Authors: Dr. Megan, Sean, Jonathan, and Brendan

Note: you can view the blog full blog post on our website for more images, interactive image comparisons, and more.

Hey folks! In preparation for the public release of the Universe Generation Update, we’re highlighting some major changes you can expect to SpaceEngine with this fresh batch of cosmos!

Planetary System Architecture

One of the biggest frustrations with 0.990’s procedural universe is the sameness between systems. Established users know the pattern: giant-terra-giant-terra, big planet, small planet, etc. With 0.991, this is no longer the case!

Planets will now generate in more realistic, diverse patterns. In this example, the procedural system shown below was once exactly what you’d expect: terra, ice giant, aquaria, gas giant, and so on. The same system (in name, at least) has taken an entirely new identity, changing its location in space. Planets 3 and 5 in this new system are gas giants, but the rest are smaller terras, subterras, and aquarias!

An example of planetary system generation in 0.990 (top) and 0.991 (bottom). View the blog post on our website for an interactive comparison.

Also, there's something you'll find much more often in 0.991: dwarf planets! Previously, dwarf planets were exceedingly rare among the generated parts of the universe, which wasn’t very realistic.

While the old meander sequence is still technically possible, you can expect far more interesting finds in unknown space!

Visual Improvements

The 0.991 release contains a couple of visual improvements worth mentioning. For starters, underwater terrain is now colored according to its depth, emulating the absorption of sunlight in the overlying water. It's a seemingly small change, but it significantly enhances immersion when underwater. No more sea floors bathed in sunlight under 5 km of ocean!

Sea floor images from a variety of depths, showing their appearance in 0.990 (left) and 0.991 (right). See the blog post on our website for more images and interactive comparisons.

We've also made some adjustments to the surfaces of stars. In earlier versions of SE, including 0.990, most stars had granulation patterns with much higher contrast than in reality, and which displayed as being cooler (redder) than they should have. These granule temperatures have been replaced with a more realistic value, with further improvements to their scientific accuracy coming in the future. Sunspots have also seen an update to their temperature generation, with their temperatures now determined by the effective temperature of the star according to an equation based on real astronomical observations.

A comparison of granules and sunspots on a G2V star in 0.990 (left) and 0.991 (right). View the blog post on our website for an interactive comparison.

The lower contrast of the new granulation may seem less visually interesting, but it's much more representative of what stars really look like. To highlight this, the following image shows a real photo of the Sun (left) taken by one of our own team members, and the Sun with the new temperature values in 0.991 (right). The photograph was white balanced to match the real color of the Sun as closely as possible, and both images were tonemapped from linear brightness values using the same function (Reinhard SE).

Left: A real photo of the Sun taken by one of our own team members. Right: The Sun in SpaceEngine version 0.991 with the revised granulation and sunspot temperatures.

Further updates to the accuracy of star surfaces are being worked on for future release.

Galactic Globular Cluster Systems

The generation of globular cluster systems (GCS) – the globular cluster populations belonging to individual galaxies – has also received some improvements. In 0.990, the number of globular clusters (GC) in a galaxy was related strictly to the volume of that galaxy's model. This didn't make much sense for spiral galaxies, as the amount of matter they contain is more closely related to the square of their radius, unlike volume, which grows with the cube of the radius. This resulted in large spiral galaxies having a very large GCSs, while smaller spirals had much smaller GCSs than they should have. While it would seem to make more sense for elliptical galaxies, it still wasn't closely based on any observed trends in GC populations. As a result, elliptical galaxies often had smaller GCSs than they should have, especially the largest galaxies.

In 0.991, GCS generation has been overhauled. Elliptical galaxy GCS generation in particular is much more realistic, following population trends observed by astronomers. One way of describing the population of a galaxy's GCS is with a term called the "specific frequency" (SF). The number of clusters in a GCS is a function of the SF and the host galaxy's absolute magnitude (intrinsic brightness, i.e., luminosity). When looking at the GC populations of elliptical galaxies, astronomers observe that small elliptical galaxies tend to have a lower SF, while larger galaxies have a higher SF. Interestingly, the SF does not significantly change within those size categories, only around a transition point in between. Based on the available research, 0.991 uses an absolute magnitude of -18.5 for this transition. That's why, on the graph below, you can see that as you move from right to left (from fainter to brighter), the graph levels off, indicating that brightness increases faster than the number of GCs. However, as the absolute magnitude increases further, the size of the galaxy's GCS starts increasing more quickly. After reaching magnitude -18.5, the relationship is amplified even further in favor of larger GCS populations. This is the result of the difference in GCS SF between fainter and brighter elliptical galaxies.

This graph shows the number of globular clusters (Y axis) that generate in elliptical galaxies in 0.991 as a function of galaxy absolute visual magnitude (X axis).

Lenticular galaxies (type S0) also have their GCS generation based on a SF relation, using a single SF value that's fairly typical of the type. The SF of spiral galaxy GCSs in nature varies based on a number of variables that are beyond what SE can accurately accommodate at present, so their GC populations scale with the square of the galaxy's radius. In all cases, this results in a more realistic number of globular clusters than in 0.990.

The generation of GC luminosities has also been significantly improved. In previous versions of SpaceEngine, GC luminosity generation used a normal distribution (i.e. a Gaussian function or bell curve) with respect to luminosity. In reality, the luminosities of a galaxy's GCs follow a normal distribution with respect to absolute magnitude. This seemingly simple change results in much more realistic and natural looking GC systems, as visible in the comparison image below.

