RF-Genesis icon indicating copy to clipboard operation
RF-Genesis copied to clipboard

Questions about pathtrarcer.py

Open fu1581112 opened this issue 6 months ago • 1 comments

Thank you for your excellent work!I successfully ran the previous version of this code half a year ago, but recently when I ran it again, problems occurred. The terminal displayed an error,Critical Dr.Jit compiler failure: cuda_check(): API error 0718,However, I solved it by upgrading mistsub from 3.5.2 to 3.6.0 and drjit from 0.4.6 to 1.0.1, but the following is the key point.

python run.py -o "a person walking back and forth" -e "a living room" -n "hello_rfgen" [RFGen]` Step 1/4: Already done, existing body motion file, skiping this step. [RFGen] Step 2/4: Rendering the human body PIRs: Traceback (most recent call last): File "D:\RF-Genesis\run.py", line 92, in main() File "D:\RF-Genesis\run.py", line 49, in main body_pir, body_aux = pathtracer.trace(os.path.join("../",output_dir, 'obj_diff.npz')) File "C:\Users\MY.conda\envs\rfgen\lib\site-packages\genesis\raytracing\pathtracer.py", line 160, in trace raytracer = RayTracer() main() File "D:\RF-Genesis\run.py", line 49, in main body_pir, body_aux = pathtracer.trace(os.path.join("../",output_dir, 'obj_diff.npz')) File "C:\Users\MY.conda\envs\rfgen\lib\site-packages\genesis\raytracing\pathtracer.py", line 160, in trace raytracer = RayTracer() File "C:\Users\MY.conda\envs\rfgen\lib\site-packages\genesis\raytracing\pathtracer.py", line 15, in init self.scene = mi.load_dict(get_deafult_scene(res = self.PIR_resolution)) File "C:\Users\MY.conda\envs\rfgen\lib\site-packages\genesis\raytracing\pathtracer.py", line 95, in get_deafult_scene 'to_world': T.look_at( TypeError: look_at(): incompatible function arguments. The following argument types are supported: 1. look_at(self, origin: mitsuba.Point3f, target: mitsuba.Point3f, up: mitsuba.Point3f) -> mitsuba.Transform4f

Invoked with types: kwargs = { origin: tuple, target: tuple, up: tuple }

The content of some specific codes is:

sensor = mi.load_dict({ 'type': 'perspective', 'to_world': T.look_at( origin=(0, 1, 3), target=(0, 1, 0), up=(0, 1, 0) ), I'm not sure what caused this. May I ask if you think it's due to the configuration environment or the code? Or other suggestions that can help solve the problem are also acceptable,thanks!

fu1581112 avatar Jul 14 '25 06:07 fu1581112