Dreamt Futures - Dev Blog #02

August 2, 2024

Mars as seen from Deimos - Chesley Bonestell

Since my last blog entry I’ve scratched off a few more shots, and while I work out some of the details for perhaps the most complex shot of the film, I thought I’d take a break and do a small progress update.

While I don’t want to share progress on every single shot (so there are still some unknowns upon first viewing!), I will share a few stills from some completed shots.

Deimos

Deimos is one of Mars’ two moons, the other being Phobos. Bonestell has paintings of both of these locations, and they are quite similar in look. For Dreamt Futures, I have taken ideas from the paintings of both locations, to create something that is still very recognisable as Bonestell’s artwork and composition, but bringing a little extra of my own flare.

I really enjoy Bonstell’s composition with Mars contrasting the view of the rocky surface of Deimos, and casting a red reflection. I wanted to take these concepts and push them a little further, to draw more attention to the astronauts in the painting. By moving the camera around and punching in a little closer, I managed to create more of a focus on the humans exploring the landscape and admiring the views of Mars.

Viewport view of the 3D scene in Cinema 4D

Deimos has an extremely low gravitational force - if you were standing on the surface and jumped, you would be at escape velocity! The Astronaut Explorer sitting on the rock in this shot is saving his friend from floating away after getting up a little too fast.

The lighting was a little tricker to set up here, I’m using multiple light sources to cast the highlights in the right areas, and still look realistic. For Mars, I created a new atmosphere shader, altering a set up I have used for Earth in the past. The technique involves remapping a Fresnel node to drive a subsurface colour, and another Fresnel node to control the opacity of the redshift material - this is then applied to a sphere which sits on top of the surface of the planet. With this set up you have good control of the depth and falloff of the atmosphere, as well as colour gradients.

Mimas

The second shot to talk about today is of Mimas, one of the moons of Saturn. Let’s first take a look at Bonestells depiction of the moon.

This is one of my favourite Bonestell paintings. The lighting on Saturn looks almost translucent, and the shadowing of the rings and moon being cast back onto the surface of Saturn are a sight to see. The biggest challenge in recreating this scene is the sense of scale. When creating landscape renders, we are used to using Atmosphere/Environment shaders. These help create a sense of depth as a viewer, and a graduation of colour as the view vanishes over the horizon. With a small rocky moon, there is no atmosphere, so it is hard to convey this concept. and keep realism in mind.

Looking at the foreground rocks in the image, I decided these would be the best way to help illustrate the scale of the landscape. The problem to solve then is how am I going to have millions of rocks active in a scene?

I used high quality proxy models of rocks for the first third of the image, and swapped them out for extremely low LOD proxies as they were positioned further away from the camera, and decreased in scale. I then instanced the proxies, and placed them into a Matrix Scatter. This allowed me to scatter a huge amount of geometry and not run out of VRAM when it came to rendering.

The base landscape was generated in World Creator, it is a 1km/1km base, which I used to scatter the geometry across. The base landscape was then exported as a TIFF, brought into Cinema 4D and displaced using Redshift. I used a displacer modifier in C4D and matched the elevation. This geometry was then used for the scatter, but swapped out for the displacement map in the final render. Most of the foreground elements were hand placed, so I had a good stage to animate the explorers wandering across the barren moon.

Lastly, to create a greater sense of depth, I created a ground ‘fog’. Although not present in real life, a subtle fog helps emulate the sense of moon dust which could be displaced above the surface of Mimas (especially when being disturbed by intrepid explorers!

While I love this Bonestell painting - its tone and colour arent quite right for what I am going for in Dreamt Futures - this will become more clear when watching the film with the shots in sequence. For Mimas, I decided to go for a much brighter look, with over exposed highlights, and darker shadows. I feel this helps create a feeling of a harsher alien environment.

That’s all I’ll be sharing for today. I hope you enjoyed reading, and having a sneak peak at a couple of shots. Remember to check back for updates in the future!

