Development

Video: Building a gigantic, snow-capped mountain by increasing elevation

Surface Grids & Lasers are not yet available in Universe Sandbox! This post is a behind-the-scenes look at our work on these features. We don’t have a release date yet, but we’re getting closer.

If you haven’t seen them yet, check out the previous Surface Grids DevLogs #1, #2, #3, #4, #5, #6, #7, and #8. Keep in mind these are documenting a work-in-progress feature. Anything discussed or shown may not be representative of the final release state of Surface Grids.

A primer on Surface Grids for anyone not familiar:
It’s a feature we’re developing for Universe Sandbox that makes it possible to simulate values locally across the surface of an object. In effect, it allows for more detailed and accurate surface simulation and more dynamic and interactive surface visuals. It also makes it possible to add tools like the laser, which is essentially just a fun way of heating up localized areas of a surface.

New Highs & Lows

See that archipelago in the middle of the ocean? Those are the great Hawaiian islands. We’ve been tweaking the heightmap for Earth to get all the details right, and we’re happy to welcome back Hawaii and the full coastline for Florida. There are some details to iron out still, like the large body of water that’s currently sitting in the middle of Australia.

It’s a bit difficult to get the coastlines and bodies of water correct for a planet we’re all intimately familiar with — any errors are quickly noticed by a geography whiz or anyone who lives near the area that’s incorrectly flooded or barren.

This difficulty is due to a current limitation in the elevation simulation: When we divide the full range of elevation for Earth into the limited steps we have available for storing the data, we get 83 meters per step. This means the maximum detail we can account for is 83 meters. And unfortunately, geography doesn’t often look like 83 meter ocean-side cliffs. A lot of coastline is much closer to sea level, and the smaller distinctions between land and water are lost. We have ideas for improving this, but for now, some hand-tuning of the heightmaps can get us the results we’re looking for.

Creating Continents

Another issue we’re grappling with is the state of heightmaps for randomly generated planets. Right now, adding some water to a randomly generated planet will create bodies of water and land masses as you’d expect, but the results look less like geological formations and more like noisy data.

This is something our lead graphics developer, Georg, has been wanting to work on, but his plate has been full with other aspects of Grids. …Enter Brendan, our new graphics developer! Brendan will be working with Georg on graphics for Universe Sandbox, and first on his list is the not-at-all-complicated (wink) task of creating a random heightmap generator capable of building known geological features in unique, random configurations. We’re happy to welcome Brendan to the team. Hopefully we’ll have some screenshots of his work in our next devlog.

Making Mountains

Right now something interesting happens when you hold shift while using the laser: it makes mountains. Or technically speaking, it raises the elevation. You can also decrease the elevation, too. So even if you don’t like the new and improved randomly generated heightmaps that we’re working on, you can go in and shape it yourself.

This functionality won’t remain attached to the laser, that’s just a temporary, developer-only solution until we build out the interface for this new tool. We’re pretty sure you’re gonna like it.

In the video above, we’ve divided a planet with a mountain range and then flooded one half, creating one hemisphere of ocean and another of land.

What’s Next

In addition to working on these various aspects of heightmaps and elevation, we’ve been doing a lot of work in other areas of the Grids simulation and tech, like getting visuals working on gas giants and planetoids (see above) and making the auto setting for lasers a bit smarter.

Up next is patching up all the holes we find as we continue to poke at the simulation. We also want to add a lot more visual detail to the edges of different areas like ice and molten surfaces to make it look a bit more natural. And there are plenty of other improvements and fixes to make.

We’re only kind of joking when we say it’s getting harder to pull ourselves away from playing with this feature so we can actually work on it. But in all seriousness, we continue to make good progress and we’re really liking the way it’s coming together. We can’t wait to share it. See you in our next devlog!
 
 
Follow @universesandbox

GIF: Filling in the Moon’s craters with ice. 

Here’s our round eight DevLog on the development status of Surface Grids and Lasers. Thanks for everyone’s patience with this blog post. We took a little break from Surface Grids to talk about our work on Magic Leap in our last post — check it out to see planets bouncing off of walls and our new, floating user interface. And if you haven’t seen them yet, check out the previous Surface Grids DevLogs #1, #2, #3, #4, #5, #6, and #7. 

