Open simin75simin opened 2 years ago
The DL methods available at the moment that are actually good and available are most of the time licensed quite restrictive and/or the code is difficult to setup. I find it curious that, there is no ready to use executable to drop a few images in and get a model out so far. Yes, there are Colab notebooks, but if you want to run the code on your own computer you need to spend a considerable time to configure the environment. I tried to build some projects using pyinstaller to wrap all the dependencies up, but sadly I don´t have the time to configure my projects. Being able to test out some of the existing models and use them within Meshroom would be great and unique. If that is something you are considering working on, I could give you some pointers to promising projects that should not be to difficult (relatively speaking) to connect to Meshroom.
If that is something you are considering working on, I could give you some pointers to promising projects that should not be to difficult (relatively speaking) to connect to Meshroom.
yeah i'd like to know could you show me some that will not slow down the runtime too much? thanks!
You may find this interesting: https://peterfalkingham.com/2022/04/26/neural-radiance-fields-nerf-and-instant-ngp-future-of-photogrammetry/
https://github.com/ventusff/neurecon
There are some projects that are OpenMVG (and therefore nearly Meshroom) compatible like https://github.com/AIBluefisher/EGSfM
There are a few more recent one, but I need to look them up.
thanks for answering.