, then at that point it's just [inefficient] programming. 2- You also need consistency for both 2d sprites and 3d models/textures/animation. *That's* the key element that requires specialized training (of humans or computers), and that's the resource- & time-consuming part that everyone seems to want to skip, but still reap great results. Tech has chased this dragon before (go read about the rise & fall of COBOL).
Sorry, but this is ridiculous. Imagine having to use a search engine like google to search for a specific imagine like that... This isn't going to catch anyone, especially nowdays
Personally, I think Unity's payment model for this feature-set is a mistake. Here's what they should do instead-- - Limit Sprite, Texture, Animation and Chat Generations for Free users to XXX Credits daily (the credits are shared across the different types of Generations) - Unity Pro users get unlimited Generations - Developers can alternatively pay $9.99 per month if they don't have Unity Pro and want unlimited Generations. - Muse Behavior without the Generation is completely free. This way, everyone can try the tools, and people using them seriously + Pro users, will get more out of it. Also, Muse Behavior without Generation would be a nice means of getting people to taste Muse and consider trying everything out while also providing a very useful tool for developers that don't want to use UVS or a paid solution for Behavior Trees / FSMs
I tend to agree completely. I actually told them something very close to this, especially with the need they have to make Pro more appealing due to the added run-time fee. This as a value add (and speedtree damnit!) would make a lot of sense. Honestly I think the biggest issue is, they have to pay out X to make this work, fees to OpenAI, whoever they license their datasets from, etc. I think they are somewhat pinched in how they can monetize Muse. That said, that's a them problem not an US problem, and the current pricing isn't overly appealing as it stands.
@@somedude5951 You're not wrong, but having a suit of AI tools specialized for rapid prototyping/development and integrated into the Unity Engine is not bad in the long-term. These tools will get better over time as Unity invests more into their datasets and weights. But for now, I do agree that-- For Image gen, SD + LoRA + ControlN_t is ideal For Textures, there's a specialized Diffusion model for them. When it comes to 3D directly, you can use M_shy or L_ma. For Animation, there's Mooti_n and C_scadeur For Codebase there's Codeium and Copilot X, but Codeium is free For Behavior AI, I guess that is technically codebase... but the context being so tightly integrated into Unity makes it fairly specialized. But to use all of these tools, you have to do multiple signups and look at a lot of examples and get used to using them on their respective platforms to get ideal results. So there's a tradeoff, but I do agree that you get more value paying into other models that can perform better than paying into Unity's current iteration of Muse.
Wow, how useless the Sprite one is, glad there is a trail so you can see for yourself. The Texture one that has the most use. The chat is barely useful. That State machine also looks useful. Not $30/month useful though.
what's going on with those AIgens, they feel extremely bad even for AI gens' low standards. it's so far behind the other tools in the same package, almost feels like they used an old SD model just to make the package seem more "bang for the buck". I know they won't, but individual tool packages would be so good.
This engine is so done. They just can’t seem to figure out how to move forward. I still remember how tedious it use to be to work with scriptable render pipelines and mechanim avatar system for animations and character. Glad I made the decision to switch to unreal. In a couple months we r getting a complete motion matching parkour sample project with 600+ AAA animations. Meanwhile Unity is wasting resources in this useless crap.
Imagine a person new to photography that just clicks the button to make an image but doesn't know about ISO, F-stop settings. That is why images look bad with AI. Also a good argument that anyone using photos for "art" is not an artist. But we sorted that issue out a long time ago so AI isn't that disruptive to begin with.
stable diffusion produces way much better results. by the looks of it seems like stable diffusion under the hood. if people start saying violating copyright. its so simple to create your own image generative models on your created art. for free!!!
Yeah. You can get far better results with a custom SD 1.5 model. And if you have a decent GPU, it's probably faster than the cloud servers unity appears to be using
I was in the Unity AI Beta and Unity was much more interested in how much I would be willing to pay versus how can they make it something I want to pay for. Very lost company.
IMO their base model are not refined enough in their datasets, it feels like they just do the random crop and put it as datasets and then just train them, as we can see there a lot of "Blank" generation. Sprite style builder feels like LoRa training, they should specify how many reference examples you should put in to get a better results and cropped as well. Their model still needs Quality embellishment like High-quality or Cinematic keywords and negative keywords, they need to put some default prompt so the User notices it more, since most likely the Base Model are from SD1.5 or SDXL that they train in-house. If they accept your own trained model or we can custom train our own model to use, it might be worthed. For now, it's Kinda Scuffed $30 subscription ngnl.
I wouldn't call this Unity, any more than Disney is Disney, or Blizzard is Blizzard. These companies are in name only and staffed by greedy suits, and bootlicking backstabber moralist grifters. What a fall from grace in such a short period of time.
My final opinions I suppose would be: Muse Sprite = AI Jank, pretty useless Muse Chat = CoPilot for Unity, works well, but CoPilot does it already so if you've got a coding assistant you like it's less useful Muse Animate = Surprisingly capable. All AI animation tools (excluding the likes of Cascadeur) have been pretty awful. This one works shockingly well, has potential Muse Behavior = Very early, not really all that AI related to be honest, it's a state machine controller with a visual scripting language, but it has potential if realized Muse Texture = You can do all of this in other programs, but it does integrate into Unity nicely There is some crap here, there is some potential here and the Unity integration is nice. Is it worth $30 for a dedicated Unity developer... that's very questionable, at least for now. It's probably too early for them to be selling it to be honest, but potential is there.
