search

rgl-epfl / differentiable-sdf-rendering

Source code for "Differentiable Signed Distance Function Rendering" (Siggraph 2022)
BSD 3-Clause "New" or "Revised" License
861 stars 53 forks source link

Local compiling issue #24

Open Start1er opened 5 days ago

Start1er commented 5 days ago

Hi, thanks for your nice work. I compile locally in windows with mitsuba 3.4.1 and drjit 0.4.4 and then compile the fastsweep. But I have met the issue when I run optimize.py

    sdf_scene = mi.load_file(ref_scene_name, shape_file='dummysdf.xml', sdf_filename=join(SCENE_DIR, 'sdfs', 'bunny_64.vol'),
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: ​[xml.cpp:1119] Error while loading "C:\research\method\diff-sdf\differentiable-sdf-rendering\scenes\dragon\...\integrator_sdf.xml" (near line 5, col 3): could not instantiate integrator plugin of type "sdf_direct_reparam": TypeError: redistance(): incompatible function arguments. The following argument types are supported:
[xml.cpp:1119]     1. (arg0: drjit::Tensor<drjit::CUDAArray<float> >) -> drjit::Tensor<drjit::CUDAArray<float> >
[xml.cpp:1119]     2. (arg0: drjit::Tensor<drjit::LLVMArray<float> >) -> drjit::Tensor<drjit::LLVMArray<float> >

Seems like the program can not find the sdf integrator. Any response would be helpful!

Start1er commented 4 days ago

I compile mitsuba and drjit for specific 3.4.1 and 0.4.4 version in my local enviroment, then compile the fastsweep, seems like it will in defualt update drjit to 0.4.6. And the data type mismatch happen as your readme saying. Any thing I can do for compiling them one by one?