Test Out Meshy.AI With Me - The Results Are Surprisingly Decent
Vložit
- čas přidán 1. 07. 2024
- Meshy.AI provides the ability to generate 3D models from a text prompt. Come along with me while I test out the features and try to generate ocean creatures for some of my projects. I'll go step-by-step through the process of generating and refining models, plus I'll bring some into Cinema 4D for a quick render to see how they look in more of a project setting.
Try Meshy.ai out yourself at www.meshy.ai/
If you decide to upgrade to a paid plan, save 20% with Promo Code: tvermilye
It helps me buy more coffee and make more videos. :)
Other creative work and info about me: www.travisvermilye.com/
My Gear:
Computer: amzn.to/41BuZH9
Monitor: amzn.to/3NF1wWU
Microphone: amzn.to/3RDlvq6
Camera: amzn.to/3NJQ4JF
Desk: amzn.to/3TCxPcL
Chair: amzn.to/47i7bZT
Like this video? I really like coffee! www.buymeacoffee.com/travisve...
#c4d #meshyai #ai #aimodeling #texttomodel #review #softwaretest #softwaredemo #conceptart
Affiliate Links: Some links that appear in this description allow me to earn money using affiliate advertising programs.
00:00 Introduction
00:48 Meshy Interface and Making a 1st model | Coral Reef
04:44 How to refine a model
06:35 Second model try | Sea Anemone
10:24 Third model try | Jellyfish - This one is pretty good!
15:12 Fourth model try | Wonky Octopuses
20:11 Fifth model try | A Blue Whale
21:50 Get Free Credits with a Quick Survey for Bonus Credits
23:20 Meshy Fee Structure to Get More Credits
24:30 Make A Quick Render in Cinema 4D
26:55 Wrapping it Up
If you decide to upgrade to a paid plan, please use Promo Code: tvermilye
for a 20% discount. Plus it helps me buy more coffee to fuel more videos. :)
I love your slow and thoughtful pacing here. I also appreciate you're testing methodology. Thanks for the video!
This was really interesting. Thanks for going through this tool and seeing what it's capable of.
You bet!
@5:14 "It looks pretty impressive." The phrase we laugh over in a short time as AI develops. Thanks for the video, great to see how you approach this.
this is quite impressive, thanks for the video!
they look like fantastic base meshes for starting a new sculpt
One thing that might end up saving some 3D artist jobs for some years (I don't think everybody will be replaced, but I think our job will be mostly as art directors for smaller teams) is the fact we already use base meshes and kitbashing so extensively that AI will have to do a whole lot better to be useful if compared to combining different base meshes together.
Now I think AI will be great for quick textuirng for those models who will end up being far in the background of a scene.
Agreed. I think the tech still has a ways to go before it will be able to directly replace a 3D modeler/texture artist for any important elements.
This! - but you better be really good. Because those tens of thousands of jobs will go down to hundreds I think. And then they will get Autonomous agents to BE the art directors do everything the human ones will do & then boom! those last people are out. Timeline 🤔 I predict AGI in about 3-5 years. That’s about the time when an Ai agent can absolutely understand a brief. The exact style you want. Type of game colour scheme. All the good stuff.
And a game director will be able to communicate to it in some pretty obtuse language too that it will understand & act on. I wish all the young artists well. It was better in my day ( started very late 80s )
As a 3d modeler, I tried it but output is too low resolution be it triangle count or texture wise. For the time being the only way I can make use of it is as placeholders on a 3d scene until I get to create the asset myself. I am pretty sure with the modal Sora is using we will get there though.
You're not using it right. The model has to be refined using a button on the website after it gives you the first output. This will also, in turn, give you the option for retopologizing to make the mesh quads only. This also stables out the textures and allows 2k and 4k texture exports if I'm not mistaken.
I've played with every txt to model out there and this is by far the best so far. I even tried to program a model in blender. Instead of a spider, it looked more like a water tank with four legs. It can only get better from here.
