There is no complete text2scene pipeline.
#4
by nikitavovch - opened
You cannot generate just from the text description + GT layout, as it says. You'll always need atleast RGB images to inference the text 2 scene pipeline. I dont get it, why you haven't made all in one script. I mean generation of rgb, depth, normal and etc. images to inference the full text to scene pipeline in one click. To inference your own generation from only a GT layout, first of all you need to run the preprocessing script, then you need to generate rgb's by yourself, with their flux wireframe model, and etc etc etc...