Edit to add: Initial comment about burnout is NOT directed at Mike, just a general comment :D I'm honestly completely burned out on the AI hype and view it as the new crypto, except that I can actually see it having a place as assistive technology going forward. In that vein, I'll admit texture generation genuinely impressed me. I love making new textures, but I _hate_ the tedious stuff after actual art asset creation, so I'm biased there. Code AI creates more work for me as an intermediate programmer than it saves, but for advanced programmers who know what they're doing, I can absolutely see it being a huge time saver. So there's definitely _potential_ here. Just nothing I'd pay $3 for, let alone $30, even if I wasn't pretty much dead set against giving Unity any money until they prove themselves to have really made meaningful change
As an experienced programmer in normal/enterprise & with Unity, it's only helpful maybe 10-20% of the time. Two big hurdles are that it doesn't know what you want to do, and that it makes stuff up. I was looking for a feature in the Timeline API - Google's AI search popped a "result" that seemed to fit. Wasted a good 5-7min trying to find the method that it hallucinated. OTOH, I was implementing a set of custom copy/paste methods, with several blocks of near-repeat code, and Visual Studio's "enhanced" autocomplete correctly anticipated/early-completed about 4 out of 5 lines (80%). Unfortunately the other times it would sometimes introduce almost-not-quite bugs, so I still had to slow down to carefully check each line, even the autocompleted ones. So it's handy, sometimes, but not ready for primetime.
Also, I agree that the hype is akin to crypto (same VCs/founders in many cases!). There's maybe a use-case for some of this stuff, or more accurately for LLM approaches. But it's not the "Easy Button" or "expensive-staff replacement tool" that it's sold as, and there are a lot of downsides, both technical & ethical.
It's just sort of a poisoned well sometimes. You have way too many sketchy companies chasing the money, namely OpenAI, but in the industry it will forever have its place because ML can destroy repetitive tasks with perceptibly perfect accuracy.
You're totally right about it being the next crypto. The big giveaway is if people had been following GPT and machine learning development, this stuff isn't actually new and there wasn't a huge boom in the technology when chatGPT came out. It was a long and hard road of research with slow incremental progress for both the chatbots and generative images. The difference that has led to the hype is OpenAI's public paid offering and subsequent bandwagoning by a bunch of small startups piggybacking on the technology being made more accessible. The explosion of "new tech" is just people using GPT to do other stuff poorly. None of the code help is going to push anything forward because it isn't able to innovate at all. It does things the way we already know how to do them and not quite as well. The quality of results is gonna keep progressing at the same slow pace as before. Now it's becoming obvious that the chatbots and even images are just regurgitating the top results from google and bing. None of it is as useful as they're claiming and it's also ridiculously expensive to maintain and let people make all these requests. There's no way it'll last the way things are right now. Now that OpenAI got their goal of being bought by microsoft I feel like the hype is gonna start to die down since none of it is profitable at all.
The only tool I can see being good right now is the muse chat (though I haven't tried it), the rest of the tools look like they'd cost you more time than you'd save using them. I'd like to see a VS Code plugin for muse chat, I think they could do great things with an autocomplete ai trained on unity docs.
You know the Visual Studio for Unity plugin already does Unity API autocomplete, right? Also picks up references in code for built-in engine callbacks like Start/Update/OnXYZ.
@@FactsNoCare No problem :) It does a lot of other cool stuff, like debugging & profiling in editor & on-device (via local network). There's probably a video somewhere on either Unity's or Microsoft's channel...
Yeah. AI coding C++, replacing blueprints 😋 It may still take some years, for AI to produce bug free code. Although should not be to hard, when implementing a virtual compiler for testing in the AI model efforts in coding.
When this was showcased before, it seemed trailblazing, but now it seems very underwhelming. The texture generation is useful, but I think it's a bit limiting done in Unity, since if you wanna do anything more on it, you have to export it and modify it somewhere else. If I would have to pick, I would probably want this for a texture artist working in substance designer and painter, not in unity
Agreed, except that I doubt Muse is for anyone actually skilled in sprites/texturing/code/animation, or interested in becoming so. The selling-pitch seems aimed at people who actually *can't* do those things (or don't want to learn). So sprites/texturing/animations for "I suck at art" people, code for non-coders, and I guess chat for ppl who won't RTFM [read the f'in manual] 😅 If you are the least-bit interested in learning any of those skills, you'd want to use the appropriate tools, so your skills are usable outside Unity.
These images are just bad. The concept is pretty cool but maybe use Stability Diffusion or Stable Cascade. They can both be installed locally and are free and open source with great images. Stable Cascade is the fastest image gen I have seen, matter or seconds per image.
Compare this to Unreal experimental features... that actually work? I dunno, Unity pretty much ended it own relevance and gave its life for the rise of Godot and Unreal
WOW! these AI tools look AMAZING! they seem super useful especially Animate, I'm a programmer so animation/art is REALLY not my thing and I would love to reduce my need on third parties for art, it's a shame that these don't have a free tier
I mean its cool they try to incorporate AI tooling in. But to be honest? This is not production ready, this should not be paid at all. Its wonky, it looks like a beta version of a tool that is still far away from a commercial product.
