Introduction to Programming with Python

How are images stored?

Vector and raster graphics

Raster graphics

Look at the pixels!

Bit depth/Colour depth

Look at the pixels!

Bit depth/Colour depth

Look at the pixels!

File formats

Compressed and uncompressed formats i.e JPEG and PNG
Image fidelity vital -> don’t compress
Intermediate workflow -> don’t compress

Once you have created an image of some sort, an important consideration is what format should you save it in?

Perhaps the main question to ask is, do I want a perfect rendition of the image I have created? There are two categories of image formats, lossy and lossless.

Jpeg is a good example of a lossy format. When you save something in jpeg, you will permanently lose some of the information as it tries to make the image file smaller. A lossless format can try and compress the file size down but not at the cost of the loss of information. PNG is a good example of a lossless format.

The picture on the slide of the cat shows how the level of compression (and distortion) can vary. The left hand side of the picture is very heavily compressed, and the right hand side has very little compression. You can see how the image quality increases as the level of compression decreases.

JPEG is good with smooth changes in colour, rapid changes, such as lines on a graph, can lead to ugly artefacts. JPEG is good at compressing “natural” images such as photos of landscapes.

To put all this in the context of you scientific workflow:
If you want perfect reproduction of an image (or as close to as possible) save images in a lossless format.

IF you are compressing image files, wait until the end of your workflow to do so. Compress once. Repeated compression will degrade your image quality.

Vector graphics

Polygons rather than pixels
Can scale up as small or large as we want -> rasterize
\

So we have seen that raster graphics are graphics composed of a grid of pixels. Vector graphics on the other hand are made up of parametrized shapes such polygons and splines. Polygons are shapes that can be created by bounding an area by straight lines, more simply triangle, squares and shapes made by combing them etc. Splines are curves generated from a mathematical expression.

The difference between vector and raster graphics becomes clear when we need to scale an image up. Thinking back to what happened when we scaled up the raster image a few slides back, the individual pixels became more and more visible. Likewise the shapes that make up a vector graphic will increase in size, but this doesn’t lead to a drop in fidelity. Vector graphics also allow for colour gradients, which means we can smoothly transition from one colour to another, regardless of the scale.

To display an vector graphic on a screen, it has to be rasterized, screens display images in pixels, so ultimately a vector graphic has to be converted for it to be displayed.

In a way, raster graphics are a special case of vector graphics, made up of only coloured squares, without any other shapes.

Vector graphics in your scientific workflow

Illustrations
Graphs