Skip to content

Latest commit

 

History

History
23 lines (21 loc) · 1.81 KB

File metadata and controls

23 lines (21 loc) · 1.81 KB

Report October 9th 2023

Quick note/summary:

I looked into LumaAI, read about Gaussian Splatting. Also created a model with LumaAI

Activities/Went through:

  • LumaAI:
    • Downloaded and set up LumaAI for my mobile device. and looked into the following tools:
      • Video to 3D API data: started working on the tutorial. As of now, I am setting up and figuring our how to run it. Idea is to run a video through it and then have a model. Note: it is $1 per capture. I also want to note that based on the models that I looked at, these NeRF models are similar enough in results as good photogrammetric models.
      • Looked briefly into the exporting formats for this app.
      • There is another application that is implemented with this called Flythroughs.
  • Gaussian Splatting:
    • Read briefly through the abstact.
    • watched a video of guys working with Gaussian Splatting:
      • https://www.youtube.com/watch?v=juRMRej2d5c, not this video is just a reaction of some testing these guys ran.
      • it seems that this is the newest technology. To an extent, this might be better than LumaAI(they use NeRF).
      • it also seems this is nicer with iffy photography or inputs. It seems we can get point clouds too(still have to check if we can use them in UE5).

Moving Forward:

  • Get a model going with LumaAI's video to 3D API data feature. Get the code going, and then get a video of an object to get going.
  • Get a model running with LumaAI's application(it is rendeing and preprocessing as of October 9th.
  • Look into interactive scenes here: https://lumalabs.ai/interactive-scenes
  • Either this week or the one after, I want to look at Gaussian Splatting and potentially run some tests on it. As of right now, this seems very promising.
  • Access to the lab would be amazing to start running the above!