Table of Contents Show
Tempus began with a thought. A thought about time and the finitude of our lives: as human beings, though we are aware of time, we do not perceive it linearly. Often, we get so caught up in life that we don’t stop to think about it and enjoy the incredible, brief time we have on this planet while we are living.
The notion was heavily influenced by the passing of many beloved fellow travellers on the road of life who suddenly vanished in the last couple of Years. I could philosophise about it a whole lot more. But this article is about the technical breakdown: the phases the project went through, the challenges, and the technical aspects that made the final short.
I carved out one month of production time from my otherwise busy schedule to produce Tempus as a whole. Looking back, that proved to be both a boon and a bane. A boon because it forced me to distil everything down to the essence. And a bane because of the tight production timeline.
I am very grateful to Lukas Guziel, who composed the score, which gives this very personal piece so much of its depth and emotion. Ameen Taher for his amazing vocal performance; OTOY and the Render Network, especially Sunny Osahn, for making the project possible; and my Patreon supporters for their continued support.
Managing scope early on:
When I sat down to write Tempus, I noticed how hard and iterative the writing process is. As a 3D artist, I learned to find the right visual output for already existing stories. But shaping one on my own was a whole new experience.
In my first draft, I made sure not to hold back and brought every aspect I had in mind to the page. Of course, it became a rather long text that didn’t flow as well. It still needed a lot of brain juice and shaping to become what it is now.
To judge the scope of what it would take to produce the piece, I created a shot table from the written words, translating what I had in mind visually—going over my first draft and dividing it into shots created an 11-shot piece with vastly different scenes. Going into production on this would not have been possible within the timeframe.
Creating this many shots at the quality level I wanted to deliver bordered on impossible, so I needed to fix it elsewhere. The solution was as practical as it was simple: build fewer different scenes and treat them as “sets” that could be used for different shots so that I could revisit them in different states (e.g., a different time of day).
While this proved to be a good solution, I still needed to distil the spoken words into a smaller form. Even though I saved some work on the scene-building side, the read-out text would still span more than two and a half minutes. And as probably all of you readers know, producing that amount of CGI is really challenging.
So in the next couple of Days I focused on writing and rewriting the text. Distilling it so it keeps the meaning and impact I wanted to convey, but in a form a Speaker could deliver in about 1:20 min.
And after some more work, the final script at last:
And of course the now-reduced shot table that went with it:
It was an interesting new experience for me working as a script writer. The most important thing to me was that the script would be 100% finished when I first opened a 3D program.
It felt a bit weird at first spending so much time refining the script. But I knew that, in the end, a locked-down script would keep the production schedule on track, and decisions down the line would be measurable, since the foundation wouldn’t move. Also, having a 1:20 min short where I could re-use some scenes felt doable now compared to the 11-shot 2:30 min behemoth I started with.
Bending a canyon
Once the script and the shot table were tied down, it was time to open the 3D program and start piecing things together. The first thing I needed to do was to prove that the images I had in my head for the short actually worked. So I sat down and created the first rough animatic.
While there was lots to improve, the first “motion test” proved me right. The piece was working. As you can probably see in the animatic, some things stand out. Especially compared to the finished piece. One thing I quickly noticed was that the first shot was not working well. The camera moved too slowly and too linearly, and it didn’t create the engagement I was hoping for.
The fix was remarkably simple. Making the landscape more organic. By bending the canyon a bit towards the left. This made the camera move faster, as the camera needed to cover more ground in the same time, but, even better, created the much more interesting reveal of the gate.
Luckily, I had a hunch I’d need to tweak the landscape later. So rather than sculpt it, I used Cinema 4D’s procedural tools to “rig” it so I could easily adjust it later.
Kill your darlings
If you are familiar with my work, it’s usually mostly, as I call it, “tabletop-sized”. So working on vast landscapes was something completely new to me and therefore proved quite challenging. The emphasis shifts completely, and you have to start with very broad paint strokes to move everything into place. At the same time, I was working on the “Foundation Gate” scene. I rendered the current state every once in a while to put into a collage, which you can watch next.
In the video, you can see some red figures standing around. Those are real-world-sized humans that helped me get a good sense of the scale of the landscape I was working with.