This is stable diffusion with just chat GPT nothing special, a lot of stuff has stable diffusion and goes to the generation, you train that with steps, like first and better and better, and yes for that needs a long time, just that is some cut out version, people better use stable diffusion don wase you money for this...
MUSE not Moose, I.e. a source of inspiration. Type the word into Google Translate and listen to how it’s pronounced. Do you call your muscles, your Moosealls?
Having used lots of different versions of ai generators, muse is trash. It does nothing(!) well and makes NO sense whatsoever. There are A LOT of possibilities to improve the engine and how the creation process works. But start with the basics. People who use Unity are already somewhat comfortable with a higher degree of complexity. Throw ALL that "one click solution" crap into the trash. Because it NEVER works. They better get their sh*t together. This is not doing that.
It's pronounced MUZE ... one S = Z sound while two S's == S sound/. :p Is this your new "foliage"? muse /myo͞oz/ noun: Muse; plural noun: Muses; noun: muse; plural noun: muses 1. (in Greek and Roman mythology) each of nine goddesses, the daughters of Zeus and Mnemosyne, who preside over the arts and sciences. 2. a person or personified force who is the source of inspiration for a creative artist. "Yeats' muse, Maud Gonne"
meanwhile everyone is fear mongering about horrible stable diffusion real artists are speeding up their workflow with ai. Interpolation also uses a form of artificial intelligence and it has revolutionized digital animation
Link
gamefromscratch.com/unity-ai-muse-tools-review/
If you caught this video earlier, you aren't going mad! ;) This is a reupload, TH-cam froze after 14 minutes with the first version, hopefully this one fixes it!
-----------------------------------------------------------------------------------------------------------
*Support* : www.patreon.com/gamefromscratch
*GameDev News* : gamefromscratch.com
*GameDev Tutorials* : devga.me
*Discord* : discord.com/invite/R7tUVbD
*Twitter* : twitter.com/gamefromscratch
-----------------------------------------------------------------------------------------------------------
Timeline
--------------
0:00 Unity AI Tool Introduction
1:00 Muse Tools Overview
1:57 Installing the Muse Tools
3:01 Muse Sprite Hands-On
5:48 Muse Sprite Style Builder
13:30 Muse Texture Hands-On
17:30 Muse Chat Hands-On
24:25 Muse Animate Hands-On
32:15 Muse Behavior Hands-On
41:29 Conclusions and Opinion
Well I don't know which produces better results frozen YT video or Unity AI, both are unusable in the end. :)
When godot and unreal engine add AI tool
It is very frustrating that Muse Behavior is basically just a AI (non-ML) behaviour designer like a behaviour tree or GOAP. Which Unity has been lacking and asked for, for YEARS. With just minimal ML integration. But it is part of Muse and locked behind the paywall for no reason.
Maybe a bit of an aside from the main topic of AI. But it really bugs me how all of the editors are inconsistent with the styling of the rest of the Unity editor. Going so far as to have custom styling for sliders and toggles. It just confuses me why they would do that.
Hi! I'm the lead developer on Muse Behavior. It is in fact free! Only the AI generation stuff is behind a paywall :)
Since Unity is a publicly traded company, it only makes sense to charge customers for using their integrated services. They have to offer integrated services such as AI to remain somewhat competitive in the game dev market. Personally, since Brackeyes is coming back to do Godot, I'm going to learn Godot with every Brackeyes' Godot tutorials. Remember back when Blender just started out against paid versions of modeling programs? Yeah Blender wasn't that good starting out, but with every new update/version, it became a "go-to" product for models. It remained free. I'm thinking that Godot will be same, especially since it's FOSS.
Godot has to get a better UI to take off; Blender didn't start to be a serious competitive concern in the professional industry until it had an interface revamp to make it more like other software.
Sadly I agree, I've stopped developing in unity for now because I'm watching where the dice land. After years learning unity its a massive disappointment for a wannabe game developer. I really don't want to learn a new engine, unity was just making sense😂. But we shall see, I'll watch brackeys as well to find out. Just saying I'm not exactly happy about it
@--NLS-- I'm fluent in c#, for me it's time. I still haven't achieved my life long dream of being a game creator, and thus only get a couple hours in a day to do what I enjoy if I'm lucky.
@@thetabestYou can use C# in Godot. I'm wanting to take a swing at C++ Godot.
Damn that was a fast re-upload, appreciate you
Thanks. I don't know what happened there. Seems to have been a TH-cam encoding glitch. Sorry for the hiccup
It's interesting to see Unity integrate these into the engine in some way, but in each case there's a better option out there that takes just a little more understanding to use (and is most likely cheaper/free).
I think this is a very good summary of it. It's a bunch of tools available elsewhere, often cheaper or free. Unity will have to win on Unity integration, ease of use and value for this to succeed.
yep, I was hyped but after seeing it I'm not impressed
For us Newbie Unity users and coders who don't know what these other options are, could you list a few examples? Because for a newbie, these integrated AI systems look awesome!!! But after reading your comments, it makes me wonder if my excitement is misplaced.
@@cwebb7178 Nah... Chat Bot? You can run locally and it won't help you like most people think it will.. it has limitations.
Image Generator? For textures is cool but for Sprites it sucks cuz u can't control it and can also be used locally, Animations? They just threw a garbage model and put a price on it.. also can be ran locally.
