LaidBackEnergy

Inspiration

Inspiration for this project comes from our friends who are developing games and are into robotics. We found out most of the time, they suffer from lack of initialization of a model. For e.g. Making a robot arm in CAD software requires 7 objects and the task is to make them separate and then put them in a file together , but seeing them stuck at point blank was something that actually forced us to think about it. So, we came up with the plan to develop an "Agentic AI", which can actually help developers to start in a specific direction rather than getting stuck at point in current time frame.

What it does

Our model tends to develop basic shapes fetches them from backend and give it to the developers, basic shapes such as donut, table , chair , mechanical arm frames, Axl frames etc. These are few of the example we tend to show and rather than that we would also be giving out advices and suggestions to the beginners or professionals how to go for a shape or model and how to make it, according to the queries asked. Generally learning a 3-D software is so tough it could time from months to years and so on. What we want to do is , make it easy for newbies to learn that software efficiently and professionally. Market of 3-D , VFX and game development are in the Buzz, every other movie or game heavily rely on graphics and for that on models and other elements such as rigging, if we tackle up these things for the indie developer or professional/ industry personnel, it will bring better results in productivity. Because better productivity yearns better result.

How we built it

We would like to use Watsonx.ai as a base/ main frame, then we will connect it to Watsonx Orchestrate ADK which would act as task manager that connects it to backend, that consists of basic 3-D models such as props, table , mechanical robotic arm etc. Then, we will connect it to the LLMs and stuff like creative skills library that holds basic 3D models. So, how exactly it works, firstly user would put out the query in the type-box for e.g. "Make a giraffe model" it will take that query to the backend where it build a very basic block by block shape of giraffe and will guide to make it more realistic and would suggest various different videos from trusted sites. Also for basic stuff like tables and chairs and pre-build models like robotic arm and axle design etc, it will provide them as a base model in .obj, .fbx format, and accordingly what direction user would like to go in, it will guide them step-by-step. “Game devs can whip up NPC models fast, while students developing robots get pre-built arm frames to tweak.”

Challenges we ran into

Integrations with Existing 3D Software; Training or fine-tuning our AI models requires high-quality, diverse 3D datasets. While automation helps speed up workflows, 3D modeling demands significant artistic skill and creativity. So, balancing Creativity with automation is a different type of hustle. 3D modelling AI often requires significant computational resources for generating, validating, and rendering models. Balancing performance with accessibility for users on less powerful hardware or internet connections is a challenge. Also, using AI-generated assets involves questions about copyright, authorship, and responsible use. Ensuring our AI agent respects these norms and provides transparent outputs is important. Other than that, User Experience and Accessibility such as understanding ambiguous and partial information is needed.

Accomplishments that we're proud of

First of all, planning and getting a platform this big to represent our idea is a big accomplishment in itself. Secondly, ideation and our mission is perfectly focused in one direction, we ain't focusing on everything at once, so that we could grow more in future. Getting access to a base ADK to develop our own agent is an accomplishment too. We were more proud if the idea would be executed perfectly, and would fetch the result and help other amateur 3-D devs.

What we learned

Basic and Intermediate understanding of 3D software such as Blender, Houdini, 3DS max, Maya etc. Learning Machine Learning and Deep learning, understanding LLMs, Javascript/typescript, basic C#. Also how the Agentic AI is made and used and maintained.

What's next for LaidBackEnergy

After this Hackathon we want to move into the field of blockchain and develop our own coin and stuff, which is in fact one of the biggest financial decision we would undergo. As technology rapidly changing from day -to-day we've to keep adapting newest technology available in the market and through the current scenario , we can understand the market better, newer protocols make the data security it's own turf , and in a world where data is key, we would like to focus on BLOCKCHAIN.