The people who really benefit are the ones who can implement Ai tools like this one into their workflow successfully to speed up the process and reduce the stress on themselves.
You think capitalism will let you keep that productivity increase to pay your bills? 🤣
@@DirtiestDeeds lmao exactly. this isnt here to improve your workflow. it is here to take your job.
Thanks a lot for the nice effort ... Thumbs up
I would recommend getting to know some Sculpting Tools in Blender and then some texture painting after and you can pretty effortlessly fix most of the problems on those.
I would say if you are going to go that route you may as well just make the models yourself, there is some satisfaction you gain when you actually create your own art.
@@sound4mation I could speed sculpt the garbage this app generates, within the time it takes to generate
Well, as a 3d artist it's good to know I don't think I'll be replaced for a job quite yet 😅. But I could totally see these types of models helping out smaller studios/projects so I guess that's cool
Yea! Give it a couple months. You'll always have a job mate, it's just gonna change, you'll implement AI into your workflow, in fact literally everyone is using it in some capabilities.
If humans always retain their hobbies and interests, even without the need to work, AI would serve as a tool to enhance these pursuits rather than replace them. AI could introduce new ways to explore and engage with hobbies, making them more accessible and offering novel methods of expression and experimentation. This would likely lead to a richer and more diverse landscape of hobbies and interests, as people find innovative ways to use AI in their personal pursuits.
that is getting very close to being a TOP tool
be interesting to see the material map in PS and if its worth refining that.
This tool is definitely useful to generate LOD meshes to fill up distant details… at least when not doing singular objects… meshy can get messy with environmental generations or objects that rely on tons of details, but gets really decent results on more simple single object (for instance the aquatic reef one, it could do much better if you generate the plants etc separate (but here is the downside with credit systems, it gets expensive really fast where just getting someone from fiverr might be the cheaper option)
Pretty good for background stuff. They look quite terrible up close but hey, it could even work as a base model for sculpting and doing your own textures. Overall, any productivity increase is welcomed in 3D since its so work intense.
Interesting, I wonder if there is any option to import your own UV'd models and with a text prompt have it texture them? Of course to be really useful it would need to export multiple texture maps, eg a PBR format.
You can already do that using stable Diffusion and projection texturing in Blender.
@@GeometricPidgeon Oh dear, won't be long before I have to learn a new trade.
Will this work off an original sketch? I’m looking for a tool to go from a traditional design sketch to 3d model
There is an image to 3D model option, but I haven't had a ton of luck with it. Not sure what it would do with sketches.
are those game ready / performance?
Wow! This is great to get a starting point for further modeling. It takes away the most boring and tedious parts of modeling (uv-unwrapping for instance) and you can still do the fun parts on top and rig the models for more fun. Thanks for the review!
I can imagine the next generation of the stool would allow you to give it a photograph and you tell it how you want to modify the thing in the photograph and it does a 3-D model of it.
When you look close at the textures, while detailed they are, just random crap thrown on to them based on the creature reference. You would have to do a lot of editing on any of those textures to make a usable model.
Creativity doesn't depends on tool, the practical puppet industry struggling when 3D came, learn the tool, use it to do the meta idea and push to the higher ground, and not bond to the tool.
Thank you for showing how this works using the example of several models. For me as a 3D artist this was interesting. As a final result, it’s apparently still too early, but using Meshy as a basis for further development is quite worthy. Thank you Keep going, it's interesting!
Do you have to go through disord to use messy?
Never mind. When he said 3d I thought he was talking about 3d printing...
not so good but they serve as base meshes which saves a bit of time. so is useful.
Love sculpting new base models and such and really looked forward to this tool, but Honestly even with many iterations of images and such, the model is just horrid on the image to 3d model function.
I can recognize some of these model from different website offering 3d models both free and for sale , in particular the octopus.
Ydah. Is in thingiverse and printables and others
Link?
