junshutang / junshutang.github.io Goto Github PK

Hi @junshutang
I find your new text/2D image to 3D quite impressive https://github.com/junshutang/Make-It-3D
It seems you don't have enabled issues nor discussions on this repository to I am making one here..

So I am quite interested in text-to-3D, which you can follow the recent development here: https://paperswithcode.com/task/text-to-3d
One of the founding paper as you probably knows is https://github.com/ashawkey/stable-dreamfusion and Make-it-3D seems to have better resolution/accuracy,
however what you might not have heard of is the recent evolution from Nvidia, Magic3D which is faster and much more accurate. Therefore I am wondering wether you could integrate/adapt some of the ideas from Magic3D into your software in order to improve the state of the art.