Dreamt Futures - Dev Blog #01

May 28, 2024

I'm deep into the post-production phase for Dreamt Futures. With storyboarding done and a previz edit complete, I'm excited to share some updates. Five shots are finished, and the remaining 20 plus are in various stages, from planning and layout to just getting started. In this post, I'll talk a bit about the production of the first two shots and share some stills without giving too much away.

Lick Observatory

The film begins in the hills outside San Jose, California, at the historic Lick Observatory. The main building houses the Great Lick Telescope, a massive 36-inch refractor. Both the telescope and the observatory are named in honor of James Lick, who dedicated much of his wealth to scientific endeavours, including the construction of the world's most powerful telescope at the time. When James Lick died in 1876, he was the wealthiest man in California and is actually buried in a tomb beneath the telescope.

Space artist Chesley Bonestell (who Dreamt Futures is focused on) born just two days before the telescope's first light in 1888, visited the observatory as a child. Viewing Saturn through the newly built telescope ignited his passion for space. This moment also serves as the inspiration for the film's opening scene.

Shot Production

Cinema 4D Viewport rendering of the scene set up.

To create the landscape, I carefully reshaped areas of the terrain where the observatory is located and filled in the background with DEM earth. This approach ensures the landscape closely resembles the actual location, using satellite imagery and Digital Elevation Model (DEM) data for accurate terrain height.

The pink and green blocks represent foliage scattered across the landscape. I added more trees around the observatory than there are in reality. This artistic choice helps break up the horizon lines and creates a more interesting gradation of shading, enhancing the sense of depth, distance, and scale. Placing trees on the more distant hills (in pink) also adds realism; the varied tree heights break up the horizon line and help the viewer gauge the distance to the city in the background.

For the outskirts of San Jose in the distance, I scattered particles across the geometry in both random and grid-like patterns to mimic streets. I used emissive materials to light these particles, simulating city lights, and slightly randomized their colors from white to gold, with some green and red lights to represent traffic signals. I also applied a subtle twinkling animation to some of the lights.

Additionally, I added clouds generated in Embergen to create further separation between the landscape layers. This adds an engaging level of contrast and balance, helping to silhouette the observatory structures and trees.

The final shot - modern day San Jose and the Lick Observatory. The shot was composited in After Effects (Including a heat haze effect over the city which was controlled via a depth pass), and graded in Premiere Pro.

The Telescope

I knew creating the telescope itself was going to be one of the more challenging and time-consuming aspects of the films production, which is why it was tackled early. It needed to be modelled in detail for two shots, one at very close range and one at middle distance. Finding good reference photos and understanding how the telescope moved and operated was quite a challenge due to the lack of available reference material. However, I managed to find some great pictures on Flickr from people's holiday snaps and got to work re-creating the scene.

The second challenge was the domed structure that houses the telescope. An intricate truss-like structure supports the dome, along with mechanisms below that open the huge doors to the stars and orient the observatory. I modeled the telescope in Cinema 4D, then textured, lit, and rendered it with Redshift 3D. For the telescope material, I created an entirely procedural shader to imitate paint wear and tear and heavy rusting in certain areas. The telescope is well over 100 years old now, so it isn’t quite as shiny as when Bonestell first saw Saturn!

Lighting this scene was extremely satisfying, as the structure creates incredible shadow patterns. Normally, the lights would be off to preserve night vision and achieve the best clarity for viewing celestial bodies through the telescope. However, for a film shot, complete darkness wouldn't work. There are some red lights for visibility within the observatory building, so I took some artistic license and increased their intensity to light up the dome.

One of the finished shots of the Lick Telescope and Observatory dome - you’ll have to wait for the full release for the other!

I hope you enjoyed a peak behind the scenes at how the filmmaking process has worked for these first two shots. There will be more to come in the near future!

ISS Cupola Moduel - Model + Animation Coming soon!