@@hraesvelgML Cannot remeber , but you will surely find in some website offering 3d models
It's an interesting tool, but I'm a little turned off by the way they're implementing credits. I'm not objecting to them having the system at all - I get that they need to be paid for their work - but I think that one would likely need to iterate the description several times trying to teach the AI how to get what one wanted. So it would be nice if there was a "revise the description" option that could be done either for only 1 credit, or for no credits but a limited number of times.
Still, for what I do, I'd really need a "text-to-STEP" AI. So now I'm going to go google that and see what I can find. :)
Any chance you could run the file on an FDM printer?
I think you’d have to subdivide the mesh to give it more resolution, but definitely a possibility.
@@travisvermilye Looking forward to this possibility as the eventual outcome of AI. 3D image generation.
When will you making a video of your first AI generated 3D print?
can you upload an image to give it an idea to work with?
There is a separate image to 3D model tool, but not the combination yet. That would be a great addition.
I think it will get so saturated to a point that "real" artists will become masterpieces themselves, people would pay more to make "human" content. Just like those dug up artifacts. Artists will become gods of these A.I creatures. Every artist will become his own studio. Their own unique style will be trained and would be patented. So in the end AI will still become a tool. Like rest of the animals down the food chain
I don’t think that you create anything…
I was with you for the first sentence
Meanwhile, the world economic forum is projecting massive job loss to AI… artists will be idolized even more (already are in hollywood) than they already are while the average person struggles in an AI driven world and the gap between the lower and upper class grows as the middle class erodes. Fast forward to the first AI protests/riots.. 2027… 300k people in the streets in every major city worldwide
That is a load of BS. Artists will get thinned out until only those remain who can afford art as a hobby. Real craftsmanship will mostly vanish in the population, because people wont be able to earn a living with it. People are greedy. It is already happening. No point denying it.
Thanks for reading this BS paragraph written by Chat GPT 😂
Have you tried Luma AI text to model generator, gives great results
I haven’t but I’ll check it out. Thanks!
@@travisvermilye Another 3d mesh AI tool you should try is CSM 3D, it's "image to 3d". Meshy, Luma and CSM 3D are the three tools I know if who all do these similar things. Would be interesting if you make a video comparing all three.
Thanks,@@Tystros I had just such a video in mind. I haven't tried CSM 3D yet, but I'll check it out. Cheers!
We are going to be in the sweetspot for this guys! It won‘t replace us but it will do enough to really speed up your workflow
All new AI tools will try to collect high pricing subscriptions, until a similar . better, free alternative tool arrives. Then they will have to start working for their money. Which will be the coming weeks for all of them😂
❤❤❤❤❤
You should have given party hat as negative prompt at 7:35
Oh you ended up with the party hat lol xd
The only issue that I see is that it’s not creating real geometry so uses for 3-D printing are limited, which is a shame
I tried a genshin impact character, and i got 2 different faces awkwardly stitched together, so it had 3 eyes. and other issues xD
Do not use it for humans xD better off with UE5 metahumans.
Yes, also tried to create some humans, I had double faces and 3-4 eyes as well...
Uh, those are pretty bad? o.O
I stopped at the jellyfish... wth
I think the 3d artists would still get to be involved, but they better create something real good...something perhaps lean more to great design, because to AI, design is a bunch of mathematics to them...
LumaAi 3D Genie is pretty legit as well.
I’ll check it out!
@@travisvermilyeyes, right now Luma Genie is free during its research phase. Meshy charges 4x the creation fee (20 credits) to download the model you already created, which is a bit nuts to me. so if I paid the monthly $16 fee to get 1000 credits, that's 100 characters I could generate. not bad. but then I have to download them (perhaps manually, individuality - they could learn from midourney's awesome mass downloader feature there) and downloading 100 characters from meshy would be 20x100 credits, so 2000 credits, so:
you create 100 characters for 16 dollars, then download them all for 32 more dollars, for a total of $48 dollars. of course it's still way less than buying or paying to have made, professional models, but still not as good as Luma, which is currently free.
