Show HN: Generate a 3D model that looks like you and apply AR animation effects
youtube.com
youtube.com
I'm guessing you didn't build the photos -> model tech from scratch in 10 weeks. What are you using for that piece?
- the core tool where generate the model was built with in 10 weeks.
- we actually wanted bring in flutter to improve the UI and smoothness of the app but that wasn’t possible.
- the head portion is from an API. & the body creation is ours. There is a few more things to be fixed.
Thank you so much for the question.
The textures (except for the face) here are obviously very low-res and sort-of generic but that's not all that important - the head and face sure look great, at least in the demo video. The person is recognizable and that's pretty awesome.
I wonder if you could add a feature to export these models for productive use (such as .fbx files) so they can be creatively used in other software (like Adobe AE, Cinema 4d etc.). Having a one-touch solution to create rigged, realistic 3D models, export them and only maybe having to refine the textures on the hands a bit and adjust clothing would be such a time saver. To me, that would add much more value to this technology than the gimmicks displayed in the demo (though they are quite funny). I could also imagine that this is a much easier (or at least a strong second) path to generate revenue. If priced competitively (between $5 and $20 per created character, let's say; I'd differentiate between commercial and non-commercial use), a lot of (semi-)professional animators, graphic designers and marketing people might buy that.
I mean, apart from Character Creator I'm not really aware of a software that even claims to do this properly outside of academic PoCs. And as I said, CC doesn't really work that well, unfortunately.
So that would be my feedback. Disclaimer: I've only seen the demo video, as I'm unable to install the mobile app at the moment.
- Thank a ton for the feedback this is highly helpful.
- We actually had a purchase button last week but removed it because there a bit of manually step involved in it. (Users need to do a purchase request, and we will sending the fbx files to them manually: this has to be automated.)
- Yes, we had to maintain the textures at a lower level to make the AR output was better.
- I truly agree with the graphic designers, marketers, new animators point of view. This could be a huge deal for them.
- We will surely discuss about this internally to take a b2b route if needed.
Cheers. Means a lot. Please do check out the app.
Which actually leads me to another feedback point: If you want to strengthen your B2B prospects, a Windows app (or web app) would probably be most helpful. Most people you will encounter in that space (especially as soon as it's gaming or rendering-related in any way) will be sitting at large Windows workstations rather than on mobile devices, and having your service accessible a click away while they have their 3d modelling software open on their main screen would be most welcome.
The UI could be as simple as a front-end where I can drag & drop the three necessary pictures (or four, if you include the backside at some point) which I took with a DSLR (instead of my cellphone cam), then it displays a 360° preview for me, allows me to click "add high-res textures for fbx" and then "buy it now". I enter my CC or PayPal and it spits out the fbx files. Done :-)
What would be the best way to contact you about obtaining an fbx file of a character? What would be the price point for a non-commercial hobby project? Would you be able to add an option for high-res textures for export only (non-AR use) that would look good when added to a full-HD video in Adobe AE?
I agree that purchasing should be automated in the long run, with as low a barrier as possible (to allow for "impulse purchasing" as well) - certainly no "contact us for a quote" please :-)
Anyway, awesome project, so much potential, I will certainly keep an eye on it and spread the word!
Cheers.
Haha! "Contact us for quote" is current way for us now - sorry for that. Automating the purchase needs a ton of effort at the moment.
Windows app seems like a cool idea as well. Feedback noted!!!
Thanks a ton again for the ideas, comments & questions.
:)
Here are some of the links for you to quickly check out the product:
Demo Video: https://youtu.be/dCEGx-thXD4
Android: https://play.google.com/store/apps/details?id=com.highavenue...
iOS: https://apps.apple.com/us/app/highavenue/id1470937706?ls=1
Tech:
Built using Unity Game Engine, Blendshape & other animation techniques, Segmentation algorithm etc.
Vision & the problem:
Unlike other AR consumer platforms, our vision is to democratize the creation of 3D animated scenes. Anybody should be able to create a short 3D animation of length 15 to 30 seconds and even be a part of it. Usually, 3D animations need big tools to build, weeks to learn, and hours to create. Not just that, it involves various stages like modeling, rigging, texturing, and then animating it. With the above version of Highavenue released, we solved the modeling, rigging & texturing part of the problem by automating the creation of 3D characters. And in the upcoming versions, with the technologies like body Tracking & depth detection, anybody can just hop into a 3D scene, select a character that they want to play, and start acting. It's as simple as that.
This opens a new realm of opportunities for creators and the way we tell our stories.
- Tiktok lowered the video creation barrier from YouTube
- Snapchat lowered the barrier for conversations
- Clubhouse did the same for audio.
We(Highavenue) will lower the barrier to create 3D animated scenes. With just your smartphone, you should be able to create 15 -30s animated scenes easily.