May 4, 2020

Hey folks, here’s something I’ve been chipping away at for a little while now. I’ll have a full post on some of the processes I went through to create this shot, but for now, here’s a teaser of whats to come!

Animate an Orrery (Model Solar System) with Cogs & Xpresso in Cinema 4D

October 15, 2019

I’ve been building a somewhat accurate Orrery in Cinema 4D, mainly for fun, as I certainly don’t have space for a physical one in my apartment. It’s a long-ish one, so grab a drink and a snack and lets get into it.

So, what on Earth is an Orrery?

It’s basically a mechanical model of the solar system, usually driven by a system of gears and cogs. Now for our 3D Orrery it’s possible to cheat it completely as we don’t need to be restricted by the real world limitations of cogs, but where is the fun in that? I thought a fun experiment would be to have functioning gears that drive the positions of the planets, as they would in reality.

Now before we get into the 3D side of things, I want to give a shout-out to Dragonator on the Instuctables forum. I based the solar system model around the design of his Orrery, and it functions in a very similar way, however our calculations are going to be slightly different. Let’s get started.

Part 1: The Moon and the Math

If math isn’t your thing, don’t worry. It’s simple and we’re going to take it step by step. We will go over it many times through this how-to. Lets first look at how cogs work in an orrery. We’re going to focus on the Earth’s orbit around the Sun, and the Moon’s orbit around Earth. We can control this whole movement with a really simple 2 cog set up.

Firstly, we need a value we can base our cog calculations on, this needs to be a whole number, as it will equal an amount of teeth on a cog. In our orrery this value will be 13. We’re going to use 13 because it’s how many times our Moon orbits us (Earth) in one year. Well, it’s precisely 13.336874634, so we have to round this number down to 13, as we want something easily divisible, and whole. You can reach much more precise numbers by adding more cogs of varying size, but this quickly gets complicated, and I’m a CG artist, not a mathematician or watchmaker! Keeping it simple, we’re going to need a Cog with 13 teeth. This first cog will sit underneath where planet Earth will be, and will drive the rotation of the Moon. We know the Moon will need to revolve 13 times as the Earth rotates around the Sun once, giving us our year. To achieve this, the 13 toothed cog will rotate around a larger cog with 13² teeth (169). (For your big cog, you need to have a number of teeth that your smaller cogs teeth can multiply to. This assures the cog teeth will perfectly intersect.)

From this set up, we can get Earth going around the Sun, and for every revolution of Earth, the moon will rotate around earth 13 times which is just what we were aiming for. Let’s have a look at how to set it up inside Cinema 4D.

Part 2: The 3D

Gif compression makes it appear the smaller cog isn’t rotating… It is, it’s just perfectly in sync with the frame rate!

Above is how the gears are positioned in the 3D scene, and below is the hierarchy in the Object Manager.

Firstly, create a Null in the centre of your scene. this will act as the rotational axis for Earth, and also the location of the Sun.

Next, create your first cogwheel spline and name it ‘Earth Orbit’ or something similar. Leave this at the centre of your scene, and set it’s orientation to ‘XZ’. You will need to do this for EVERY cog we are creating, so get in the habit now! Give the cog the 169 teeth we talked about earlier (13²).

Create a second cogwheel spline and name it ‘Moon Cog’, and give it 13 teeth (√ 169). Position it by adjusting its position in Z space, so that the teeth are just intersecting the gaps between the teeth of your larger ‘Earth Orbit’ cog.

Now we need to animate all of this. It doesn’t really matter how long the timeline is, but for simplicity’s sake I went for 360 frames, so we have 1 degree of rotation for Earth, for each frame in the animation. Note that if we are starting at frame 0, frame 359 is the 360th frame, not frame 360, as 0 still counts as an integer!

The Null that we created to act as the parent for our Moon cog is what we’ll animate first. You can rename this Null to something like ‘Rotate H’, as that is what we’re going to do next. Set a keyframe at frame 0 in the H rotation, then go to frame 359 and set another keyframe with the value of 360. Make sure the keyframes interpolation is set to linear, so we don’t get any acceleration or deceleration. All keyframes involved with cog movement will need to have a constant (linear) speed.

