THuman Dataset preprocess

[mmmcn]

Hi, I found the program was running so slowly when I ran bash render_batch.sh debug all, I figured out it stopped at hits = mesh.ray.intersects_any (origins + delta * normals, vectors), and the number of rays is the millions, is that the reason why it was too slow?

[YuliangXiu]

Pyembree is required for computing ray intersection inside trimesh (https://trimsh.org/), please uninstall and reinstall pyembree and trimesh, and then it should work well.

Same issue in PIFu

[mmmcn]

thx，now intersects_any() works fine, but I ran into another problem:

-> rndr = PRTRender(width=size, height=size, ms_rate=16, egl=egl)
(Pdb) c
Traceback (most recent call last):
  File "./render_single.py", line 125, in <module>
    rndr = PRTRender(width=size, height=size, ms_rate=16, egl=egl)
  File "/home/private/project/3d-holographic-reconstruction/ICON/scripts/./../lib/renderer/gl/prt_render.py", line 35, in __init__
    CamRender.__init__(self,
  File "/home/private/project/3d-holographic-reconstruction/ICON/scripts/./../lib/renderer/gl/cam_render.py", line 32, in __init__
    Render.__init__(self,
  File "/home/private/project/3d-holographic-reconstruction/ICON/scripts/./../lib/renderer/gl/render.py", line 96, in __init__
    glTexImage2DMultisample(GL_TEXTURE_2D_MULTISAMPLE,
  File "/home/private/Software/anaconda3/envs/icon/lib/python3.8/site-packages/OpenGL/platform/baseplatform.py", line 415, in __call__
    return self( *args, **named )
  File "src/errorchecker.pyx", line 58, in OpenGL_accelerate.errorchecker._ErrorChecker.glCheckError
OpenGL.error.GLError: GLError(
	err = 1282,
	description = b'invalid operation',
	baseOperation = glTexImage2DMultisample,
	cArguments = (
		GL_TEXTURE_2D_MULTISAMPLE,
		16,
		GL_RGBA32F,
		512,
		512,
		GL_TRUE,
	)
)

Is PyOpenGL version or something else? my PyOpenGL==3.1.6 and PyOpenGL_accelerate==3.1.5

[YuliangXiu]

#60

[mmmcn]

Sorry to bother you again, after running bash render_batch.sh debug all successfully, I got results below,

Snapshot on mesh	Normal(SMPL, Front)	Normal(Front)	RGB Image

I am confused why RGB image and normal image are incorrect

[YuliangXiu]

Could you send me the ID of this mesh?

[mmmcn]

0028， I've tried 0028 and 0000 actually, but both of them can not got reasonable results

[mmmcn]

By the way, I use headless mode.

[YuliangXiu]

I checked my rendering results, and they are correct. These kinds of artifacts are often caused by the disordered face topology, and should be resolved by

mesh = trimesh.load(mesh_file, skip_materials=True, process=False, maintain_order=True, force='mesh').

BTW, I am using trimesh==3.9.35

[mmmcn]

Thank you so much for your reply! everything worked fine after I rebooted the mechine and ran the script again, though I'm not sure what caused this problem.

[mmmcn]

BTW, is there a way to get depth images that correspond to the RGB images?

[YuliangXiu]

set depth=True in render_single.py, I just updated the dataset readme.