How to use this viewer on self trained scenes
Cholezlh opened this issue · comments
hi,thanks for your work,it's very interesting.but if I have trained a new scene through mobile nerf,how can i import it to use your unity viewer?
Hey! Pretty cool that you've managed to train your own data! It's currently not easy to import your own files, but it should not be too difficult to add that.
I can do it, but I would need some help with what the output looks like for you. Did you use the official code to train the MobileNeRF?
Would it be possible for you to send me the output files? (I understand if that's not possible)
Imho the easiest way would be if we agree to put all the assets in a folder and use the same naming convention that the demo scenes use:
- OBJ files with naming convention: shapeX_Y.obj
- multiple PNG files with naming convention: shapeX.pngfeatZ.png (where Z goes from 0 to 1, so 2 .pngs per shape)
- The weights of the MLP in a file called mlp.json
Then I can implement a menu that lets you choose a folder with source assets, which will automatically import all these.
Let me know if that looks god to you.
I created a branch Custom_Data
It has a new menu: MobileNeRF -> Import from disk
Feel free to try it and report any issues that you might find.
sorry for late relay.this is trained materials through official mobileNeRF, https://drive.google.com/file/d/1R3-vHWrLZEauaX7bGWXwJ4yFLdSRxuME/view?usp=sharing, and the data is from mipnerf, it's a image collection of a room.I will try your new code, thanks very mush.
Hey, thanks for sharing the data. That was really helpful!
It seems there are cases when the individual shapes are not split into several .obj files like in the demo scenes. These shape files then have the naming convention: shapeX.obj (X being the total number of objects specified in the mlp.json)
I just pushed some changes to make this work. After making these, I was able to import your scene in Unity.
Would you mind if I posted a small video about this on Twitter? I can tag you if you want.
I use your new pushed code, It worked very good.Thanks for your work again. just feel free to post the video :)