# Creating New Textures from Scratch¶

## The PNMImage Class¶

This class is how Panda3D handles regular images (.gif, .jpg, and the like). This class allows you to manipulate individual pixels of the image. You can load existing images using the function read(fileName) where filename is the path to the image file (in Panda Filename Syntax) wrapped in a Filename object. Or, you can create a brand new image from scratch, by passing the x, y size to the constructor.

PNMImage my_image;

PNMImage my_empty_image(256, 256);


You can get the size of the image you have read using the getXSize() and getYSize() functions. Although you cannot change the size of an image directly, you can rescale an image by filtering it into a larger or smaller PNMImage:

PNMImage full_size(Filename("testImg.png"));
PNMImage reduced(256, 256);
reduced.gaussianFilterFrom(1.0, full_size);


You can get individual RGB values using the getRed(x,y), getGreen(x,y), getBlue(x,y) or getRedVal(x,y), getGreenVal(x,y), getBlueVal(x,y) where x and y tell what pixel to look at (lower-left is 0,0 upper right is getXSize()-1, getYSize()-1 The difference between these functions is that the get*Val functions return a number between 0 and 1 while the get* functions return their value as an integer. For example, if your image uses 8-bit color calling getGreenVal on a green pixel will return 255 and calling getGreen will return 1. You can also get all the RGB information at the same time using getXel(x,y) and getXelVal(x,y) which return a 3-component vector with red in the x, green in the y, and blue in the z.

# The pixel at 0,0 is red and we're using 8-bit color
myImage.getRedVal(0,0) # Returns 255
myImage.getRed(0,0) # Returns 1

colors = myImage.getXelVal(0,0) # Returns (255,0,0)
colorVal = myImage.getXel(0,0) # Returns (1,0,0)


The functions for setting pixel information are setRed(x,y,value), setGreen(x,y, value), setBlue(x,y, value) or setRedVal(x,y,value), setGreenVal(x,y, value), setBlueVal(x,y, value). There is still the same dichotomy as above when it comes to regular sets and using setvals. You can also use setXel(x,y,colorVec) and setXelVal(x,y, colorVec). You can also fill an image with a color by using fill(r,g,b) and fillVal(r,g,b).

myImage.setGreenVal(0, 0, 255) # If pixel (0,0) was red before, now it is yellow
myImage.setBlue(0, 0, 1) # Pixel (0,0) is now white

gray = Vec3(0.5, 0.5, 0.5)

# Both of these set the origin to gray
myImage.setXelVal(0, 0, gray * 255)
myImage.setXel(0, 0, gray)

# Makes every pixel red
myImage.fillVal(255, 0, 0)
# Makes every pixel green
myImage.fill(0, 1, 0)


There are also gets and sets for the alpha channel using the same interface as above. However, if you use them on an image that doesn’t have an alpha channel you will cause a crash. To see if an image has an alpha channel use hasAlpha() which returns True if there is an alpha channel and False otherwise. You can add an alpha channel using addAlpha(). You can also remove it using removeAlpha().

You can also make an image grayscale by using makeGrayscale(). You can now use sets and gets for Gray too. Using getGray* on a color image just returns the value in the blue channel. If you want to get the grayscale value of a pixel regardless of whether the image is a grayscale or a color image, you can use getBright(x,y), which works equally well on color or on grayscale images. If you want to weight the colors use getBright(x,y, r,g,b) where r,g,b are the weights for their respective channels.

There are several other useful functions in the class, which are described on the PNMImage page in the API Reference.

## Getting the Image of a Texture¶

The Panda Texture class does not allow for pixel manipulation. However the PNMImage class below does. Therefore, if you want to change the image in a Texture object you must call its store(myImage) which saves the image of the texture into myImage.

myImage = PNMImage()

# After this call, myImage now holds the same image as the texture
myTexture.store(myImage)


Once you have changed all the data in the image you can now load it into a texture using the Texture objects load(myImage) function, where myImage is the PNMImage to make the texture from.
// Assume we already have myImage which is our modified PNMImage

Remember however, that most graphics cards require that the dimensions of texture have to be a power of two. PNMImage does not have this restriction and Panda will not automatically scale the image when you put it into a texture.