Meshy is your 3D generative AI toolbox for effortlessly creating 3D assets from text or images, accelerating your 3D workflow. With Meshy, you can create high-quality textures and 3D models in minutes.
Hello Product Hunt community!
I'm Ethan, co-founder and CEO of Meshy. I got my Ph.D. from MIT in computer graphics and AI, and I've spent a decade in CG research and industry, building strong connections in the game, film, and XR industries. A common challenge I've heard from my network is the difficulty in creating high-quality 3D models, which led to the founding of Meshy. I feel it's my destiny to tackle this challenge, and our entire team at Meshy shares this commitment.
Today, we’re thrilled to introduce Meshy-1 designed with a singular vision: to turn everyone’s 3D dreams into reality. Our 3D generative AI empowers content creators, enabling them to craft stunning 3D models from text and images in a mere minute.
Meshy-1 boasts three user-friendly modes:
1. Text to 3D: Text prompt in, 3D model out.
2. Image to 3D: Front-view image in, 3D model out.
3. Text to Texture: Upload your model and text, and Meshy generates high-quality textures.
Below is a quick tutorial of one of the latest features, Meshy-1 Text to 3D, on Discord just like Midjourney. Using this feature, you can create whatever 3D model you can describe, be it a pineapple Mario, Einstein on a skateboard, etc. The only limit is your imagination!
Here’s how Meshy-1 stands out from its predecessor:
🚀 Blazing Fast: Meshy-1 delivers in just 60 seconds! Gone are the days of waiting hours.
🎨 Style Control: Choose from a range of styles, be it realistic, anime, cartoon, or comic.
🔧 Seamless Integration: We’ve expanded our file format support to include glb, fbx, and usdz. Plus, for the Unity enthusiasts, we’re rolling out a Meshy-1 plugin.
🔍 Crisp Textures: Experience textures in 4K resolution with top-notch PBR!
Who’s the ideal user for Meshy-1? Anyone looking for a swift, efficient method to create 3D!
🧑🎨 3D Artists: Looking for a rapid modeling solution? Use Meshy to transform your ideas into ready-to-use models!
🎮 3D Hobbyists: Tried mastering Maya or Blender but felt overwhelmed? Meshy is easy to use.
👓 XR Developers: Elevate your VR experiences by instantly converting text or images into 3D!
🏡 Indie Game Makers: Scoured SketchFab but didn’t find the perfect fit? Just describe what you want in Meshy!
🔮 Designers: Eager to visualize in 3D during brainstorming sessions? Meshy brings your 3D ideas to life!
For a deeper dive into the journey of Meshy-1 and our thoughts on the future of GenAI for 3D, check out our blog post. If you are making a 3D app and wish to integrate 3D AI generation, our API doc is here.
We’re eager to hear your thoughts!
@yuanming_hu congrats on Meshy-1 launch, Ethan! Amazing to see 3D modeling made easy and fast - models in 60 seconds is a game-changer! Great for pros and hobbyists alike. Can't wait to see how Meshy-1 changes the 3D world. Best of luck! 🌟
@yuanming_hu@jasonh19 Hi Jason! To celebrate our new pricing model and our launch on ProductHunt, the discount will be valid for the next month! The discount code is applicable for all subscription-based products :)
Began exploring Meshy about a month ago. AI still has a ways to go particularly with text to image and text to 3D. This is common across all AI generators.
Where I have found Meshy most useful is text to texture. Using other AI I had to provide models on all axis to get a decent texture all round. Meshy does this with a single model, though occluded faces would be blanked, projection is being improved I am told. They already increased texture resolution to 4k which is a decent step towards use in a commercial production sense. None of these AI generators are useful at production level at this time and requires a bit of work on my end (fixing small errors is no great chore and worth the time saved in other areas), but I believe they will get there eventually and are worth getting into now to follow progress.
Developers are friendy and approachable, in fact they encourage it.
@jim_logan_rolfy_ Good to see you here, Jim. 3D AI is much harder to build compared to 2D, and our daily development is full of challenges. But every time we see what you create with Meshy, our entire team is so motivated! You are right that 3D AI is not yet fully production-ready, and we agree, and we feel like it is our destiny to make it production ready!
I've played with some of these text-to-3D tools. It's super impressive. But many still require a ton of tweaking to get to production-level quality. What's the typical workflow or use case for Meshy, @yuanming_hu?
@rrhoover Good point, CG production puts a high bar on mesh & texture quality. Currently, people have been using Meshy Text to 3D for a few interesting cases, such as generation props for environment art (they are not heroes in the scene, but you need a lot of them!) Indie game developers are ideal users for Text to 3D. You can also use Text to 3D to create a quick conceptualization for level art design.