A primer on Surface Grids for anyone not familiar:
It’s a feature we’re developing for Universe Sandbox that makes it possible to simulate values locally across the surface of an object. In effect, it allows for more detailed and accurate surface simulation and more dynamic and interactive surface visuals. It also makes it possible to add tools like the laser, which is essentially just a fun way of heating up localized areas of a surface.

Keep in mind this is a development log for a work-in-progress feature. Anything discussed or shown may not be representative of the final release state of Surface Grids. Read: Surface Grids & Lasers are not yet available in Universe Sandbox! To make this more clear, we’re now calling these posts DevLogs instead of Dev Updates.

Random Improvements

Everyone likes Earth and the rest of the cool, popular planets that make up our Solar System, but sometimes you want something a little different. That’s why we have the randomly generated planets. And with Surface Grids, the randomly generated planets are getting a little makeover.

We now have new textures and new elevation maps (which you can now see in the data map) that spruce these up and make them more unique. And we have further plans to improve how these are generated to widen the range of possibilities and customization.

Water Your Planet

It’s been possible for a while now to add water to planets in Universe Sandbox. But it’s never looked so good as it does with Surface Grids.

The key components to this improvement are great examples of what make Surface Grids an awesome and powerful new feature. It’s all about the localized data: First, you can see that the water spreads locally. Before Surface Grids, water would just fill in across the whole surface, regardless of where it hit. Second, you can see the direct correlation between elevation and water level. This was sort of possible pre-Surface Grids, but now there’s a data map for both of these that makes it even easier to see this in action (see screenshot below). And third, you can see the water freeze and the ice melt locally as well. That’s the nice localized temperature part of it.

Watch Your Step

Detouring from how awesome Surface Grids is, let’s look at one of its fundamental challenges. As with most simulation features in Universe Sandbox, Surface Grids has to battle against the simulation time step (the rate at which the simulation runs, for example 10 days per real-time second). We know that it would be awesome if you could just set the time step to whatever rate you wanted and it would just work, but that’s unfortunately not possible, at least not without sacrificing accuracy.

This fact is most obvious with gravity simulation in Universe Sandbox. The simulation automatically sets a limit to how fast the time step can go while still maintaining relatively accurate orbits. And if you try to set it faster than this limit, you’ll see a message that it’s not safe to do so. For Surface Grids, we’ll need to add a similar warning system, as it too has a limit for accurate results.

A good example of time step limits for Surface Grids is in the simulation of water flow. This is simulated by moving X volume of water from one cell to its neighboring cell each step. The maximum volume of water that can be moved in each step is the maximum volume contained in that cell — it can’t transfer water that it doesn’t have. So if you’re running at a time step that is already moving the maximum amount of water and you try to increase the time step further, the data starts to get a little weird.

We have some ideas for improving this, but ultimately there’s no way around time step limits. Our hope is that we can make a smooth experience by communicating where these limits are while allowing you to exceed them at the risk of accuracy loss.

What’s Next

We’re feeling really good about our progress on the Surface Grids feature. Now that we have the visuals and systems working for random planets, it’s time to turn to planetoids and gas giants. We also want to work on visualization for vapor, but that’s one of the few remaining items that still need a connection between data and graphics.

There are some more apparent issues to work through yet, too, like getting water levels to initialize properly on Earth. Then there are the new elements of the user interface (UI) that we’ve designed but have yet to add, and some questions about UI interaction that come up as we continue to play around with Surface Grids. And of course there are the inevitable giant bugs running around that have to be squashed.

All of this will definitely keep us busy, but the pile of tasks gets smaller every day as we get closer to releasing an experimental build and getting out the official update. We still can’t say when either of these will come yet — we thank you for your patience.

We’re also still working on a smaller update that will introduce some improvements and bug fixes for the new galaxies added in Update 23, plus a new introductory experience. Hopefully we’ll have this ready soon!

Follow @universesandbox

GIF: Lasering away the ice on Earth.

Here’s our round seven update on the development status of Surface Grids and Lasers. If you haven’t seen them yet, check out Dev Updates #1, #2, #3, #4, #5, and #6.

We hope you’ve been enjoying the new galaxies we added in Update 23. We still have another round of improvements and bug fixes planned for galaxies, but we’re proud of what we released and we’re shifting our excitement back to Surface Grids & Lasers.