AI is a tool that should be manipulated by the end user not a random company... is what it makes useful. These features on the video is just Unity making low effort stuff to increase the price.
Animate, specifically, is an in-house tool that was covered in Unite '22 (Using Machine Learning & AI to create natural character poses). IIRC from the talk, the advantages were that it produced animations usable for real-time (no mocap jank or 1000-keyframe nonsense) and that it was built for use in Unity, but could also easily export for editing in a modeling/animation program like Maya/Blender.
I'm a bit disappointed that it's getting bundled with the genAI stuff, since I have no interest in a $30 sub for that (already have access to texture libraries and the API docs, and Unity already has tree-graph visualizers).
Heya! Lovely video. I'm the lead developer on Muse Behavior and i wanted to point out a small correction: It is in fact free! Only the AI generation stuff is behind a paywall :)
Oh that is intereating to hear. I would highly suggest making that prominently known in future materials, as I do not Beleive it was mentioned anywhere in the recent blog.
I've asked the marketing person to help clear up the confusion, I'm hopeful 😅
Great video! I agree with your review. The muse texture, chat, animates, and behaviour are impressive. Great potential! I look forward to using them one day.
These AI tools seem very lackluster compared to alternatives, I’m honestly glad epic isn’t trying to shoehorn AI into unreal and is spending their efforts on actual features and tools devs can use in a commercial product.
13:13, you said this is typical AI art? Have you used DALLE 3 or Midjourney? These muse sprite generations look like version 1 AI models from mid 2022
Yeah, this is really low quality. I'm disappointed.
I think it is a MUST to know if Unity is using the users of this services to training AIs which will later be used to build and self other products.
Every AI service is retraining on their [paid] users, you only have to read the EULA to see that. Another reason to avoid them IMO
@@mandisaw In this particular case I think it is worst because you are teaching the AI how to make games!!!
@@walterpalladino1965 Nah. None of this is the least-bit usable in a game. Even if the demo was rock-solid, anyone can already buy a bunch of beautiful, handcrafted assets, hire a team of experienced staff, and still make a shitty game. You can't teach what you don't know.
Hey Mike, a centaur has human torso and head, so it wasn't even close 😂
Thanks for the overview of the Muse tools. I've been curious but haven't given them any attention as yet. The sprite and texture generators pale terribly compared to what I can already do myself with ComfyUI / Stable Diffusion. The animation AI is one I'll be keeping my eyes on as I can see it improving massively in the next year or so. I was also curious about the behavior tools, but they look like a scaled back copy of Behavior Designer, and I've already got that and several other similar packages. Looks like I can save my money, at least for now, haha! 😅
This is some half-baked AI. You can do much better than this for free with Stable Diffusion.
Yeah, that was my thought too
Heads up!
For $30 /month you get 3000 credits. It's not an unlimited service.
Chat prompt = 25credits (max 120 / month)
Sprite generation = 4/image (max 750 / month)
So in my opinion this is a very unattractive offering. Being able to make only 120 prompts in unity chat for $30 (they even charged credits when the prompt failed and gave nothing is ridiculous given the state of the market.
The external Ai tools im already using are like 10 times better and still need tons of manual refinement, only barely speeding up my workflow than making it from scratch. Ai is still mostly disappointing for me
Great overview video! Thank you for going over everything.
Muse Behavior should be free and built-in like uneal engine behavior tree
Hi! I'm the lead developer on Muse Behavior. It is in fact free! Only the AI generation stuff is behind a paywall :)
@@user-gk1ty4qk3f if that correct it will be super awesome 💖
@@user-gk1ty4qk3f and what is the use without the generation?
@@user-gk1ty4qk3f also i checked and it is just free for 15 days, its not free
The idea with these AI art tools right now is really for fast concepting. So instead of you having to make a bunch of placeholder art you can use AI to do it. That would be one advantage of having it in engine. In the end if you were going to use AI for the finished game you would be much better off using something external like Stable Diffusion combined with image editors and other art apps that are going to give you far more control and options.
A lot of people have the idea that AI is some kind of easy mode button but it's really not, if you want something good that is based on a concept and not just close enough random art you actually have to put in some time and effort. AI right now isn't really going to benefit anyone with no art skills outside of just placeholder art. It can get you 75% there very quickly but it still needs time and skill to turn it into something finished that doesn't look like low effort AI art.
At the moment most of these tools are not worth the price especially when there's free and better options. It's impressive they have training in there though.
I think too, if you're someone in charge of putting art into a project, placeholder or otherwise, and you don't know how to use an image editor, you're at a huge disadvantage too, so for beginners it's also a bad practices trap that we really ought to advise against in case people end up missing important skills.
i will do you one better. is UNITY any good. Edit: in all seriousness brackeys being back just really solidified me never having to consider unity as even being an option anymore.
Yeah. For me Godot for messing around and prototyping games and 2d projects and Unreal for more demanding 3d stuff as Godot is not the best with 3d but slowly improving.
