For the last three years, Emissive have been developing and refining The Horizon of Khufu, a VR ‘immersive expedition’ that allows people to visit the Pyramid of Khufu. The experience allows visitors to explore sections of the pyramid interior, and recreates historical scenes to provide a view of everyday life in Egypt four and half millennia ago. To date, the VR experience has received more than 50,000 visitors, and is currently open at Lyon Confluence. Pierre Gable is Creative Director at Emissive; Guillaume Duflos is Concept Artist and Senior 3D Artist at Emissive. Here, they break down their objectives and motivations in creating this VR experience, and detail some of the practical elements of their work.
Pierre: I’ve been working with Emissive on various projects around the Pyramid of Khufu, and the surrounding necropolis of Giza, for nearly 16 years. Early on, I frequently worked with Peter Der Manuelian, an Egyptologist and professor at Harvard University. In one notable case, a partnership with Dassault Systèmes allowed us to reconstruct in 3D the mastabas, the small tombs around the Pyramid of Khufu. These were incredible learning experiences for me – Peter is fantastically knowledgeable about Egyptian history, and this really showed me that when you’re working on this kind of historical reconstruction, precise details are vital. You need to consult expert historians on every point. If you want to build a wall in 3D, or even just to paint a wall, you have to speak with the historians first, and they’ll explain what sort of methods and materials would have been used in that period.
The Origin of the VR Experience
Later on, in 2016 , we were fortunate enough to work with a non-profit scientific mission called Scan Pyramids, who were using muography – that is, scans for muon particles – to detect open spaces within the Bent Pyramid of Snefru, then the Pyramid of Khufu. We provided 3D experience on this project and so, to be able to provide an accurate 3D mesh of the pyramid, we visited the pyramid to take photos, and scan sections of its exterior.
And once we had all this data, and being specialists in VR, we began to think about how the pyramid could look through a VR headset. This was the origin of the Horizon of Khufu VR immersive expedition.
The tent from which the expedition begins.
We had a couple of main objectives in creating this experience. First, we wanted to be as historically accurate as possible, with the data available today. We’ve been in constant contact with historians throughout this project, and I think we’ve had some success overall. At any rate, we’ve been visited by Egyptologists from the Louvre and elsewhere, and they’ve been satisfied with what they’ve seen.
Second, we wanted to tell the story of the people who lived in the period of Khufu, two and a half millennia before Christ. In the experience, we focus on the human side of that time, and on the feelings of the people in that place, who were similar to us, today, in so many ways. We don’t talk about the treasure of the pharaohs; that wasn’t important to us. We wanted to tell the stories of the people of that time.
And VR is an excellent tool to communicate in both areas. When you’re physically inside the pyramid, in real life, it can be difficult to visualize exactly where you are, or how the different volumes of space within the pyramid interconnect. With VR, we can provide ways to appreciate this sense of space more clearly.
And, indeed, VR allows us to transport the visitor back through time to see the lives of the people of that time, and to better understand the challenges they faced – in the VR experience the visitor travels back in time, for instance, and witnesses a funeral ceremony. These are of course the kinds of ordeal that we still experience today.
Historical Accuracy
The most exciting part of this project for me is to provide a plausible view of the past, supported by scientific fact. I also find it very rewarding to work with historical experts – they don’t typically have access to 3D models, or 3D technology, and so we find that historians are often happy to collaborate with us, because our 3D expertise can help them with their research, and help them to make new hypotheses. It’s a win-win situation.
Early on in the project, we were thinking that we could use the models of the pyramid for teaching – and this has happened; Peter Der Manuelian is using our models in his courses at Harvard University. But historical accuracy was important as a matter of principle, as well – just because you’re intending to make the experience available to the general public, and even accessible to children, doesn’t mean that it’s okay to provide an inaccurate representation of the pyramid. Frankly, when we were constructing the experience, there were times when it would have been more convenient to us to add a staircase, or move the walls, or create an additional room in the pyramid. But we never misrepresented the facts that we had . We wanted the experience to remain as faithful to reality, and history, as possible.