A comparison between globular cluster luminosity generation in 0.990 (left) and 0.991 (right). View the blog post on our website for an interactive comparison.

There's also a small bonus feature related to the above work on globular clusters: you can now specify the number of globular clusters in catalog galaxies! Multiple galaxies in SE's catalog have already been updated to generate the correct number of globular clusters based on astronomical observations. Once 0.991 comes out, you can do the same for any catalog galaxy you create by using the GlobClusters parameter in the galaxy's catalog entry, followed by the desired number of clusters. Enjoy!

Brown Dwarf Effective Temperature and Radii

Finally, we have made improvements to the effective temperatures and radii of ultra cool dwarfs (UCDs), which include the dimmest red dwarfs (spectral type M7 and later) as well as all brown dwarfs (including L, T, and Y spectral types).

There are several ways to define the temperature of a star or brown dwarf, depending on what part of the body you’re looking at (e.g., core, photosphere, chromosphere, etc.) and what wavelength you’re looking at (e.g., optical, infrared, all wavelengths, etc.). The effective temperature (T_eff) is a particularly useful temperature quantity that allows us to estimate the surface temperature of a body, and it is defined as:

T_eff = [ L / (4 π R^2 σ) ]^0.25 ,

where L is the bolometric luminosity (the luminosity measured across all wavelengths), R is the radius of the body, and σ is the Stefan-Boltzmann constant. The effective temperature is the temperature of a blackbody (a theoretical object that perfectly absorbs and emits radiation) that emits the same amount of energy as a star or brown dwarf of the same radius. Stars and brown dwarfs are, of course, not perfect blackbodies. The atoms and molecules present in their atmospheres cause absorption and emission of light at different wavelengths. But the T_eff is still a useful quantity for categorizing and understanding stars and brown dwarfs. The T_eff of the Sun is 5778 K, and for UCDs T_eff ranges from about 2700 K all the way down to just hundreds of Kelvin. Some theories estimate that the faintest Y dwarfs may be as cool as 90 K.

In the 0.991 update, we revised the T_eff of UCDs using values measured and published by astronomers. The figure below shows the T_eff for different spectral types. In the case of our UCDs, it is immediately noticeable that T_eff does not change linearly with spectral type. This is because spectral types are a sequence of spectral features (light absorbed by atoms and molecules in their atmospheres), not a temperature sequence. The transition between the late-L brown dwarfs and the early-T brown dwarfs is especially interesting, as the T_eff levels off over a few steps in the spectral sequence. This is because the diversity of molecules present in the atmospheres of brown dwarfs at these “low” temperatures (around 1200 K) can make spectra look very different, despite having approximately the same T_eff. At these temperatures, the main carbon-bearing molecule changes from carbon monoxide to methane. This switch is the hallmark of the transition from L- to T-type brown dwarfs and results in some very interesting chemistry in the atmospheres of these objects.

Effective temperature (T_eff ) as a function of spectral type for Ultra Cool Dwarfs. The relationship used in SpaceEngine 0.990 is shown with green diamonds, and the new relationship in 0.991 is shown with red squares. The new relationship is based on data from Filippazzo et al. (2015), Leggett et al. (2017), and Kirkpatrick et al. (2021). We adopted the fit from Filippazzo et al. (2015) for M, L, and T dwarfs, and extended that fit through the early-Y dwarfs using T_eff values available in the literature. We extrapolated through the mid- and late-Y dwarfs to a fixed value of 90 K for Y9.9.

We have also improved the radii of field-age (>5 Gyr) brown dwarfs in this update. As substellar objects, the forces in the cores of brown dwarfs have an interesting balancing act that results in fairly constant radii over nearly two orders of magnitude in mass. Despite their cool atmospheres, the cores of brown dwarfs can still be millions of Kelvin – not quite hot enough to overcome the Coulomb barrier and ignite nuclear fusion, but certainly hot enough to ionize hydrogen. The Coulomb barrier fixes the inter-particle distance in the interior of brown dwarfs and leads to the radius being proportional to mass to the power of +1/3. But the ionization of the hydrogen means that the outer layers of a brown dwarf are supported from gravitational collapse by electron degeneracy pressure. Under this condition, the radius is proportional to mass to the power of -1/3. These two effects cancel each other out, resulting in a fairly constant radius, regardless of mass. Of course, these effects are not always perfectly balanced, and the electron degeneracy effects matter more for high mass-brown dwarfs, while the Coulomb effects matter more for lower-mass brown dwarfs, and so the radii vary by about 15% over the range of brown dwarf masses. The figure below shows our new mass-radius relationship compared to our old relationship, based on brown dwarf evolutionary models.

Radius as a function of mass for the typical mass range of field-age (5 Myr) brown dwarfs. The old SpaceEngine radius relationship is shown in green, and the new, revised relationship is shown in cyan. This relationship is a 6th-degree polynomial fit to data from the Sonora brown dwarf evolutionary models (Marley et al. 2021). The Sonora model data are shown as white circles.

This is only a selection of notable improvements from the coming update. There are a number of other changes coming in 0.991, and you'll be able to read the full changelog and experience it for yourself when public beta testing begins next week!

Stay Tuned!