Yep it's quite good it has quite a lot of very good games made with it, it's easier than Unreal yet more flexible, and it's way more capable than Godot plus it's viable for production because it won't introduce breaking changes every now and then. The reason why a lot of people, including Brackeys, are switching, is ideological not practitcal, this kind of reasons can seduce and satisfy very simple surface tutorial makers like Brackeys and very unexperienced unambitious hobbyist, but it won't satisfy the need of an actual team (of any size) trying to actually make an actual game and a living out of it.
godot!!! godot my beloved
@@pacco7641 idk mank I have tried all engines so far and godot just feels better overall, why is suddenly every single thing on the itnernet about "ideology"
@@CarTastic-fv6eo I'm the same, using UE for now until the 3D in Godot is ready (hopefully)
I switched to Godot before the whole unity runtime fee and glad that I did. I mean 2D is as easy as it gets used to use raycast from a camera get the intersection with a plane just to get the mouse position in a 2D game and convert the vector math into 2d. in godot it is just a one line of code. Ai tools won't change my mind
Me personally, as a dev I don't care how long it takes for development, how long financial problems have to last during development, AI ruins it for me.
I mean all of this seems like best use for prototyping. Unfortunately I know a lot of people who think they can just make a sellable game with just these tools and wonder why no one buys it. But still, won't get me to get the newer build of unity.
It would be great if your viewers don't need to use a magnifying lense to actually see your unity interface. please make it larger when you record lol!
If you have an RTX it doesn't make sense to pay for an AI service or an AI Tool that is paying for an AI service.. you can simply run locally and if you get the knowledge, fine tune it based on your work and needs.
does chat have a way to have it learn your project so you can ask it something specific about your project?
People should check out the history of shareholders & board members before further enriching gen AI ghouls. The EA/Unity guy did not step out of a vacuum. We get ourselves into theses messes & and make *them* so rich that we can't get out. PS #StopKillingGames
Oh trust me, I have VERY much covered Unity's missteps and the trials and tribulations of them going public and their ex-CEO on this channel. Probably moreso than anywhere else on the internet.
@@gamefromscratch yes Mike you have indeed. No need to trust you, Ive watched them. I was commenting more among we gaming & dev community. Its time we stopped sowing the seeds of our own destruction. I mean I bought games in the early 80s and companies taking the money and running isn't new. Utterly scraping/scamming/stealing and spying are newer along with denying us the ability to play games we bought. I can still load a tape into my ZX Spectrum even if the publisher became a global mega corp by the 1990s. But all this DRM, surveillance, online only, subscription, gen AI thieving and 'disruption' of society whether laying people off or burning fossil fuels and draining reservoirs to run server farms has got to stop yesterday. See you got me started! 😁
Don't buy subscription based games and don't support greedy companies. Simple as.
@@therealpeter2267 it is simple yet it seems we still need to spread the word.
I will use muse chat for now. I currently use chatgpt and Gemini for coding assistance. I look forward to muse behavior and muse animate
Ooh Steam is gonna love this.
Key questions:
1) What's license on Muse output?
2) What is the training data?
I think most of that is covered in the FAQ, but the direct source of data they licensed isn't fully clear.
unity.com/ai/faq
@@gamefromscratch I read that page 5 times already, and no - there is no answer to that questions there :(
@@VertexRage From the AI Faq linked:
"For producing sprites, textures, animations, and AI-enabled behaviors, we custom-built our own model trained entirely on data, images, and animations that Unity owns or has licensed." I feel it's safe to assume that this means commercial use is acceptable. There is a point about DMCA removal for your own content within the model-- I don't like this, it worries me. This is not the blanket "This will not happen" statement I wanted to see.
@@anchorlightforge I wouldn't assume anything atm. Too many lies about data sources in the tech industry recently. Especially with this DMCA addition (they wouldn't add that if they were 100% sure that their data is 100% clean, would they?)
@@VertexRage A part of me says "owned and licensed by Unity" is enough to sufficiently pass the buck. If there is a legal issue, you weren't aware of it.
The reality is I quit Unity in September and YMMV. 😅
I do not undestand why Behaviour Graph tool is part of Muse, why it can't be standalone with Muse extensions. It could be nice for users who would like to code actions manualy.
Some Ministry of Silly Walks vibe at the 26:00 minute mark. th-cam.com/video/iV2ViNJFZC8/w-d-xo.html
💯 Well they trained on memes & gifs, so makes sense
Then blank ones could be a automatic inappropriate censorship filter. AI can generate that kind of stuff even if you tell it not to.
That could be. With Muse Sprite I actually got "inappropriate results" when that happened, and it actually happened quite a bit.
So its a game dev tool that tells you what you may or may not put in your game. Do I have to "jail break" my own software to have it make what I want? No thanks.
@@rodvik It's a safeguard because there could very well be illegal content generated through the generator. What concerns me more is why said results are happening-- what's going into the dataset to trigger such results?
If this is using the same dataset OpenAI and all the ethical-- understatement of the century-- PROBLEMS that entails, it's gg for this whole operation.
@@anchorlightforge The idea that fake images and sculptures can be illegal is absurd.
@@BlueSpawn I don't use that word lightly. If you're shocked by the idea that I would use that word in that context, you'll be even more shocked by some of the horrendous content OpenAI is responsible for including in their model. They may be generated images but those patterns are developed through source images that don't belong in a public dataset, or anywhere for that matter.
Whatever the usual game censorship arguments are, I assure you this is a completely different and valid fear.
Unity , I think, is proud of how active their asset store market is. I wonder how well AI scripting and logic could be incorporated into projects using various 3rd party extension assets...
damn 40min upload!!
