Use Numpy array for panda3d texture, video player example

numpy_texture_panda3d.py

from __future__ import division
import os


import cv2
import numpy as np

from panda3d.core import Texture, CardMaker
from direct.showbase.ShowBase import ShowBase


fp = 'path/to/video.avi'

# use opencv to read from avi
if os.path.exists(fp):
    cap = cv2.VideoCapture(fp)
    success, frame = cap.read()
    h, w, depth = frame.shape # source is RGB video but will be used as gray scale image
else: # if no input, generate random noise image
    h, w = 480, 640
    cap = None

# setup panda3d scripting env (render, taskMgr, camera etc)
base = ShowBase()

# set up a texture for (h by w) 8 bit gray scale image
tex = Texture()
tex.setup2dTexture(h, w, Texture.T_unsigned_byte, Texture.F_luminance)

# set up a card to apply the numpy texture
cm = CardMaker('card')
card = render.attachNewNode(cm.generate())
card.setPos(-0.5, 1.5, -w/h/2) # bring it to center, put it in front of camera
card.setScale(1,1,w/h) # card is square, rescale to the original image aspect


def updateTex(task):
    if cap: # video input
        success, frame = cap.read()
    else:  # random noise
        frame = np.random.randint(0,255,(h,w,3)).astype(np.uint8)

    buf = frame[:,:,0].T.tostring() # slice RGB to gray scale, transpose 90 degree, convert to text buffer
    tex.setRamImage(buf) # overwriting the memory with new buffer
    card.setTexture(tex) # now apply it to the card
    
    return task.cont

taskMgr.add(updateTex, 'video frame update')

base.run()