Bringing JWST images to life

One year ago, the public got its first look at images from the James Webb Space Telescope (JWST). The magnificent images showcased the 6.5 m IR telescope’s ability to see through cosmic dust, discover ancient galaxies, present stunning nebulae in new ways, and more. To mark the anniversary, Physics Today explored how astronomers transform the raw, black-and-white filter data from the IR telescope into the spectacular images that can be appreciated by both scientists and nonscientists.

Consumer cameras, including the ones in cell phones, use filters that capture multiple wavelength ranges simultaneously; each photograph is quickly processed and combined into a full-color image. Astronomers, however, are far more interested in individual filter images from telescopes than they are in color composites. Each filter produces a single black-and-white photo that depicts the intensity of light emitted within that filter’s wavelength range.

If those ranges fall within the visible spectrum, it can be relatively straightforward to create a color image by mapping the intensity to a single color for each filter. Combine the red, green, and blue and you have a decent start. But the output from most telescopes is not that simple. There are broadband filters that cover a wide range of wavelengths, often in regions of the electromagnetic spectrum that are undetectable by the human eye. Then there are narrowband filters that capture the precise colors produced by specific elements, such as hydrogen.

We turned to Travis Rector, an astronomer at the University of Alaska, Anchorage, to walk us through the steps for colorizing telescope imagery. A coauthor of the book Colorizing the Universe, Rector has been doing astronomical colorization since 1998, when he was asked to demonstrate the capabilities of a new telescope camera at Kitt Peak National Observatory in Arizona. Today he is part of a small team that processes images from NSF’s NOIRLab, which manages Kitt Peak and other observatories.

“We’re turning what the telescope can see into what our eyes can see,” Rector says. Like translating one language to another, you must make some assumptions; it’s an art as much as it is a science. There is one main rule to follow for broadband filters, he says: chromatic ordering. You start by putting the individual filter images in wavelength order. The shortest wavelength is colored blue, and the longest is colored red. All others fall in order between those.

It's not quite a straightforward color-by-numbers process, though. Different filters cover different ranges of wavelengths. For example, the F200W filter of JWST’s Near-Infrared Camera (NIRCam) is a general-purpose filter with a wide bandwidth (see figure 1). The F470N filter has a much narrower wavelength range because it is designed to detect H₂ emission. Often a narrowband filter image includes wavelengths that fall within those of a broadband filter.

Narrowband filters are designed to show a certain phenomenon, such as knots of star formation in an irregular galaxy or an oxygen-rich region of a planetary nebula. Rector tries to make that information stand out. Choosing a color that contrasts with the corresponding broadband filter can do this, as long as chromatic order is preserved among all the narrowband images. Other common procedures include the use of nondestructive editing, such as using a curves tool to adjust an image’s highlights and shadows and de-emphasize the background. Defects in images are also removed so they aren’t confused with real structures.

With those basic principles in mind, it was time to create our own color images with standard photo-editing software. Rector and I downloaded the JWST NIRCam raw filter images for galaxy NGC 1300, a barred spiral galaxy in the southern constellation Eridanus. (JWST images, along with data from Hubble, Kepler, and other telescopes, can be found in the MAST catalog.) As shown in figure 2, there are four publicly available images: one wide filter (F200W) and three medium-width filters (F300M, F335M, and F360M).

Looking at the filter chart for the instrument reveals that F335M and F360M partly cover the same wavelength range. Such overlap is something Rector pays attention to when deciding on colors. “I’m not going to make the two colors too radically different,” he says.

Both Rector and I used the curves tool to highlight the interesting structure in the spiral arms, where new stars are being born (see figure 3 below).

Even though Rector and I had access to the same tools and raw images, it’s not difficult to see which processed image was made by the professional. Rector’s image, shown at left in figure 4 below, has a color scheme that allows the image to look more natural, and the bulge in the galactic center is bright yet does not wash out the central bar.

For this exercise, we were using a set of images that had been through some processing to prepare them for public use. It’s not always so simple. Sometimes the data quality isn’t optimal; a part of the sky could be missing due to dead pixels, or the stacking of multiple exposures could leave unequal depth on the edges. When that happens, Rector tries to supplement with other data from the same part of the sky or see if cropping can make the presentation better. He’s found that a close crop often makes the object seem even more grand and awe inspiring. Data quality issues are the most challenging part of the job, often requiring hundreds of hours of tedious, meticulous cleaning.

We stopped here, but there’s still more processing to do. Later steps “reduce the noise, deal with cosmetic defects, and bring out the contrast,” Rector says. There are also processing tricks that can be used, for example, to bring out the ring in the center of the galaxy that is obscured by the brightness of the bulge. It’s in these small adjustments that personal preference comes into play. Final processing also includes making the colors more vivid, for more of a visual impact.

Over the years, Rector has advanced his thinking about the images, learning from artist friends about how people perceive color and what tends to look good. “You can use the color and composition to draw people’s eyes to the things you want them to see,” he says. Tight cropping, rotation, and adjustment of the general color scheme can help create a lingering impression of an image.

This was Rector’s first time working with JWST data, but he is familiar with IR telescopes. Foreground stars have fewer distinct colors in the IR than in visible light, he says. Our eyes have evolved to take advantage of the range of light produced by Sunlike stars; the differences in color and temperature for stars is most dramatic in the visible. Observing at IR wavelengths also reveals the influence of cosmic dust (see figure 5 for a vivid example from JWST). In near-IR, the dust is transparent, enabling views inside stellar nurseries. Some dust emits light at far-IR wavelengths, which creates an ethereal glow in the images. “This gives a really spectacular feel to images,” whether they are from JWST or other IR telescopes like the Spitzer Space Telescope, he says.

Choose a filter image: Combined image F090W F187N F200W F335M F444W F470N

Rector says he is impressed with the astronomers who have been colorizing the JWST images. He appreciates the depiction of images that are familiar—often because the celestial targets were captured previously by Hubble—through a new IR lens. “I think one of the ways an image can be really successful is if people look at it and say, ‘I’ve seen that before, but never like this.’ ” The transformation from the familiar to the new is due to not only the JWST image processing team but also a telescope that has exceeded astronomers’ expectations in its performance.

As Rector observes, people seeing the images may understand all the astronomical science behind an irregular galaxy, or they may simply see a planetary nebula that inspires fascination. In either case, he says, astronomers have accomplished their goals.