Matplotlib render

render 3D mesh with matplotlib library
I made this just for fun, so the code is a bit messy and there are things that still needs to be done too!

Reference: https://matplotlib.org/matplotblog/posts/custom-3d-engine/

TODOs

• Clean up duplicated codes..!!!
• Image
  • curling based on normal
  • rendering mesh difference (e.g. MSE ...)
  • gouraud rendering with matplotlib.tri
  • There is some mistake in depth sorting triangles
• Animation
  • add slider feature to tour around the animation frames
  • add feature for comparing difference between two animation
  • interactive ui

Requirements

pip install -r requirements.txt
pip install .

NEW!

render with gouraud shading!
also supports visualizing difference/error

from matplotrender import *
import trimesh

mesh0 = trimesh.load('your_mesh_file0.obj')
mesh1 = trimesh.load('your_mesh_file1.obj')

v_list = [mesh1.vertices, mesh2.vertices]
f_list = [mesh1.faces, mesh2.faces]
rot_list=[[0,0,0] ,  [0,0,0]] # you can control rotation for individual mesh

plot_mesh_gouraud(
    v_list, 
    f_list, 
    is_diff=True, 
    diff_base=mesh2.vertices, # calculates difference based on this mesh
    rot_list=rot_list,
)

The meshes will be rendered from left to right (mesh1, mesh2)

How to use

from matplotrender import *
import trimesh
mesh = trimesh.load('your_mesh_file.obj')
print(mesh.vertices.shape)
>>> (5023, 3)

# figure size
SIZE = 2

# mesh that you wanna render
v_list=[ mesh.vertices ]
f_list=[ mesh.faces ]

# xyz Euler angle to rotate the mesh
rot_list=[ [0,0,0] ]

plot_mesh_image(v_list, f_list, rot_list=rot_list, size=SIZE, mode='mesh') # default
plot_mesh_image(v_list, f_list, rot_list=rot_list, size=SIZE, mode='normal')
plot_mesh_image(v_list, f_list, rot_list=rot_list, size=SIZE, mode='shade')

You can also render video!
let's say you have an mesh animation saved as a npy file as below:

# your mesh animation
vertices_anim = np.load('your_mesh_animation.npy')
print(vertices_anim)
>>> (100, 5023, 3)

you can use plot_mesh_video to render a video.

v_list=[ vertices_anim ]
f_list=[ mesh.faces ]
plot_mesh_video(
        v_list, 
        f_list, 
        size=2, 
        bg_black=False,
    )

you can also mux audio too!

audio_dir='your_path/audio.wav'
v_list=[ vertices_anim ]
f_list=[ mesh.faces ]
plot_mesh_video(
        v_list, 
        f_list, 
        size=2, 
        bg_black=False,
        audio_dir=audio_dir
    )

Rendering difference (errors)

you can render L1 error between meshes (only for same topology)

from matplotrender import *
import trimesh

mesh0 = trimesh.load('your_mesh_file0.obj')
mesh1 = trimesh.load('your_mesh_file1.obj')
mesh2 = trimesh.load('your_mesh_file2.obj')
mesh3 = trimesh.load('your_mesh_file3.obj')

v_list = [mesh1.vertices, mesh2.vertices, mesh3.vertices]
f_list = [mesh1.faces, mesh2.faces, mesh3.faces]
v0 = mesh0.vertices # (V, 3)

plot_image_array_diff3(
    v_list, 
    f_list, 
    v0,
    rot=[0,-10,0], 
    bg_black=False,
)

The meshes will be rendered from left to right(mesh0, mesh1, mesh2, mesh3)