---

2.6 Color images with full dynamic range ¶

Color images are fundamental tools to visualize astronomical datasets, allowing to visualize valuable physical information within them. A color image is a composite representation derived from different channels. Each channel usually corresponding to different filters (each showing wavelength intervals of the object’s spectrum). In general, most common color image formats (like JPEG, PNG or PDF) are defined from a combination of Red-Green-Blue (RGB) channels (to cover the optical range with normal cameras). These three filters are hard-wired in your monitor and in most normal camera (for example smartphone or DSLR) pixels. For more on the concept and usage of colors, see Color and Colormaps for single-channel pixels.

However, normal images (that you take with your smartphone during the day for example) have a very limited dynamic range (difference between brightest and fainest part of an image). For example in an image you take from a farm, the brightness pixel (the sky) cannot be more than 255 times the faintest/darkest shadow in the image (because normal cameras produce unsigned 8 bit integers; containing \(2^8=256\) levels; see Numeric data types).

However, astronomical sources span a much wider dynamic range such that their central parts can be tens of millions of times brighter than their larger outer regions. Our astronomical images in the FITS format are therefore usually 32-bit floating points to preserve this information. Therefore a simple linear scaling of 32-bit astronomical data to the 8-bit range will put most of the pixels on the darkest level and barely show anything! This presents a major challenge in visualizing our astronomical images on a monitor, in print or for a projector when showing slides.

In this tutorial, we review how to prepare your images and create informative RGB images for your PDF reports. We start with aligning the images to the same pixel grid (which is usually necessary!) and using the low-level engine (Gnuastro’s ConvertType program) directly to create an RGB image. Afterwards, we will use a higher-level installed script (Color images with gray faint regions). This is a high-level wrapper over ConvertType that does some pre-processing and stretches the pixel values to enhance their 8-bit representation before calling ConvertType.

• Color channels in same pixel grid
• Color image using linear transformation
• Color for bright regions and grayscale for faint
• Manually setting color-black-gray regions
• Weights, contrast, markers and other customizations