A primer on Surface Grids for anyone not familiar:
It’s a feature we’re developing for Universe Sandbox that makes it possible to simulate values locally across the surface of an object. In effect, it allows for more detailed and accurate surface simulation and more dynamic and interactive surface visuals. It also makes it possible to add tools like the laser, which is essentially just a fun way of heating up localized areas of a surface.

Keep in mind this is a development log for a work-in-progress feature. Anything discussed or shown may not be representative of the final release state of Surface Grids. Read: Surface Grids & Lasers are not yet available in Universe Sandbox!
 

Gotta See It to Believe It

In the past few weeks, our graphics developer, Georg, has been taking the visual foundation that was built into the last version of Grids and plugging it into the new model.

This includes 1) getting the basics down for the shaders handling different materials and phases, from water to snow to molten silicate 2) getting all of Earth’s vegetation in the right spots 3) adding lighting, including effects for diffusion and specular and rim lighting for solar and atmospheric effects 4) and adding normal mapping.

The normal mapping creates the effect of bumps and ridges, or in other words elevation changes for things like mountains and crater rims. For a lot of visual effects, we often start with more exaggerated settings then dial it back and tweak it until we reach a more realistic appearance. In the screenshots below, the bumps are very pronounced and make for surface features that wouldn’t be discernible when viewing from space. While it’s not realistic and we plan to tone it down for the default setting, we are considering exposing this value and letting anyone set it to however they prefer.

 
Please note that you can safely ignore the incorrect water levels right now. We’re well aware that Europe is completely flooded and we’re working on a fix (even if it may be an accurate representation of what human-caused climate change will do to Earth…).

This visual foundation now works for planets with known heightmaps, but it will eventually be applied to all procedurally generated objects. So whether you’re lasering Earth or a randomly generated rocky planet, you’ll see the nice lighting and normal mapping.

Graphics for Surface Grids are definitely still incomplete, but everything is coming together nicely. The screenshot below shows a tidally-locked, near-Sun Earth that is frozen on the far side and molten on the near side (note: it looks especially “clean” now because it doesn’t yet include additional visual noise from blurring, blending, and randomness).
 

Lasers!

We haven’t shared much about lasers yet, so let’s fix that. Before, lasers were hooked up to the data side of things, so you could see its effects in the data map, but that was about it. Now that we have the visual foundation described above and can see the effects of temperature on the planet itself, playing around with lasers just got more interesting.

We recently added a radius setting that multiplies the area of effect (the laser visual itself is not yet affected by this). So whether you want to melt the whole ice cap or you want to carve your name, you can do it.

In the case of the GIF shared at the top of the post (and maybe I’m just hungry right now), this reminds me most of decorating a cake. Especially the part at the end where I do away with careful decoration and just throw a bunch of icing (er, lasering?) on the top.

We often see debates in the Universe Sandbox community about these lasers — some say they’re unrealistic and not a good use of development time, and others will stop at nothing to get their hands on them.

To the first, naysaying side we say this: Yeah, lasers like this don’t exist (at least, not yet). But… what if they did? Universe Sandbox has always been about realistic simulation of fantastical scenarios, and we think this fits very nicely into that. As for development time, lasers are a tiny fraction of the larger work on Surface Grids. They’re essentially a nice side-effect of having a system like Grids that can simulate localized surface properties, much in the same way Grids will allow us to add basic life simulation in the future as well.

And to the laser fanatics we say this: Yeah, they’re fun. Thanks for your patience!

What’s Next

There are a few higher-level things we want to get working, like being able to locally deposit water and having vapor flow connected to axial tilt. And then there are some must-fix issues like proper water level on Earth and proper phase-handling for Venus (it shouldn’t look like it’s covered in liquid water, right?). There’s also more visual work to do with blending materials and phases and handling procedurally generated objects.

So we’re a long way from the starting gate, but we’re not quite in the final stretch yet. It may be fair to say we’re at the point that was described all the way back in the second dev update: “… often times the longest and most challenging part of development comes later, when we’re tweaking, polishing, finding and fixing bugs, and making sure all the complexities of the fully fleshed out model play nicely with each other and the rest of the simulation.”

We’re definitely interested in making experimental builds available to the community in the future, before we’re ready for a full release. We’re not ready to say when these will come yet, but we hope they’re not far away.

As always, thanks for your patience and support while we work on this next big, complex feature for Universe Sandbox!
 