One aspect of the gate scene I was really proud of was the riverbed.
After watching a lot of references, I re-created the nuanced feel of the water interacting with the landscape. I made the stones beneath the water surface slightly mossy brown and gave the water some greenish absorption and scattering to simulate the look bodies of water in high-altitude mountains often exhibit.
After putting a lot of work into dialling in the values, one thing became apparent. The camera angle and lighting hid most of the riverbed work from the viewer’s eye. I decided against changing the shot to show off my work and instead to “kill my darlings”. This is a term I often heard in the editing room, where editors sometimes have to sacrifice their favourite shot in exchange for an edit that better serves the story. That’s exactly how I felt with the shot and riverbed I was working on.
To honour the work, I might re-render the scene with different lighting, as it will probably make for an amazing shot in a future demo reel.
Temple scene
After creating the landscape, I went on with the temple scene. The workflow was a bit different here. Luckily, a temple is less organic and smaller in scope than the landscape. So for the temple scene, I opted for a workflow that created hero assets I could distribute and repeat throughout the scene. To add more atmosphere and visual density, I also added a volume.
I noticed that the landscape breakdown timelapse ended up a bit wild, since I didn’t care about continuity in the captures I took. Because the temple scene was smaller, I could use a fixed camera view to create my timelapse.
With the viewpoint locked, I could show successive additions against the same framing. As a result, the temple build is much easier to follow than my landscape timelapse. The documentation isn’t equally complete for every shot, though: I only later realised that I barely saved any renders for the 3rd temple shot, as I must have been really in the flow working on it.
Set dressing
Though I saved some production time by condensing the script and scene usage, one month for a whole short is still not that much. So I tried to save some time without sacrificing production quality. And what better way to save time than not having to model assets for the scene?
For the landscape, I used Quixel assets for the different grounds as well as the smaller stones scattered throughout the scene. I hand-modelled the gate and the “alignment stone,” then refined them with Cornelius Dämmrichs’ Edgy plug-in, which was a massive help in ageing down those big stone structures.
On the temple scene, I used the same logic as real set dressing for a production: I used pre-made quality assets I either bought on Gumroad (e.g., Max Bitiukov Procedural Book Rig for C4D) or brought Quixel assets from my local hard drive into the scene.
You can see me using a beta version of DaForge from artist Adam Zen. A plug-in that sources assets and material libraries from your HDD and turns them into a “one-click add to your scene” item.
One additional thing I usually do to make my scenes feel more alive is add particles. Either floating (which I call “floaties”) for wide shots or sitting on the geometry “dust” for close-ups. Rather than moving the floaties with an airflow simulation, I use Cinema 4D’s Cloners and Effectors in a combination I have refined over the years, creating movement behaviour that comes close to dynamically simulated particles.
While real particles advected by an airflow simulation would probably be a bit more realistic, I wanted to avoid the time-consuming step of caching. And since I already knew I would render the shots on the render network, I avoided uploading large cache data through my slow German internet connection.
A gradual sun
To add more mood and gravitas to the scene, I fell back on a technique I learned quite a while ago. Creating more saturated half-shadow areas. To do so, we need to give the sun a gradient.
While Octane’s sun and sky system is quite flexible, it does not let you use custom shaders on the sun emission. So I replaced the procedural sun with an Octane area light coming from the same general direction as the original sun. Once done and adjusted to the same brightness level, adding a gradient from orange in the middle to red on the rim to the “sun” will produce those moody red-saturated fringes on the shadows. While not realistic in the real world, it’s a technique even old masters used in their oil paintings hundreds of years ago.
The gradient lets me shape colour across the emitting area and introduce a red transition within the soft shadow boundary. A uniformly coloured light would tint the illumination as a whole; the gradient helps here because I want the penumbra transition to contribute to the mood.
Note about the videos above: In addition to adding a gradient, I also made the custom area-light sun redder overall, so the two examples aren’t colour-matched. You should still be able to see that the penumbra area in the 2nd example is more saturated and red.
Fine-tuning lighting
If you follow my tutorials, you might have noticed that I have a very special way of multi-layer compositing. Rather than using the usual workflow of adding the material element passes as Diffuse, Reflection, Refraction, SSS, etc. together, I generally separate the different lights in my scenes.