Sora will leapfrog over the need for CGI. My career was a Lightwave, Maya, and Blender generalist.
Right! I don't think a lot of these commenters have seen Sora yet! It will soon replace most 3D, Filmmakers and artists.
I went to 3D after AI taking over 2D..now i need to get a "real" job
Kadiem is the best
Tried a couple prompts. Tool is totally incapable of creating anything but 3d cartoonish low-res shapes, even in the "Realistic" mode. Thanks for the video, but it definitely needs a ton of work before it is ready for prime time.
imagine in 1 year from now.
Tried a bit, not very convinced. I think there are already better ones out there, especially for characters.
CZcams as a platform is not behind the quantum computing race, but the content creators in the platform, are.
thats only possible because the use already existing 3d models...
I am sure with time the app will generate pretty solid models, but I think the pricing is not that great. You really do not have a lot of room for experimenting unless you pay a lot of money.
Not useful for what I need. I wish it could create things that are VERY specific. Like, "A Heinkel He70 German WWII attack plane" etc.
slot machine :)
It's crap. It just creates a very low poly mesh, the refine just slaps on a more detailed image onto a low poly mesh All the time eating your credits.
Not impressed.
I'll pass thanks.
I still remember 2DArtists making fun of A-IArt a few years ago. Good old times...this is basically the same
This might be an expensive tool. Lets get real, 1000 credit for $20 isnt a lot. Refining a model and downloading is like 40 credits. So its 25 models for $20.
i stoped watching the moment i heard "200 credits" yep, Bye!
I wonder how long before people aren't needed anymore for creative industries. I also wonder when art and design courses at colleges will cease. It's interesting to watch. I pretty much stopped learning blender last year because AI will be way better than humans in a few years probably. All the best
I really think people will always be needed in the creative industries. AI is only good (now at least) at generating images and art based on or drawing from what has already been done by humans. If we all stop making things, wouldn't it eventually stagnate and just keep spitting out the same old stuff? To me, it is the job of creative people to figure out how to use this tech but also keep creating new things - maybe using what AI makes as a starting point. Only time will tell, I guess, but I'm watching with cautious enthusiasm.
If you get discouraged by advances in technology then art was never meant for you in the first place.
@@jeanpierrepolnareff9919 I'm making a case that advancements in AI might very well replace creative people. Many different careers could be replaced as well as artistic ones. This is more than just an advancement in technology such as a new version of photoshop or a new software. It could be much bigger than that. Is it wrong to ponder the possibility of that happening?
I don’t think AI will ever replace humans when it come to making new design that have never been thought of before like someone else has already said.
Before AI, many rubbish music, movies, art is made. After AI, even more rubbish music, movies, art will be made. As long humans are capable of defining what they like and dislike. Or at least a percentage of humans. Because most people just eat what you feed them.
How soon should Award ceremonies and Award shows be done away with ? at what point is true excellence no longer a requirement when soon AI will be doing all the work ? Hundred of years of creative excellence and celebrated heritage, finally burnt to the ground in the age of pretend and pursuing better $$. I don't mean you Travis, i'm just talking in general about the soul destroying situation all of us artist find ourselves in.
It's impressive, but their pricing model is terrible. I have to pay money for their system to generate garbage that I throw out? Heck no! Charge more to download and let us generate for free. Charge 100x more to download if you want, but pay to gen and pay to upres is just stupid. They're shooting themselves in the foot before they properly launch.
No. Saved you 28 minutes.
now do something symmetrically mechanical......
Try to make an heinkel 111-P with historically accurate full interior and then you will understand that AI is really dumb!
while the ai stuff is cool its still gimmicky give me an ai where i can upload a base mesh and have it create an awsome character with game ready topology...