I wanna see a ”Free Trial game Jam”
I'm so glad they developed these AI tools. This is a big relief for me.. no longer having to cry about Unity when I've gotten used to Unreal.
They should've made Muse free for Pro users, and let indies use it for free (or at least give some free tokens to prevent people from spamming).
By doing so, Unity can win some of the trust they lost from their users.
I've been doing A LOT of work with AI the last few years. Your biggest issue is the prompts. You will NEVER get good results with a prompt like "knight in shining armor" there's just not enough information to go off of and depending on the model and samplers it can be wildly different and inconsistent. Something better would be like...
8k, masterpiece, raw photo of an adult male knight without helmet, winter landscape with mountains in background, standing by a lake, short brown hair, medium length beard, muted colors, dynamic lighting, trending on ArtStation, trending on CGSociety, Intricate, High Detail, Sharp focus, dramatic, photorealistic painting art by midjourney and greg rutkowski, VSCO, film grain, vignette, volumetric lighting
Stuff like that. Where you give a little bit more information on not only the subject, the scene, but also what it might need to look like (The overall picture, lighting, whether it's a painting, photorealistic, concept art, etc). Also Negative prompts imo are MORE important than the normal prompt. You can have a decent prompt, but without a negative prompt it can still look like some cursed horror.
1- If you have to provide detailed prompts, with clear in/out-group signifiers & delimiters, then at that point it's just [inefficient] programming. 2- You also need consistency for both 2d sprites and 3d models/textures/animation. *That's* the key element that requires specialized training (of humans or computers), and that's the resource- & time-consuming part that everyone seems to want to skip, but still reap great results. Tech has chased this dragon before (go read about the rise & fall of COBOL).
Sorry, but this is ridiculous. Imagine having to use a search engine like google to search for a specific imagine like that... This isn't going to catch anyone, especially nowdays
No, it makes just a less error styled graphics, not a more accurate design intended.
@@jamesmillerjo ... Lol. sure thing
Personally, I think Unity's payment model for this feature-set is a mistake. Here's what they should do instead--
- Limit Sprite, Texture, Animation and Chat Generations for Free users to XXX Credits daily (the credits are shared across the different types of Generations)
- Unity Pro users get unlimited Generations
- Developers can alternatively pay $9.99 per month if they don't have Unity Pro and want unlimited Generations.
- Muse Behavior without the Generation is completely free.
This way, everyone can try the tools, and people using them seriously + Pro users, will get more out of it.
Also, Muse Behavior without Generation would be a nice means of getting people to taste Muse and consider trying everything out while also providing a very useful tool for developers that don't want to use UVS or a paid solution for Behavior Trees / FSMs
I tend to agree completely. I actually told them something very close to this, especially with the need they have to make Pro more appealing due to the added run-time fee. This as a value add (and speedtree damnit!) would make a lot of sense.
Honestly I think the biggest issue is, they have to pay out X to make this work, fees to OpenAI, whoever they license their datasets from, etc. I think they are somewhat pinched in how they can monetize Muse.
That said, that's a them problem not an US problem, and the current pricing isn't overly appealing as it stands.
More mistake is, any one buying it.
All these AI options can be done for free, and in better quality, on the web, except for Cascadeur.
@@somedude5951 You're not wrong, but having a suit of AI tools specialized for rapid prototyping/development and integrated into the Unity Engine is not bad in the long-term.
These tools will get better over time as Unity invests more into their datasets and weights.
But for now, I do agree that--
For Image gen, SD + LoRA + ControlN_t is ideal
For Textures, there's a specialized Diffusion model for them. When it comes to 3D directly, you can use M_shy or L_ma.
For Animation, there's Mooti_n and C_scadeur
For Codebase there's Codeium and Copilot X, but Codeium is free
For Behavior AI, I guess that is technically codebase... but the context being so tightly integrated into Unity makes it fairly specialized.
But to use all of these tools, you have to do multiple signups and look at a lot of examples and get used to using them on their respective platforms to get ideal results.
So there's a tradeoff, but I do agree that you get more value paying into other models that can perform better than paying into Unity's current iteration of Muse.
Wow, how useless the Sprite one is, glad there is a trail so you can see for yourself. The Texture one that has the most use. The chat is barely useful. That State machine also looks useful. Not $30/month useful though.
what's going on with those AIgens, they feel extremely bad even for AI gens' low standards. it's so far behind the other tools in the same package, almost feels like they used an old SD model just to make the package seem more "bang for the buck". I know they won't, but individual tool packages would be so good.
I can see this developing into a powerful tool, but I think it should be a lot cheaper or even free in its current state.
i can see making something funny with the animation tool
I want to see it try to animate other creatures movements on the human 😆 maybe fly like a bird, flap your wings
This engine is so done. They just can’t seem to figure out how to move forward. I still remember how tedious it use to be to work with scriptable render pipelines and mechanim avatar system for animations and character. Glad I made the decision to switch to unreal. In a couple months we r getting a complete motion matching parkour sample project with 600+ AAA animations. Meanwhile Unity is wasting resources in this useless crap.
Imagine a person new to photography that just clicks the button to make an image but doesn't know about ISO, F-stop settings.
That is why images look bad with AI.
