1
votes

I copied a code of YouTube, about displaying 3d cubes on a screen in Python, without the use of external modules (like PyOpenGL). It works fine, but the moment you go between two cubes, the display gets messed up. Here is my code:

import pygame, sys, math, random

def rotate2d(pos, rad): x,y=pos; s,c = math.sin(rad),math.cos(rad); return x*c-y*s,y*c+x*s

class Cam:
    def __init__(self, pos=(0,0,0),rot=(0,0)):
        self.pos = list(pos)
        self.rot = list(rot)

    def events(self, event):
        if event.type == pygame.MOUSEMOTION:
            x, y = event.rel; x/=200; y/=200
            self.rot[0]+=y; self.rot[1]+=x

    def update(self, dt, key):
        s = dt*10

        if key[pygame.K_q]: self.pos[1]+=s
        if key[pygame.K_e]: self.pos[1]-=s

        x,y = s*math.sin(self.rot[1]),s*math.cos(self.rot[1])
        if key[pygame.K_w]: self.pos[0]+=x; self.pos[2]+=y
        if key[pygame.K_s]: self.pos[0]-=x; self.pos[2]-=y
        if key[pygame.K_a]: self.pos[0]-=y; self.pos[2]+=x
        if key[pygame.K_d]: self.pos[0]+=y; self.pos[2]-=x
        if key[pygame.K_r]: self.pos[0]=0; self.pos[1]=0;\
                                         self.pos[2]=-5; self.rot[0]=0; self.rot[1]=0


class Cube:
    faces = (0,1,2,3),(4,5,6,7),(0,1,5,4),(2,3,7,6),(0,3,7,4),(1,2,6,5)
    colors = (255,0,0),(255,128,0),(255,255,0),(255,255,255),(0,0,255),(0,255,0)
    def __init__(self,pos=(0,0,0),color=None,v0=(-1,-1,-1),v1=(1,-1,-1),v2=(1,1,-1),v3=(-1,1,-1),v4=(-1,-1,1),v5=(1,-1,1),v6=(1,1,1),v7=(-1,1,1)):
        if color != None:
            if len(color) == 3 or len(color) == 4:
                self.colors = tuple((color for i in range(6)))
            if len(color) == 6:
                self.colors = color
        self.vertices = (v0,v1,v2,v3,v4,v5,v6,v7)
        x,y,z = pos
        self.verts = [(x+X/2,y+Y/2,z+Z/2) for X,Y,Z in self.vertices]


pygame.init()
w,h = 400,400; cx,cy = w//2, h//2
screen = pygame.display.set_mode((w,h))
clock = pygame.time.Clock()
cam = Cam((0,0,-5))
pygame.event.get(); pygame.mouse.get_rel()
pygame.mouse.set_visible(0); pygame.event.set_grab(1)
occupied = []

cube1 = Cube((0,0,0))
cube2 = Cube((0,0,2))
objects = [cube1, cube2]

while True:
    dt = clock.tick()/1000

    for event in pygame.event.get():
        if event.type == pygame.QUIT: pygame.quit(); sys.exit()
        if event.type == pygame.KEYDOWN:
            if event.key == pygame.K_ESCAPE: pygame.quit(); sys.exit()
        cam.events(event)

    screen.fill((0,0,0))

    face_list = []; face_color = []; depth = []

    for obj in objects:

        vert_list = []; screen_coords = []
        for x,y,z in obj.verts:
            x-= cam.pos[0]; y-=cam.pos[1];z-=cam.pos[2]
            x,z = rotate2d((x,z),cam.rot[1])
            y,z = rotate2d((y,z),cam.rot[0])
            vert_list += [(x,y,z)]

            f = 200/z
            x,y = x*f,y*f
            screen_coords+=[(cx+int(x),cy+int(y))]

        for f in range(len(obj.faces)):
            face = obj.faces[f]

            on_screen = False
            for i in face:
                x,y = screen_coords[i]
                if vert_list[i][2]>0 and x>0 and x<w and y>0 and y<h: on_screen = True; break

            if on_screen:
                coords = [screen_coords[i] for i in face]
                face_list += [coords]
                face_color += [obj.colors[f]]
                depth += [sum(sum(vert_list[j][i]**2 for i in range(3)) for j in face) / len(face)]

    order = sorted(range(len(face_list)),key=lambda i:depth[i],reverse=1)

    for i in order:
        try: pygame.draw.polygon(screen,face_color[i],face_list[i])
        except: pass

    key = pygame.key.get_pressed()
    pygame.display.flip()
    cam.update(dt,key)

And here is what happens when you try and navigate between the two cubes: Example of error

Can someone please suggest an edit to the code that would stop the cubes from becoming distorted when the cam.pos is close to the cubes?

1
Is the camera inside the cubes? - Kingsley
No, I made sure of it. It is between cube1 and cube2 - User 12692182
Although, a function with inside-cube display would be nice, it is not necessary, as the player dies the moment he/she goes inside a cube - User 12692182

1 Answers

1
votes

The application does not correctly draw the geometry, when apart of a faces (primitive, side of a cube) is behind and the other part in front of the eye position. That happens if the transformed z coordinate (vert_list += [(x,y,z)]) is positive for the some vertices and negative for negative for some other vertices that form primitive (face).

You can test that behavior with ease, if you skip all the faces, where at least one z coordinate is negative (behind the eye):

while True:
    # [...]

    for obj in objects:
        # [...]

        for f in range(len(obj.faces)):
            face = obj.faces[f]

            #on_screen = False
            #for i in face:
            #    x,y = screen_coords[i]
            #    if vert_list[i][2]>0 and x>0 and x<w and y>0 and y<h: on_screen = True; break

            # draw a face if any projected coordinate (x, y) is in the viewing volume 
            on_screen = False
            for i in face:
                x,y = screen_coords[i]
                if x>0 and x<w and y>0 and y<h: on_screen = True; break
            # skip a face if NOT ALL z coordinates are positive
            if on_screen:
                on_screen = all([vert_list[i][2]>0 for i in face])

            if on_screen:
                # [...]


The issue can be solved by clipping the geometry at hypothetical near plane. See Viewing frustum:

while True:
    # [...]

    for obj in objects:
        # [...]

        for f in range(len(obj.faces)):
            face = obj.faces[f]

            on_screen = False
            for i in face:
                x,y = screen_coords[i]
                if vert_list[i][2]>0 and x>0 and x<w and y>0 and y<h: on_screen = True; break

            # clip geometry at near plane
            if on_screen:
                near = 0.01
                for i in face:
                    if vert_list[i][2]<0:
                        x, y, z = vert_list[i]
                        nearscale = 200/near
                        x,y = x*nearscale,y*nearscale
                        screen_coords[i] = (cx+int(x),cy+int(y))

            if on_screen:
                coords = [screen_coords[i] for i in face]
                face_list += [coords]
                face_color += [obj.colors[f]]
                depth += [sum(sum(vert_list[j][i]**2 for i in range(3)) for j in face) / len(face)]