Apart from CG production, you can also use these models for 3D printing - a use case without the limit of UVs, topology, and polycount!
@rrhoover game assets should account for the bulk of the use case. 50% of gaming is mobile based, current genAI 3D cannot serve this segment due to mesh density problems, 30% of gaming is console (AAA), also challenging due to insufficient mesh quality / downstream customization, the remaining 20% PC (browser/boxed) could be interest, I can show you some interesting ways to play with this.
I’ve used Meshy, and it’s incredibly powerful. This is a must tool in the toolbox for anyone working with 3D.
The fact that we’ve come this far in such a short amount of time is crazy.
Super excited to see the team launch here, and wishing them all the best on launch day.
Now go check out Meshy-1
@linusekenstam
Thank you Linus! We aim to make the best generative AI tool in 3D. And we are keep evolving! The new version of text to 3D will make you surprize :)
@linusekenstam Thank you very much, Linus, for being an early user of Meshy, and for your support! Indeed, It's not an easy journey, but we are happy with where we are today. 3D generative AI is moving so fast, and even we ourselves couldn't imagine one day we could deliver something like Meshy-1 half a year ago.
In the future, we will continue to deliver as much as we can, and we really wish we can lead the adventure in using AI for 3D creation!
@yuanming_hu Congrats on the launch! Meshy will be super useful for game creators to create basic objects super fast and shorten the development lifecycle. Impressive product.
@yuanming_hu Recommend adding Japanese translation, Japanese people tend to select mediocre product with Japanese translation rather than best product only in English UI👍
Looks promising! It would be cool to have more detailed tips on how to write promt, because no matter how many times I tried to create something usable, I couldn't
@eugene_lipski Good point... We are planning to release a prompt & result sharing feature, together with a marketplace for users to trade their models! I hope that would further lower the barrier for people to make use our product.
One attempt we've made: Meshy supports 20+ languages so for non-native English speakers (including me!), hopefully using their native language makes life easier!
Awesome looking platform! Once the 3D object is rendered, can you pull it into blender or whatever app you're using and touch it up?
I think this will save 3D designers TONS of time!
Congrats on the launch!
Congrats on the launch, @yuanming_hu! 🚀
Meshy-1 looks like a really cool AI-driven platform for 3D model generation.
Given the teams expertise in computer graphics and AI, I'm curious about the future of AI in this field.
How do you see AI evolving in the context of 3D modelling, and what steps is Meshy taking to stay ahead of the curve and minimize any risks associated with AI-generated content?
@dzaitzow Thank you, Daniel! My answers:
> How do you see AI evolving in the context of 3D modelling.
It's moving really fast. I've been working in the CG research & engineering field for a decade, and have never seen such as strong & focused attention from the entire academia on a single problem: generating 3D models from text or images.
At this point, I think we have only solved 10% of the 3D GenAI issues (assuming 100% means Midjourney quality in 3D). But I expect the remaining 90% will be solved in an accelerated manner. My estimation is in 2 years, there will be a "3D Midjourney" out there.
@dzaitzow
> what steps is Meshy taking to stay ahead of the curve
Great question. Probably no magic, just work hard and stay close to the user's needs :-) We have a solid background, but our competitors are smart too.
@dzaitzow
> minimize any risks associated with AI-generated content
We are still figuring that out. For copyright issues, we are following Midjourney way to deal with them. One good thing, though, is that generated 3D content is less prone to "deepfake", since converting 3D models to something photorealistic still needs expertise (actually very hard), which is different from 2D, where you can easily generate fake news photos that are hard to distinguish from real ones...
@yuanming_hu Very cool! and how do you see Meshy fitting into that '2 years from now' world? I guess a better question is what are you hopes for the platform as 3D GenAI continues to reach that '3D Midjourney ... or beyond' landscape?
@yuanming_hu oh it looks like you answered some of this here! I think in such a fast, ever-changing landscape that is the best approach! You can't tell what is going to be around the corner to keeping informed and listening to your users sounds spot on!
Incredible text to 3D and more! No limits to your imagination. If you have a creative mind, you can indulge yourself in creating everything that comes to mind, in a 3D version. The latest update has brought further improvements, furthermore in just 60 seconds 8 results of what we are going to create are generated. Of these, we can request an upscale or rather a refinement of the result. There are 3 3D generation possibilities which are as follows: from text, from image and the fantastic retexture function. I am one of the lucky winners of the retexture contest which gave away color stickers printed with a 3D printer! It is possible to export your creation in video format and in the glb format exportable to 3D editing software. Rating: 5 stars deserved!!
@max_casu Wow, Max, good to meet you at PH! Really appreciate your insightful comments. If it's not too much trouble, could you kindly replicate your review on our official review page at https://www.producthunt.com/prod...? That will allow much more people to hear your thoughts. Thank you so much for your time and effort, and we will try our best to improve the Meshy product!