Now select the ‘Moon Cog’, and at frame 0 set a keyframe in the B rotation. Now go to frame 360, and in this same field, type in 13*360. This will result in a value of 4680, which equals 13 full rotations around Earth. Agian, be sure that the keyframe is set to linear interpolation.

The set up for animation is now complete for the Earth, Moon and Sun. We won’t be going into details about adding in the geometry, however if you do, simply apply these same animated parameters to the extrude on the spline itself, as otherwise at render time, the cog geometry will have to be re-calculated on each frame, rather than just on frame on, and could cause phong & normal shading discrepancies between frames.

Part 3: The Planets

Save what we’ve made so far, and open a new scene, so we have a blank canvas to work with.

Let’s get started on some of the other planets to see how they function. The idea in practice is the same, however we’re going to use some very simple Xpresso to automate the animation for us. These cogs will function more like a gear, so a smaller cog will be driving a bigger cog in the opposite direction.

Just as a heads up, you can handle all your Xpresso in one tag, or have an individual tag on each gear system, personally I’d recommend having everything in one tag to keep things simple, as you may want to be copy and pasting nodes to speed things up.

To start off, we need to do some calculations to work out how far around the orrery each planet will move throughout one year (1 rotation of Earth). Let’s first compile some data on how long a year is on each of the planets. I’ve compiled a simple graph to visualise the speed each planet will rotate for reference, and we’ll dive into the numbers below. The higher the number on the Y axis, the the slower the orbit.

Planetary Orbits in Earth Years

What I really loved about compiling the data was seeing that the graph was exponential, with each orbit almost doubling in time taken as you move from one planet to the next. Here are the actual values rounded to the closest 1st decimal place, other than Mercury. You can be as precise with these numbers as you want to be however.

Mercury - 0.25
Venus - 0.6
Earth - 1
Mars - 1.9
Jupiter - 11.9
Saturn - 29.5
Uranus - 84
Neptune - 164.8

These numbers will be used as a divider of 360° to get our orbit speed, which we will get into shortly. We again need to make 2 cog splines to drive the rotation. Lets start with Mercury.

Mercury Gear Set up, 10 teeth on smaller cog, 20 on larger.

Similarly to how we set up the Earth, Moon and Sun system, we need to find 2 numbers that divide into each other as whole numbers. However lets firstly think about how this gear is going to be moving.

From the data we have compiled, we know a year on Mercury is around .25 of an Earth year, meaning it makes 4 rotations around the Sun for every 1 Earth year (1/.25 = 4). Now we need to take 360 and multiply that by 4, this will give us how many degrees of rotation Mercury will make, which is 1440°. This gear system will be moving the fastest out of all the cogs, so choosing numbers to represent the cog teeth that a relatively similar will stop one of the gears from moving at a crazily fast speed. For my set up I went with 10 for the small gear and 20 for the large gear. I suggest you do the same to make following along easier. Make sure you have this set up with the smaller cog positioned so it has a tooth perfectly between two of the teeth on the larger gear.

Before we start with the initial keyframes, we need just one more piece of data. We need to find where the planets are in relation to Earth around the Sun. For my orrery, i looked up current positions of the planets on https://www.theplanetstoday.com/. I then roughly estimated how far around the orbit each planet was, imagining Earth is at 0°.

Planet locations as of 10/06/2019: https://www.theplanetstoday.com/ as seen from the Northern Hemisphere.

So, using the above image as an example, we’ll take Earth as our starting point (0°), and estimate that Mercury is 270° degrees around it’s orbit. Starting with the Large cog, keyframe this to linearly animate from 270° at frame 0, to 1710° at frame 359 (360*4+270) in the Rotation. H around it’s centre.

