Simple triangle example doesn't work (Mesh culling func misses)

[JerryYan97]

Hello! What should I do if I also want to keep the back face of a dynamic triangle?

https://github.com/taichi-dev/taichi_three/blob/master/examples/simple_triangle.py

[archibate]

Thank for asking, you may use the newly added t3.MeshNoCulling node. It was not available in v0.0.9 yet. To use it:

1. you may clone this repo and install it manually to get this latest update (recommended):

python3 -m pip uninstall -y taichi_three  # remove the old version
python3 -m pip uninstall -y taichi_three  # double uninstall for sure
git clone https://github.com/taichi-dev/taichi_three.git --depth=1
cd taichi_three
python3 -m pip install wheel  # required for building wheel
python3 setup.py bdist_wheel  # create the wheel file
python3 -m pip install dist/*.whl  # install the latest version

2. copy and paste the source of t3.MeshNoCulling locally in your file:

import taichi_three as t3
import taichi as ti

@ti.data_oriented
class MeshNoCulling:  # the same one as taichi_three/model.py have in latest master branch
    def __init__(self, mesh):
        self.mesh = mesh

    @property
    def shape(self):
        return self.mesh.shape

    @property
    def static_shape(self):
        return [2, *self.mesh.static_shape]

    def before_rendering(self):
        self.mesh.before_rendering()

    @ti.func
    def get_face(self, i, j: ti.template()):
        face = self.mesh.get_face(i, t3.ts.vec(*[j[_] for _ in range(1, j.n)]))
        ret = t3.DataOriented()
        if ti.static(j.x == 0):
            ret.__dict__.update(
                pos = [face.pos[i] for i in [0, 1, 2]],
                tex = [face.tex[i] for i in [0, 1, 2]],
                nrm = [face.nrm[i] for i in [0, 1, 2]],
            )
        else:
            ret.__dict__.update(
                pos = [face.pos[i] for i in [0, 2, 1]],
                tex = [face.tex[i] for i in [0, 2, 1]],
                nrm = [face.nrm[i] for i in [0, 2, 1]],
            )
        return ret



N = 64

scene = t3.Scene()
mesh = t3.DynamicMesh(n_faces=N, n_pos=N + 1)
model = t3.Model(MeshNoCulling(mesh))  # here
scene.add_model(model)
camera = t3.Camera()
scene.add_camera(camera)
light = t3.AmbientLight(1.0)
scene.add_light(light)


@ti.materialize_callback
@ti.kernel
def init_mesh():
    mesh.nrm[0] = [0, 0, 1]
    mesh.pos[N] = [0, 0, 0]
    for i in range(N):
        a = i / N * t3.tau
        mesh.pos[i] = [ti.sin(a), ti.cos(a), 0]
        mesh.faces[i] = [[i, 0, 0], [(i + 1) % N, 0, 0], [N, 0, 0]]


gui = ti.GUI('Dynamic', camera.res)
while gui.running:
    gui.get_event(None)
    camera.from_mouse(gui)
    mesh.n_faces[None] = gui.frame % N
    scene.render()
    gui.set_image(camera.img)
    gui.show()

3. Or, if you don't mind - create two faces with different 'orientation' and normal will make it visible from both side.