Our Story:
(Bootstrapped Startup)We are a team of four members — Pradeep Baskaran, Soorya Sankar(Developer), Terrance Abilan (3D Full-stack Animator), & Renold J (3D Generalist - Full-time employee) hailing from small towns & tier 2/3 cities in Tamilnadu, India like Anupuram, Bhavani, Thoothukudi, Kanya Kumari respectively (you wouldnt heard about these places before). We are huge fans of anime, games, and other 3D-based content. And working on Highavenue seemed the best fit for us.
During Covid-19, we rented a small apartment outside of Anupuram Township, Tamilnadu, India, and converted the living space to Office — 4 desks, 2 systems, and a rented workstation. We juggle our work within the systems, and there were times when we couldn’t afford a Mac and we would borrow it from other startups for an hour to build and Ship iOS versions.
We bootstrapped our way here after providing services to clients in the US and India. Covid-19 almost made us out of business since almost every client we had and deals we were about to do were called off. And a lot happened:
- We saved up from our previous services work.
- We borrowed some to manage operations.
- We paid back all the debts.
- A few Investors we cold emailed said "No" to our AR consumer platform idea.
- We got the clients back. We saved up again.
- With the little amount we had in the bank — we decided that it was time. We concentrated purely building Highavenue.
We built this product within 10 weeks and shipped it out a few weeks back (later mid of Jan 2021). We are iterating the product rapidly, fixing bugs, building minor features, talking to users repeatedly. It's the love for animations, cartoons, & consumer platforms that drive us.
Any feedback/suggestion would be very much appreciated - It will help us improve. Get in touch with us - email: pradeep@highavenue.co or admin@highavenue.co
thanks! - Highavenue Team
Edit: I realized that my comment is probably more dismissal than intended. The core functionality of your app
„a mobile app using which anyone can generate a 3D Character (within a minute or so) that looks very similar to them with just three photos as inputs. They can also apply different ready-made animations to the model and record some fun, attractive videos with it in Augmented Reality.“
doesn’t inherently need an account to work. So a login/signup screen without anything else as a landing screen isn’t particularly encouraging
I might be wrong. May be introducing a guest login without having to sign up but just to try out the product/tool makes more sense. I’ll note this down :)
Apologies about the delay. There are chances when the capture has some issues.
- At the moment its about: rectifying your model from the backend if the ML system failed to plot the segmentation properly.
- The Selfie photo alone is shared with the AvatarSDK API for head generation.
- I think going the Privacy policy or t&Cs is a huge pain for a lot of users. So I'm planning to create a bulletin and showcase at the start of creating the model so that the users are well informed :)
Somehow missed it. We work on IST (Indian Standard Timing) & we were awake till 5:30 in the morning. Apologies that I somehow missed to reply the above comment. Privacy is something which is very important to us as well. We are and we will be taking a lot measures to make sure that we don't mismanage a user's data in anyway.
And we can make sure it works cross platforms. We will collaborate soon on this :)
I say this as an early employee of a startup that failed, where people have repeatedly asked us for a feature and we wanted to build it "later as we grow". We never got to that point because we never got enough users. Had we built the feature, we would've at least gotten users.
"Sign in with Highavenue" doesn't sound too good at an early stage, to be honest. Even as a future plan. For game/AR/whatever developers, requiring their users to create an account with Highavenue will probably be a dealbreaker.
I also notice you mention in another comment that your goal is to create a network around it, and that's why it requires a login. I admire the sentiment, but at this stage I don't think it's the way to go. I see a very cool tool that I'd take for a spin, but no reason why I'd join a network around this. I think, as a medium, 3D animations are far less general and expressive than video/audio/images [for the same amount of effort put into creating], so I'd be wary of trying to emulate what TikTok/Snapchat/Clubhouse did. It might very well be that the medium is different enough that no one can use their creativeness effectively with it, making it DoA as a social network per se.
(Aside: we also had a mandatory login screen, and even with a landing page, when we implemented an automatic "guest mode" which dropped you into the webapp instantly, people actually started using it and signing up.)
I'd say pack it up as an API, offer to developers at first. Then you're B2B and that's much easier to charge for than B2C. When you see cool stuff that developers do with it, you can partner with them, you can go consumer facing gradually (and with limited risk in case it doesn't work out) and you'll have some capital to put back into that growth.
I see a lot of similarities to earlier me in your responses - you're very defensive of your work and the choices you made, and that's fine, but you're walking a fine line with just dismissing initial feedback because of the sunk cost fallacy. I was dead set on "listening to feedback" as well but ended up doing the same thing. Don't take any of this the wrong way, not trying to disparage or discourage, it's a cool product, and in terms of startups I know nothing on the grand scale of things. So if it feels right ignore all I said and best of luck. But also consider what I'm saying even for just a second :)
Since we are bootstrapped we couldn’t build everything at once. :)
You might be right, B2B play seems straightforward. There is this powerful intuition that we can make it work this way. Thank you so much for the question.
It did a pretty good job with pictures of Tommy Wiseau and Nicolas Cage in my testing, though it does raise the potentially creepy issue of impersonating non-celebrities in VR social environments.