This gives me a much more useful workflow, since I usually don’t need to change a material’s look; I just want to tweak the lighting. I use a floating-point workflow with linear images. I generally use ACEScg .exr files. This gives you the flexibility to fine-tune the lighting or completely alter it if needed. And all of that without re-rendering the scene.
Small technical insert about .exr’s, floating point, colour spaces and light passes:
For anyone building a similar set-up in OctaneRender, OTOY’s beauty and lighting AOV Guide explains the general reconstruction. A complete lighting stack accounts for the contributions that form the beauty image. A combined Light AOV contains its direct and indirect contributions; use that combined pass or the two components, because adding all three counts the light twice. Light Pass IDs can separate emitters or group several sources for joint control. The Guide is a useful reference for setting up that separation and checking the resulting reconstruction.
This workflow depends on scene-linear floating-point images. Linear sRGB and linear ACEScg are possible working spaces; 16-bit half-float and 32-bit float are numerical storage precision options. They are separate choices. In linear-light values, doubling a contribution doubles the represented light, so you can add passes and rebalance them meaningfully. OTOY’s Guide to linear workflows provides the wider colour-management background.
For a practical file-format reference, the OpenEXR technical introduction distinguishes HALF, a 16-bit floating-point channel type, from FLOAT, a 32-bit type. Its scene-linear documentation also makes a useful warning explicit: an EXR file is not guaranteed to contain linear-light data. A reliable hand-off therefore needs both the channel type and the interpretation of the stored values to be clear.
As you can see in the demonstration above, I use my light passes to alter the completelyscene’s look by adjusting the strength of the single lights. The output of the “remixed” lighting below looks like a completely different shot, and all of that without the need to re-render.
Light passes in action
While I used the light-pass workflow in all the scenes of Tempus to balance and make final tweaks to the lighting, the hourglass scene is a more drastic example. I lit and rendered this scene with a strong red backlight because I thought it would make sense narratively, in connection with the red slit of light seen through the marker stone one shot later.
As it turns out, some things that should work narratively do not work visually, and the red light looked awfully out of place. So rather than re-rendering, I turned off the red-light contribution in the comp and ended up with a much more fitting shot.
You might notice that clip 1 is not the same as clip 2. This is because I changed something in the sand simulation later on and therefore needed to re-render the shot. The good thing is that I only changed the sim and didn’t fiddle with the lighting. So the re-render still included the red light in case I wanted to reintroduce it later. Which, as you see, never happened.
More light passes
To show you the light sources I would use to light my scene, here’s a breakdown for the different light passes I would put together in comp for the first temple shot. Most of them are outside the field of view and light the scene. But some are also visible in frame, like the dim incense glow and the candle flames, which were a last-minute addition.
In the next temple shot, which symbolises a passage in time, the sun has set, and it’s night, so the lighting is different. Personally, I feel night shots are always harder to light, since a real night shot would be pitch black. In film, we have to invent lights to make the viewer see something but don’t make the light sources too obvious and directional. It’s a balancing act, and another great use case for the multi-light comp workflow to tweak and polish each light’s exact contribution to the scene.
For the last temple shot, the lights changed a bit. The key light now comes from the left instead of the right. One could see it as a continuity error. But overall, I made it this way to create the most impact with my lighting. Also, if the lighting keeps the overall continuity in colour and tone, almost no one will ever notice!
Finalising the shots
For the compositing workflow, I used Blackmagic Fusion Studio and its node-based workflow. The following sequences go through the different steps of the compositing tree. From the raw rendered image to the final colour correction that you can see in the finished piece.
In the lower third of the image, above, you can see the flow (left to right) of the nodes that make up the compositing and, therefore, the final look of the shot.
Going through the accumulation of light passes and effects of frame 125 of the Gate A shot. I especially like the halation effect, which I created to simulate the strongly colored glow that happens when light moves between the layers of real chemical film stock. Together with glow, grain, vignette, and edge blur, I use it to make the image feel less digital and give it a more analogue appearance. Below, you can see the same process for frame 250 and frame 575 of the same Gate A shot.
