The buffer consists of a main bitmap (an array of data) that contains your working image, a swap bitmap (or swap buffer) and an alpha channel.  The alpha channel is an 8 bit bitmap that can contain a matte or mask that can be used like a template or selection.

The swap buffer can be useful in many ways.  You can use it to store a spare copy of an image, or to store a checkpoint for your work.  You can rub one buffer through to the other, or you can use it with some filters to displace, merge, composite, or generally modify your main buffer.

To swap buffers, press the j key.

Pressing the J (capital j) key will copy the current buffer to the swap buffer.

Swap buffers swaps the Main and Swap buffer
Copy to swap copies the current buffer to the swap buffer
Copy swap buffer to alpha copies the swap buffer to the Alpha channel.
Copy alpha to swap buffer copies the alpha channel to the swap buffer.

Generally since PD is a dedicated painting application, you work on one image at a time, however should you wish, you can store a buffer or a number of buffers to access at any time.

Replace image replaces the current buffer with the one that was stored.  If your buffer has changed size, it will be cropped into the new size.

The combine menu lets you combine the stored image with the current image in a number of different ways.

Replace alpha replaces the current alpha with the one that was stored with the buffer.

Use image as alpha converts the stored buffer to a grayscale image and uses it as the alpha channel.

Replace existing buffer replaces the current buffer with the stored buffer, obliterating whatever you are working on.  If needed, the buffer will be resized.

RGB Channels copies any of the Red green or blue channels of the stored image into the main buffer.  Say you only wanted to work with one channel, you could store your image, clear your main buffer and get just the one channel you wanted to work on.  When you are done, you can add in the other two channels.

HSV Channels converts the stored buffer to HSV space and copies any of the Hue, saturation, or Value buffer into the main buffer as a grayscale image.

YUV Channels converts the stored buffer to YUV space and copies any of the YUV buffers into the main buffer as a grayscale image.

As an example, this image was cleaned up extensively by modifying only the blue channel, which contained most of the film grain in the image.

The process was to store the original image and clear the buffer to black.  The blue channel was then added back, and a median filter was applied to it several times to reduce the grain.  Then the red and green channels were added back to the image.

New from clipboard creates a new buffer from an image contained on the system clipboard, destroying any previous buffer.

Overlay from clipboard places the contents of the system clipboard into the current buffer.  If the image on the clipboard does not fit in the buffer, it will be cropped.

If you want to place an image from the clipboard into your buffer manually, one way to do it would be to load it from the clipboard into a custom brush, then stamp it down manually.  See the section on custom brushes for more information.

It is converted to a bitmap...

There are several transformation options available that work on buffers.  (It's worth noting that all of them also work on entire animations)

Flipping a buffer creates a mirror image.
In the examples, the image is flipped on the horizontal axis, and on the vertical axis.

Rotating a buffer tilts the image, like turning a photograph sideways by 90 degrees.

Resampling, or scaling a buffer changes the size of the image by stretching it to fit a new size.

If you do not want to use one of the preset sizes, you can enter your own image size in pixels into the text boxes on the right.

The Constrain check box forces your new size to match the same aspect ratio as the original image.

The dropdown control lets you select the type of interpolation used to resample the image.  Since images are composed of a series of pixels, it is necessary to interpolate them to get a smoother result.
The interpolation types are nearest neighbor, bi-linear, and bi-linear+.
The "+" in bi-linear+ means that the image is also super-sampled, or multiple samples are taken of every pixel when they are scaled, for a better result.

The flip control   at the bottom changes the orientation of the image by flipping the width and height.

Resizing

You can change the Buffer size without scaling the image.  This results either in the image being cropped, or in new areas being revealed.

The Anchor lets you specify where in the new image that the old image should be placed.

The new area will be filled with the secondary color

Crop to Selection

You can crop an image to keep only the part that interests you.

This is like chopping of parts of a photograph with scissors.

Use it to create a marquis around the area.

Select 'Crop to selection' from the buffer menu, and your image will be cropped.

You can also create a selection numerically.

Result of cropping.

See the section on the Alpha channel for more information.

The next few features act like filters to change the appearance of the image in the buffer.  The difference between the features on the buffer menu, and the ones under the Filter menu is that the alpha channel is ignored, and that all the buffer features work on animations.

Grayscale

In the first case, the red, green, and blue channels are mathematically averaged with the formula (R+G+B)/3

With Video weighting averaging, the channels are averaged in a way that is more consistent with human vision and video systems.  The formula is 0.299 * red + 0.587 * green + 0.114 * blue.

Inverting an image

Expanding dynamic range is a quick way to automatically adjust the contrast of an image.

In a 24 bit image, with 8 bit per channel, the highest value for any channel is 255, and the lowest value is 0.  If the value for a pixel in all three channels is 0, then the color is black.  If all the value for all channels is 255, then the color is white.

However, not all imaging devices capture perfect blacks or whites, and may give you an image that looks muddy.  Expanding the dynamic range can often make these images look nice.

A histogram shows the effects of expanding dynamic range.  The closer the range of values is to the edges of the histogram, the more contrast your image will have.

There is also the option to clear all the frames of an animation.  They will be cleared to the secondary color.

Paint fill is a special tool that fills the buffer or selection with a paint like effect.  The values in the alpha channel have an impact on the result of the paint fill.

Adjustable paint fill goes a step further and lets you adjust the effects orientation with a rubber band control.

You can make your paint fill fade out like a brush stroke.  The effect is also heavily effected by your current opacity.

Right click on the icon to bring up the menu.

Open layers panel opens the layers panel should you need it.

Set current layers mode lets you change the current layer's mode, or in this case, the mode for the swap mixing.  If you where working with layers, you could change the current layers mode.

Set tracing paper opacity lets you change the opacity of the trancing paper mixing mode.

Show alpha causes the image thumbnail to represent the alpha channel.

Note Layers are not saved with a file, so you will have to merge your layers before saving your image.
 

Additive Adds the RGB values of the main and swap buffers

Subtractive subtracts the main buffer from the swap.

Multiply mode multiplies the pixels of the swap and main buffers (and divides them by 255.) The result is similar to painting with ink on ink. The result is always darker.

Divide is the complement of multiply. Use it as you see fit.

Screen is similar to additive but keeps an image from ‘blowing out’ or becoming overly bright. It’s often useful for effects like stars and fire.

Around gray combines two colors around their middle brightness level.

Tracing paper lets the swap buffer show through, a lot like tracing paper.

Here, the animation has been inverted.

See the section on the timeline if you want to apply FX filters to an animation.