@yuanming_hu Yes, I'm more than happy to do it. I put a review in the comments because it doesn't allow me to give a rating and write the review. I can only read reviews from other users. 😢 I'll try again.
@max_casu Roger that! Not sure what that happens - maybe you can retry later. In any case, I feel so happy to have your review and support, regardless of where it is posted!
@yuanming_hu I solved Ethan! From the PH app it doesn't allow me to write reviews and give a rating. From browser it works... You deserve a review and I wanted it published! ♥️
Hi Ethan and the Meshy Team!
You've created an impressive tool that undoubtedly has the potential to revolutionize 3D modeling. However, I have some concerns and questions regarding Meshy.
Firstly, while the idea of converting text and images into 3D models sounds appealing, I'm concerned about the accuracy and detail of these models. How detailed and accurate are they, especially with complex and abstract requests? Secondly, how does Meshy handle copyright issues, considering users might request the creation of models that are protected by copyright?
And a question for you as the founder: What challenges do you foresee in the development of Meshy, and what strategies are you using to ensure continuous technological advancement in 3D generation? How do you plan to evolve and improve Meshy to stay at the forefront of innovation?
Thank you for your work, and I'm looking forward to seeing how Meshy develops and impacts the 3D modeling industry.
Best of luck!
@drsudba Thank you for your detailed comment! Great questions!
> while the idea of converting text and images into 3D models sounds appealing, I'm concerned about the accuracy and detail of these models. How detailed and accurate are they, especially with complex and abstract requests?
At this point I wouldn't say the generated models have the quality level required by hero assets (e.g., main characters or items) in games, but they may serve well for props in an environment that is far away from the camera.
Regarding complex and abstract requests, AI may not do well since it may not understand. But for a "abstract" request, what's the expected behavior, if that's given to a human modeler? :-) So I would say it's the user's job to find a way to clearly specify what they want.
> Secondly, how does Meshy handle copyright issues, considering users might request the creation of models that are protected by copyright?
The users own the copyright of the generated models. Our copyright issue follows Midjourney, so if you are trying to use what Meshy generates for commercial usage, attention should be paid to compliance.
@drsudba > And a question for you as the founder: What challenges do you foresee in the development of Meshy, and what strategies are you using to ensure continuous technological advancement in 3D generation? How do you plan to evolve and improve Meshy to stay at the forefront of innovation?
Great question... The journey to where we are today wasn't smooth, but we enjoyed it, and we are happy with our position now. We are a team of researchers, engineers, and artists, and we work hard to ensure our generation algorithm is the state of the art, and our product has a good user experience. The 3D GenAI sector is highly competitive. I would say as for all startups, finding the first group of (paying) customers and achieving product-market fit is the most important step. Our barrier towards PMF could be generation quality, or an immature market (3D market is much smaller than 2D). But we are happy to take the risk and move on!
Congratulations for the launch Ethan, super excited to see your team kicking off the generative AI on 3D. Any plan introducing enhanced 3D models for characters with auto rigging? That would be super useful for storytelling sites like us at Artflow.ai.
@tim_zhang_artflow Thank you Tim! Auto rigging could be a potential next step! For now, you can upload your model to Mixamo and rig & animate on your own :-)
I've always been interested in 3D. I even tried to create something. Installed Blender. Created a donut 🍩 It was difficult, but I think that thanks to your tool it can now be made much easier
P.S. I’ll take Meshy now and create another donut 😂
@gipetto Thank you Petto! Hope you are having fun with Meshy. I tried to learn 3D modeling in Blender, never succeeded (couldn't even create a donut!) :-)
@pontus_bjorkberg The API usually comes 1 month after we release the features on Discord/WebApp. You can already use our API for Text to 3D, Text to Texture and Image to 3D following the doc here! https://docs.meshy.ai/api-introd... Snapchat is using our API to allow its creators to easily create 3D, and we hope our API will bring the same value to you! More details: https://www.linkedin.com/posts/e...
Hi Ethan, and congrats on the launch! The only thing I know about 3D-modelling is that it takes time and quite a lot of effort, especially in case of complex images. As a product manager in edtech creating content for school programs, I wonder if I could create realistic 3D-models of elements of cells, or nature (like the cycle of waterflow in the nature), myself, or would I still have to find a creator? And can I find a creator in your discord community?
@arina_spiridonova Thank you Arina! For your question: you can simply join the Discord channel, type "/create a cutaway view of a plant cell", and see how it works! I made an attempt for you: https://discord.com/channels/108...
@arina_spiridonova My feeling is that expecting it to generate the fully correct complex structure of a cell is still too much at this point. For simpler structures, it should work!
Meshy