(Incomplete Submission) - Disqualified
Welcome team to Hackatron S5, this looks amazing. I like the graphics. I am waiting for a demo video.
All the best
3 Likes
Oh wow, this is giving some sci-fi vibes and quite intriguing having through read through each paragraph of your offering. I see “Blend: Tron’s Ultra-Realistic Game Character 3D Face Reconstruction Engine using Monocular Depth Estimation and Hybrid CNN Deep Learning Architecture with Real-Time Post Skeletal Mesh Fusion” is a groundbreaking initiative aimed at revolutionizing the blockchain gaming and metaverse experience on Tron. The project uses AI and deep learning for 3D face reconstruction, merging the user’s face into a game character skeleton and storing the 3D files as TRC-721 tokens on BTFS.
Out of curiosity, I’ve few questions I need clarity on;
How does Blend ensure the privacy and security of users’ data, especially when using a single selfie for 3D reconstruction?
How does Blend plan to tackle potential challenges related to different facial expressions and emotions during 3D face reconstruction?
2 Likes
Hi @Prince-Onscolo, happy to hear that. The demo video will be open public by today or tomorrow. I will make an announcement in this forum post. Thanks for waiting.
1 Like
Hi @manfred_jr
-
Our on-cloud computation will only retrieve and use the user’s uploaded single-view selfie once during the process and will not save this data again. The Blend’s NFT Passport, which the user mints, will be responsible for storing these 3D-related files as well as acting as a link between the user and any blockchain game developed with Blend.
-
Our deep learning pipeline is able to predict and process situations like multi-angle faces, extreme illumination conditions, light occlusion, and extreme facial expressions in the single-view selfie image the user provided. As Blend is trained with a wide range collection of different face conditions and utilizes a modified version 3D Morphable Model.
3 Likes
Welcome to Tron Hackathon season 5, your project is good but there are some important questions.
Q1
What are the specific challenges of reconstructing a 3D face from a single selfie image, and how does the Blend pipeline address these challenges.
Q2
In what way does Blend’s NFT Passport concept work and what are its benefits, and what are some of the key technical challenges of integrating Blend with Unrea.
Q3
Elaborate the advantages of using a hybrid CNN deep learning architecture for 3D face reconstruction, and in what manner does Blend plan to achieve even greater heights in the post-hackathon period.
Q4
What is the accuracy of Blend’s 3D face reconstruction pipeline,and enumerate the computational requirements for running Blend in real time.
Q5
By what means does Blend ensure the security and privacy of user data,and identify some of the potential use cases for Blend beyond gaming and the metaverse.
Cool, thank you
I will be waiting
Thanks for replying, it was fun reading through your response. It’s great to learn that Blend emphasizes data privacy by only using the user’s uploaded single-view selfie once during the process and not saving it again. The concept of the NFT Passport securely storing 3D-related files and acting as a link to blockchain games is a commendable approach, the ability of Blend’s deep learning pipeline to predict and process various facial conditions like multi-angle faces, extreme illumination, light occlusion, and diverse facial expressions is impressive. It shows the robustness of the system and its potential to handle real-world scenarios. I’ve a few questions moving forward;
How does Blend ensure scalability to handle a large number of users concurrently and maintain high performance during the 3D face reconstruction process for real-time integration into games?
What are the potential monetization models for developers using Blend, and how does Blend plan to sustain its operations in the long run, ensuring continuous support and enhancements?
1 Like
Thanks @manfred_jr, happy to hear that! Hope you love our project and show your interest on us in the AI track.
For the scalability and performance when going to on cloud computation or real time processes, it really depends on the computation resources of the virtual machine which utilise by the game provider, and not us. However, our deep learning pipeline do designed for optimum performance during the computation process itself and also made to host on an AWS instances and pixel streaming optimize.
Blend itself don’t monetize, but the blockchain game provider do can benefit a lot by using Blend as the game can be level up to unpredictable level in the terms of graphics quality and immersiveness. Our research team set this project as a community open source project and welcome any talented one to join the development or research tasks. So in the long run, our work are mainly supported by fund source like hackathon (this time), grants, or any open source contributions in the form of donations from different parties (but not limited to).
1 Like
Oh there you go again with the really detailed response, thanks for replying and for providing insights into scalability and performance aspects, clearly understood and I really do appreciate.
I did miss a very important question bordering on education initiatives.
Are there plans to organize workshops, webinars, or other educational programs to educate both developers and potential users about Blend’s capabilities and how they can leverage them effectively?
That will definitely be a thing to arrange in the future. However, for this hackathon, our team will focusing on AI-related researching task and development more, and hope to speed up the progress for this project public release. But thanks for your recommendation and it will be on our list. Thanks @manfred_jr
1 Like
You’re welcome, thanks once again for replying to every inquiry, goodluck on your buidl.
Hey everyone, the demo video for Blend has finally been uploaded on YouTube ~ We spent a lot of time preparing this video to explain the concept and idea of Blend, and also demonstrate how Blend can be integrated into your next Tron blockchain game. As Blend is built with complex AI technology, our team knows that the video will be a bit lengthy, but you definitely can learn a lot from our project.
Let me know in the comment section how you think after watching it~ Enjoy!
1 Like
alright I will check it out
Oh wow, checking it out in a bit.
The thumbnail has this description; “Tron’s Ultra Realistic Game Character
3D Face Reconstruction Engine
using Molecular Depth Estimation and Hybrid CNN Deep Learning Architecture with Real-Time Post Skeletal Mesh Fusion” can you break this down to noob language haha
And first few seconds of the video my eyes did a typo by thinking that was "Studio Universal’ had to look again over and over to discover that was actually "Unreal"
Haha @manfred_jr , it’s Unreal Engine~ Thanks for watching the demo video. The title in the thumbnail is the title of this research and development project, hope you can understand. But we do explain about what are the technical terms are in a breakdown way (if you watch through this video)
1 Like
I thought it was complex machine learning type language lol.
1 Like
Greetings @blend ! Your project seems to be missing details on:
Any smart contract links:
Project Website:
Project Milestones:
Please add them to your project. Thank you!