You can use this same approach to find the starting point for each of the planets. (For example, Jupiter would be 0°, Saturn would be 20° and Venus would be around 125°, and so on.)

Tip: To make the numbers a little easier to read, after setting the initial starting position for each planet’s larger gear, try freeze transformations. It’s important to do the same for our smaller gear too once its positioned.

Our smaller cog should ow be animated to rotate. Now it’s time to set up the Xpresso, which is very straightforward!

Tip: In the timeline manager, go to Functions -> Track After -> Repeat After. This will make the animation loop.

Part 4: Xpresso

So hopefully you’ve made it this far, and have your 2 cogs set up, with the larger of the 2 animated. This large gear is the one which will drive the orbit speed. We’ve cheated by animating the larger one first, as in reality it would be driven by the smaller cog. In the example above, a spline has simply been drawn and swept to act as an arm to hold the planet, and set as a child of the large gear so it inherit it’s rotation.

Create a Null at the centre of your scene, and call it ‘Mercury_Gears’ or something similar. Create another Null at the centre of the scene and call it ‘Mercury Xpresso’ or something similar. Drop this into the first null we created (Mercury_Gears). Now we can put everything else we’ve made for Mercury into this main null too, to keep everything tidy and easy to find.

Here’s how I had everything set up in the Object Manager.

Drop an Xpresso tag on the ‘Mercury_Xpresso’ null, open it up, and lets get started. I would suggest reading thorough the section below before trying it in practice, as you will have a better understanding of why we are doing certain things that way.

This is what our finished Xpresso set up will look like.

The first thing we want to do is drag our animated ‘Large’ cog into the Xpresso workspace so we can access it’s data. Dragging it into Xpresso will create a node for us to work with. Click the red box in the top right corner of the node, and add Coordinates > Rotation > Rotation. H as an output. Now we want to add a Math node (You can search for these nodes but clicking the magnifying glass icon on the left of the workspace, the node we want is called ‘Math:Add’). Once we’ve added this node to the workspace we want to change it’s function from add to Multiply, and connect the Rotation. H from ‘Large’' to Input [1] of this node.

Alright, lets hold here and talk about why we are multiplying this value. We’re going to need to go back to our cogs and teeth, in Mercury’s case, 20 for the large one, and 10 for the small one. We have to take our large number and divide it by the smaller one. The result of that division is the factor we need to divide our large gears rotation by to drive the smaller cogs rotation speed.

In this case, 20/10 is 2, so we must set the parameter for Math nodes Input [2] to -2. We MUST add the ‘-’ as this turns the multiplication into a division instead. The smaller gear has to rotate in the OPPOSITE direction to the larger one, at double the speed. This function is the same for all the planets, for example:

Let’s say we’re working with some larger cogs for a planet like Jupiter for this example. We will have to chose whole numbers that divide into each other with no remainder. Lets say our large cog has 77 Teeth, and our smaller one has 11. The simple calculation we have to do is 77/11 = 7. So our Input [2] for our Multiply node must be -7. Remembering the ‘-’ symbol to set the smaller gear to rotate in the opposite direction, and in this case at 7x the speed of the large one.

Now we understand the math, set the output of the Multiply node to drive the Rotation .H of the smaller gear, and optionally use a result node to check what the output value is. You can also add a result node to Rotation .H value of our ‘Large’ cog. You’ll notice the output rotation of our Math node is double the input, but negative.

You can now merge this scene into our first one and see how it looks with with them animating together.

Final Notes

And that is how the set up works! The process needs to be repeated for each planetary system. Remember you can use whichever numbers you’d like for the gear teeth as long as they divide into each other. The most important factor for accuracy is the degrees of rotation that a planet will make it around the sun, which we manually keyframe (check back to Part 3 if you need a re-cap!)

Hopefully this write up taught you something new, and you’re now equipped with the knowledge to create the rest of the planets and animate them with cogs.

Please leave any questions below, or email me directly at contact@jamestf.com