Follow @universesandbox

 
Here’s our round six update on the development status of Surface Grids and Lasers. If you haven’t seen them yet, check out Dev Updates #1, #2, #3, #4, and #5.

This will be a smaller update than usual because we’re focusing our efforts on the last sprint for the new galaxies we’ve been working on. We hope you won’t have to wait long for their official release, but if you’re feeling impatient, you can check them out by opting into the experimental version.

A primer on Surface Grids for anyone not familiar:

It’s a feature we’re developing for Universe Sandbox that makes it possible to simulate values locally across the surface of an object. In effect, it allows for more detailed and accurate surface simulation and more dynamic and interactive surface visuals. It also makes it possible to add tools like the laser, which is essentially just a fun way of heating up localized areas of a surface.

Keep in mind this is a development log for a work-in-progress feature. Anything discussed or shown may not be representative of the final release state of Surface Grids.

Set Phases to Accurate

Our last dev update focused on the upcoming galaxies, but at the end we mentioned how Jenn, astrophysicist and Universe Sandbox developer, was working on vapor flow for the Grids model. We also joked that the challenging part here is creating this “accurately and performatively without single-handedly developing Weather Simulator 2020.” Unfortunately, the joke is all too real!

That doesn’t mean we are actually developing a complex weather simulator, but there is nonetheless complexity. While developing the vapor flow model, Jenn has been doing her homework with research into fluid systems and geophysics (if you’re looking for some light reading, you can check out Lectures On Dynamical Meteorology, Geophysical Fluid Dynamics, and An Introduction to Planetary Atmospheres).

The first part of vapor flow is determining what exactly is vapor; this is done by phase tracking, which determines what phase each material is in, based on temperature and elevation. For water, vapor is its gas state, then we need to see where it’s headed from there — whether it’s evaporating or condensing at the dew point, depositing, if it should be boiling away completely, etc. If it’s remaining as vapor, then the model accounts for a few factors to move it around: there are the prevailing winds which vary by latitude, are affected by planet rotation, and cause east-west movements (as seen in the GIF at the top), and there are temperature differentials from thermal flows and elevation that cause north-south movement (still a work in progress).

Image: The “heat map” shows vapor amounts (not vapor flow), where red is higher amounts and blue is lower. The poles are both much drier, and because there’s much less evaporation over land than oceans, you can see the outlines of the continents.

We want to stress that this is necessarily a very simplistic model, largely limited by its low resolution and the amount of memory we can allocate for this single component of Grids. There are lots of things missing that make this very different from more complex weather simulations — there are no vortices, so there won’t be anything like hurricanes, it is only a 2D simulation with no layers through the vertical dimension, and it’s fairly low resolution.

But we hope to use this data for the resulting local vapor amounts to have rough approximations for clouds, ice caps (for Mars, this effect happens with CO2 vapor flow), and for the future implementation of basic life simulation (vegetation), it could affect growth in dry and wet areas.

Loading…

In our last post, we also mentioned Chris’s work on saving and loading with Grids. This component of the feature obviously isn’t as interesting as, say, lasers, but at the same time, it’s essential to get it right and it’s another good representation of challenges on the edges of new feature development.

Here’s a shortlist of some of the questions and challenges that doesn’t even get into the technical weeds: How can we deal with file type and size limits for different platforms, like Steam Workshop, mobile devices, etc.? How can we maintain file size for fast, background autosaving and quicksaving? How can we get it to play nicely with previously saved simulations with objects that didn’t have all of the Grids data?

We had similar saving and loading questions with the new galaxies: What should happen if you load simulations that had the old galaxies? They won’t look and function the same. Should we change their shape and motion to use the new model, or should we preserve appearance? Is it okay to change sims on Steam Workshop that are very popular?

Saving and loading is something we all hope just works seamlessly and shouldn’t be something the player ever has to think about — which are both telltale signs that there is little room for bugs, errors, and bad user experience (UX). Thankfully, we have answers to all of these questions!

What’s Next for Grids

We’re hoping to make some good progress again on the visual side of Grids, rendering all of that wondrous data into some nice planet graphics. We’ve been recruiting our graphics developer, Georg, to work on some other projects (like the now so gorgeous galaxies), but it’s time for Grids attention again.

Thanks for reading! We’ll be back in two weeks with another update on development. And hopefully before that, we’ll have our next big update with new galaxies.

Follow @universesandbox