Pro tip: If youre gonna use ai then at least dont say you are going to "work" towards to "creating" anything. When it isnt you doing any of actual creating.
The mediocrity of ai material will actually make the market more welcoming for actual artists. Yes, ai material could catch up. But, it has been 2 years since its release (generative ai) and the whole hype, where a lot of people in my surrounding are commenting that its application is useless and disappointing in real life, and that artists are literally necessary even with the introduction of these artificial generation models.
I am not here to put flack or to flame the tech and your video, just wanted to share and uplift anyone who is/was stressed about ai "replacing humans".
I agree with you completely. I feel like it is very much in the experimental stage and will likely be there for a while. Like some others have commented, maybe good for an initial model to block something out, but not yet at a stage where anyone should be frightened. To me it is similar to what happened when computers entered the art space in the 80s. It changed things but did not remove the need for artists and creators. This is a new tool and it is up to artists to figure out how they can use it best.
Yeah but this is bad for the industry in general because it is dissuading newcomers from learning the craft then the cost of hiring cg artists will go high for a while until ai catches up then everything will be done for everyone. Except for artists making art for the sake of art .
@@travisvermilyeHow are you comparing it to that 😂 yall are coping at the fact that this will be plummeting the demand for artists in a few years. If anything, computers only switched the medium through which art was made, it didn't open the door for the average joe to just hop on a computer and prompt it to make something for them. Not to mention the huge learning curves all these programs had prior to AI
@@ohgeezus_ I can see where you're coming from but think about typesetters and graphic designers for example. They went from doing paste-ups, working with their hands, using photography equipment, and manually setting type, to it all being done on a computer. People in general were either excited or afraid. This happened in nearly every industry (like AI), and it made many people afraid of the changes it would bring. This of course happened much more slowly than what is happening now with AI, but the overall sentiment is what I'm comparing it to. Any time new technology is introduced, there is a mix of fear of change and excitement for it. I personally don't believe the demand for artists will go away. If anything, artists will be better at using these tools than other people because we already think creatively and will demand more from the technology. Anyway, nobody knows for sure what's going to happen. We'll just have to wait and see and adapt as we go.
The point of creating anything is to do it yourself. It is a product of YOUR supposed talents and imagination. As someone who has been involved with technology all my life and understands AI, it really pisses me off to see another technology used poorly, by the greed driven and idiots. People are completely missing the point why HUMANS f-ing do things. I have always believed that technology should ONLY be a tool to assist and never replace.
There we see an example of it being used as a tool to assist. Nothing of value is being replaced.
you cant pay me to use ai
and i will never use it and its really cheating making artwork
pronounced Sea EN-NEN-IMY. That said, Thenk Yew!
HIGH KNIGHT #5
Hi, I just stumbled upon your vid with an intriguing title regarding a category, “IFFA Ayyye iIEEYyye!!!..” is a threat to the social economic order (aka) “c r e a t i v e s…” imagines a world of it’s destruction!! We’ve been swimming in it for quite some time now, this AI thing that we’ve develop 50 years ago! But that is not why I’m leaving this comment, NO! I noticed your background reflects your “eNnerre...” emotions and brightness and contrast need not apply. “My Dude!” as all the clueless Gen Alpha and Zeeyuurzz like to say to each other… that painting or photo at your set magically blends in with your own hair which reminds me of a scene in the movie where Chris Tucker is dressed in his role as a transgender MC in the “5th Element” which maybe the case in his private life… “yoohua know H O L L Y W O O D!!” -Thris Cucker. Anyways, I just made myself laugh there! (APPLAUSE SIGN) Ya man, I think you look like Chris Tucker or Thris Cucker at the intro of your video which is information that hedges my future bets in ME or “mooarRh…” appropriate as being grammatically abbreviated from “I am” is the reason why selfishness is on the rise! Take control of your life by investing in yourself and not feed the machine!
And this is the WORST that it will ever be! The possibilities 🫠