Also a good argument that anyone using photos for "art" is not an artist. But we sorted that issue out a long time ago so AI isn't that disruptive to begin with.
stable diffusion produces way much better results. by the looks of it seems like stable diffusion under the hood. if people start saying violating copyright. its so simple to create your own image generative models on your created art. for free!!!
Yeah. You can get far better results with a custom SD 1.5 model. And if you have a decent GPU, it's probably faster than the cloud servers unity appears to be using
I was in the Unity AI Beta and Unity was much more interested in how much I would be willing to pay versus how can they make it something I want to pay for. Very lost company.
IMO their base model are not refined enough in their datasets, it feels like they just do the random crop and put it as datasets and then just train them, as we can see there a lot of "Blank" generation.
Sprite style builder feels like LoRa training, they should specify how many reference examples you should put in to get a better results and cropped as well.
Their model still needs Quality embellishment like High-quality or Cinematic keywords and negative keywords, they need to put some default prompt so the User notices it more, since most likely the Base Model are from SD1.5 or SDXL that they train in-house. If they accept your own trained model or we can custom train our own model to use, it might be worthed.
For now, it's Kinda Scuffed $30 subscription ngnl.
Unity is distracted, because they can't do their main jobs well anymore.
I wouldn't call this Unity, any more than Disney is Disney, or Blizzard is Blizzard. These companies are in name only and staffed by greedy suits, and bootlicking backstabber moralist grifters. What a fall from grace in such a short period of time.
any tl;dw?
My final opinions I suppose would be:
Muse Sprite = AI Jank, pretty useless
Muse Chat = CoPilot for Unity, works well, but CoPilot does it already so if you've got a coding assistant you like it's less useful
Muse Animate = Surprisingly capable. All AI animation tools (excluding the likes of Cascadeur) have been pretty awful. This one works shockingly well, has potential
Muse Behavior = Very early, not really all that AI related to be honest, it's a state machine controller with a visual scripting language, but it has potential if realized
Muse Texture = You can do all of this in other programs, but it does integrate into Unity nicely
There is some crap here, there is some potential here and the Unity integration is nice. Is it worth $30 for a dedicated Unity developer... that's very questionable, at least for now. It's probably too early for them to be selling it to be honest, but potential is there.
@@gamefromscratch Animate isn't LLM/GPT-based, it was their own project from before '22 - that's probably why it's the only gem in the pile
Edit to add: Initial comment about burnout is NOT directed at Mike, just a general comment :D
I'm honestly completely burned out on the AI hype and view it as the new crypto, except that I can actually see it having a place as assistive technology going forward. In that vein, I'll admit texture generation genuinely impressed me. I love making new textures, but I _hate_ the tedious stuff after actual art asset creation, so I'm biased there. Code AI creates more work for me as an intermediate programmer than it saves, but for advanced programmers who know what they're doing, I can absolutely see it being a huge time saver.
So there's definitely _potential_ here. Just nothing I'd pay $3 for, let alone $30, even if I wasn't pretty much dead set against giving Unity any money until they prove themselves to have really made meaningful change
As an experienced programmer in normal/enterprise & with Unity, it's only helpful maybe 10-20% of the time. Two big hurdles are that it doesn't know what you want to do, and that it makes stuff up. I was looking for a feature in the Timeline API - Google's AI search popped a "result" that seemed to fit. Wasted a good 5-7min trying to find the method that it hallucinated.
OTOH, I was implementing a set of custom copy/paste methods, with several blocks of near-repeat code, and Visual Studio's "enhanced" autocomplete correctly anticipated/early-completed about 4 out of 5 lines (80%). Unfortunately the other times it would sometimes introduce almost-not-quite bugs, so I still had to slow down to carefully check each line, even the autocompleted ones. So it's handy, sometimes, but not ready for primetime.
Also, I agree that the hype is akin to crypto (same VCs/founders in many cases!). There's maybe a use-case for some of this stuff, or more accurately for LLM approaches. But it's not the "Easy Button" or "expensive-staff replacement tool" that it's sold as, and there are a lot of downsides, both technical & ethical.
It's just sort of a poisoned well sometimes. You have way too many sketchy companies chasing the money, namely OpenAI, but in the industry it will forever have its place because ML can destroy repetitive tasks with perceptibly perfect accuracy.
You're totally right about it being the next crypto. The big giveaway is if people had been following GPT and machine learning development, this stuff isn't actually new and there wasn't a huge boom in the technology when chatGPT came out. It was a long and hard road of research with slow incremental progress for both the chatbots and generative images. The difference that has led to the hype is OpenAI's public paid offering and subsequent bandwagoning by a bunch of small startups piggybacking on the technology being made more accessible. The explosion of "new tech" is just people using GPT to do other stuff poorly. None of the code help is going to push anything forward because it isn't able to innovate at all. It does things the way we already know how to do them and not quite as well. The quality of results is gonna keep progressing at the same slow pace as before.
Now it's becoming obvious that the chatbots and even images are just regurgitating the top results from google and bing. None of it is as useful as they're claiming and it's also ridiculously expensive to maintain and let people make all these requests. There's no way it'll last the way things are right now. Now that OpenAI got their goal of being bought by microsoft I feel like the hype is gonna start to die down since none of it is profitable at all.
The only tool I can see being good right now is the muse chat (though I haven't tried it), the rest of the tools look like they'd cost you more time than you'd save using them. I'd like to see a VS Code plugin for muse chat, I think they could do great things with an autocomplete ai trained on unity docs.