Here is a very similar breakdown for the temple shot. Of course, it uses a slightly different mix since the lights are set up differently and the shot therefore needs a Shift in balance. But as you can also see, most of the main ingredients, such as the halation, glow, grain, vignette and edge blur, are back to form a very similar smooth analogue-looking outcome.
For the follow-up shot, I reshuffled the light mix. But our main ingredients are back. In the night shot, you can see that I use light mixing more to tweak the shot to fit my vision and match the overall look of the surrounding shots. This also applies to the following close-up shot, as the lighting situation there is very similar.
P.S. If you have keen eyes, you might have noticed that I left the candle lights on in the light pass breakdown above. This was another case of uncertainty about whether the shot looks better with the candles lighting up the space a bit. In the end, I decided against it because the narrative impact was greater seeing the cool deserted space without warmth.
And last but not least, very similar to the first gate shots, the closing shot of the short.
Again, all our trusted ingredients are back, just slightly tweaked to match the new perspective and light angle.
Again, if you paid attention to the breakdown, you can see that the light coming through the slit is split into an extra light pass. I did that with a spotlight and some Octane Trace Set tricks. Having the “slit of light” as an extra controllable element was important for the short, as it is the main point that brings the whole narrative together. So I naturally wanted to have full control. And indeed, I ended up making the contribution a bit stronger than it would have been, to balance out the shot. This pass also helped me with the second “Soundtrack Only” version where everything but the light coming through the slit is black and white, seen just below.
A different relationship with the score
I also made an alternate version to give Lukas Guziel’s music the space to stand on its own. Presented as a black-and-white, soundtrack-only version, it removes the voice performance and changes the relationship between the score and the images. A few changes can give the same material a surprisingly different character.
Overall, I managed to pull off creating an engaging piece of work within the
self-imposed deadline. And though it proved challenging at times, it gives the viewer a much more emotional and personal view of my heart and soul than my past work. It taught me a lot about writing, narrative, and large-scale scene creation, and it helped me tackle story-based projects more confidently in the future.
| Project | TEMPUS: A Reflection on Time, Transience, and Memory |
|---|---|
| Artist | Raphael Rau / Silverwing-VFX |
| Format | Animated short; main YouTube upload runs 1:47 |
| 3D / rendering | Maxon Cinema 4D / OctaneRender |
| Final shot count | 7; reduced from an initial plan of 11 |
| Compositing | Blackmagic Fusion Studio |
| Asset workflow | Locally stored Quixel assets imported with a DaForge beta by Adam Zen |
| Floating particles | Cloner and effectors; no airflow simulation for these elements |
| Music | Lukas Guziel |
| Voice performance | Ameen Taher |
| Project support | OTOY and Render Network; Sunny Osahn acknowledged |
| Behance project date | 9 October 2026 |
About the author
Raphael Rau is an artist specialized in photorealistic physically based computer generated imagery / animation based in Ludwigsburg, Germany.
His work ranges from art direction to creation of logo animations, pack-shots and CG shorts.
Topics
3D & CGI736 postsBuild what the camera cannot find. Explore computer-generated imagery, digital worlds and production techniques that take an idea from its first rough shapes towards a convincing final frame.
Lighting180 postsSet the mood before anyone says a word. Explore digital lighting, scene illumination and practical approaches to guiding the eye, shaping depth and making an image feel deliberately lit.
Rendering887 postsThe scene is ready; now the pixels have work to do. Follow rendering engines, optimisation and production workflows that balance image quality, iteration speed and the available computing budget.
Compositing504 postsBring the pieces together and hide the seams. Explore keying, tracking, integration and node-based workflows that help live action, digital elements and effects convincingly share the same frame.
Workflow759 postsKeep the handoffs from becoming cliffhangers. Explore pipelines, collaboration and production practices that connect people, applications and assets, making room for creative work between the deliveries and revisions.
Blackmagic Design Fusion135 postsConnect a few nodes and the picture starts telling a different story. Explore Fusion compositing, motion graphics and visual-effects workflows, including techniques for combining sources into a coherent finished shot.
Article details
- Published
- Author
- Raphael Rau
- Reading time
- About 17 min
- Word count
- 3,306