You know the Visual Studio for Unity plugin already does Unity API autocomplete, right? Also picks up references in code for built-in engine callbacks like Start/Update/OnXYZ.
@@mandisaw Oh that's really cool, I never knew about it. Thanks for telling me
@@FactsNoCare No problem :) It does a lot of other cool stuff, like debugging & profiling in editor & on-device (via local network). There's probably a video somewhere on either Unity's or Microsoft's channel...
Sorry to be this guy but. It's prounounced MEWZE. Excellent video though.
waiting for the unreal engine
Yeah. AI coding C++, replacing blueprints 😋
It may still take some years, for AI to produce bug free code. Although should not be to hard, when implementing a virtual compiler for testing in the AI model efforts in coding.
When this was showcased before, it seemed trailblazing, but now it seems very underwhelming.
The texture generation is useful, but I think it's a bit limiting done in Unity, since if you wanna do anything more on it, you have to export it and modify it somewhere else.
If I would have to pick, I would probably want this for a texture artist working in substance designer and painter, not in unity
Agreed, except that I doubt Muse is for anyone actually skilled in sprites/texturing/code/animation, or interested in becoming so. The selling-pitch seems aimed at people who actually *can't* do those things (or don't want to learn). So sprites/texturing/animations for "I suck at art" people, code for non-coders, and I guess chat for ppl who won't RTFM [read the f'in manual] 😅 If you are the least-bit interested in learning any of those skills, you'd want to use the appropriate tools, so your skills are usable outside Unity.
"I am ripping off Warcraft" LMAO I wonder how many ppl actually make prompts like that 😂
AI is both amazing and hilarious
Sorry, I don't mean to be that guy, but it's pronounced "myooz"😅have a great day!
These images are just bad. The concept is pretty cool but maybe use Stability Diffusion or Stable Cascade. They can both be installed locally and are free and open source with great images. Stable Cascade is the fastest image gen I have seen, matter or seconds per image.
Unity is my ex now, I really don't care if it gets a facelift or butt enlargements . 🤣🤣
it is really sad to watch that my favourite game engine getting these useless features just trying to be like ue or godot :(
Compare this to Unreal experimental features... that actually work? I dunno, Unity pretty much ended it own relevance and gave its life for the rise of Godot and Unreal
Insane that they're charging for this lol
Oh so you have to pay...no thanks...
Lol. I thought AI was the future. That is what all the midwits are saying.
WOW! these AI tools look AMAZING! they seem super useful especially Animate, I'm a programmer so animation/art is REALLY not my thing and I would love to reduce my need on third parties for art, it's a shame that these don't have a free tier
I mean its cool they try to incorporate AI tooling in. But to be honest? This is not production ready, this should not be paid at all. Its wonky, it looks like a beta version of a tool that is still far away from a commercial product.
AI slop yippeee that will surely be super duper helpful in creating high quality games yaaay
*end me please*
Unity is trash, not AI. I see a lot of potential for generative AI, but asset generation is not one of them
This is stable diffusion with just chat GPT nothing special, a lot of stuff has stable diffusion and goes to the generation, you train that with steps, like first and better and better, and yes for that needs a long time, just that is some cut out version, people better use stable diffusion don wase you money for this...
not only unity engages in shady business practices, the software sounds bloated as hell with feature creep
people are definitely gonna lose their jobs... but ok, at least not soon (we've been seeing it some of that since last year tho)
MUSE not Moose, I.e. a source of inspiration. Type the word into Google Translate and listen to how it’s pronounced. Do you call your muscles, your Moosealls?
Not reallt interested in generative AI thanks.
Having used lots of different versions of ai generators, muse is trash. It does nothing(!) well and makes NO sense whatsoever. There are A LOT of possibilities to improve the engine and how the creation process works. But start with the basics. People who use Unity are already somewhat comfortable with a higher degree of complexity. Throw ALL that "one click solution" crap into the trash. Because it NEVER works. They better get their sh*t together. This is not doing that.
Great More garbage in unity.
Sadly i will have to use this sh#t to mod Life by you.
It's pronounced MUZE ... one S = Z sound while two S's == S sound/. :p
Is this your new "foliage"?
muse /myo͞oz/
noun: Muse; plural noun: Muses; noun: muse; plural noun: muses
1. (in Greek and Roman mythology) each of nine goddesses, the daughters of Zeus and Mnemosyne, who preside over the arts and sciences.
2. a person or personified force who is the source of inspiration for a creative artist.
"Yeats' muse, Maud Gonne"
🤦♂️
Its horrible, even a kid could draw those sprites
It looks pretty worthless.
some of them look like that now but they have a lot of potential, plus Animate already seems like something I would use, despite being in beta
Because it is
underwhelming
Not a polished product 😕
After all that being said about Unity, I’ll still use Unity. They did a mistake, big YES., they tried to reverse it, another big YES!!
Unity is dead at the moment. Where is CryEngine? These Engine is better.😂
Man, Unity is so trash nowdays
meanwhile everyone is fear mongering about horrible stable diffusion real artists are speeding up their workflow with ai. Interpolation also uses a form of artificial intelligence and it has revolutionized digital animation
You're writing prompts with 5 words, dont expect good results on any AI