In the VR experience you also see the area around the pyramid as it was thousands of years ago, and what you see is absolutely based on the latest research available. Even just a few weeks ago, new historical research was published concerning the Giza area. Today, the Nile is in the middle of Cairo, eight kilometers from the pyramid; flooding no longer occurs there thanks to the construction of the Aswan Dam in the 1960s . But four and a half millennia ago it was much nearer the Giza Plateau. This latest realization concerning the river’s position came by analyzing the pollen of the plants in that area – the presence of certain types of pollen reveals the location and level of water at various points in history . The methods that historians use to uncover the secrets of the past are remarkable.
In cases like this, where new data becomes available, we’ll update the VR experience accordingly. These are often subtle details that a typical visitor to the experience won’t notice – but, because the experience is used for teaching, and because we sometimes use the experience to create pictures or videos that might appear in documentaries, it’s important to maintain that level of historical precision.
First Steps of Creation in 3D
Guillaume: The first stages of creating the recreating the pyramid in 3D involve bringing together as many reference images as possible, and planning out how the scene should appear. This is the part of my role as a concept artist; because the pyramid of Khufu is already so well known, I have to think about ways to step outside the ‘classic’ image of the pyramid that everybody tends to have.
We’re fortunate in that we’ve been able to scan sections of the pyramid, using both laser scanning and photogrammetry. Using the data from laser scans, we can create accurate HD 3D meshes, along with some texture information. We also use photogrammetry in the same areas, to capture data allowing us to enhance our textures and materials. When scanning, we mostly focused on rocks and stones, and the summit. The summit in particular is tricky to scan, as the scanning is often carried out from the ground.
Of course, the pyramid is just so enormous that it’s difficult to scan everything. The entire surface of the pyramid has been scanned by drones, at medium range – but even in this case, the resolution isn’t as high as we’d like.
And so our scan data is a starting point for us, but we can’t just directly send that information into the app. We have to rework that information enormously. For instance, any textures that we scan will include the lighting conditions on the day the scan was carried out. In such a case, we have to remove this lighting data as much as possible, so that we’re able to realistically display the texture in a range of situations – in direct sunlight or in shadow, for example, or during the night.
The Substance 3D apps are helpful here. They’re very practical for dividing up the different channels that we need to work on. We can separate out things like the reflectivity, or the normal map, as needed.
Scan-based Assets Versus Assets Created by Hand
Our scan-based media reached an excellent level of photorealism, I feel. At the same time, the project contained a lot of assets that we had to create ourselves, from scratch. One of the challenges of this project was therefore to ensure unity between the level of quality of our scan-based assets and those we’d crafted by hand. We had to ensure that visitors to the experience wouldn’t be able to tell the difference between those two types of assets.
This notably applied to all the interactive elements of the experience, as well. If you think back to old cartoons from the 80s or 90s – any of those old Hanna-Barbera cartoons, for example – if you see a bunch of rocks on top of a cliff you can see right away which one’s going to fall. We wanted to avoid that visible difference in quality between the general scenery and the interactive elements of each scene.
Optimization
Optimization is the biggest element of our work. Scanning just a part of the pyramid, we’ll come away with a scan of around 20 to 30 million polygons – far too many for the app to work with. To show one whole face of the pyramid, we have to bring that number down to about 30,000 to 40,000 polygons. This means we have to extensively rework each scan, and create by hand any parts that we haven’t scanned.
And creating for VR, you have to keep in mind that your scene will be rendered for two screens within your headset at the same time – one for each eye, with a slight difference between them to allow for binocular vision. This is the big difference between a VR scene and, say, a video game displayed on your desktop: VR requires twice the rendering power. So it’s especially important to pay attention to elements such as the number of polygons used, and the size of your textures.
A lot of effort goes into optimization, and into trying to make things look good, using few resources.
The Emissive Workflow
Guillaume: The workflow for creating a scene for VR is essentially the same as for any other 3D scene, with the extra consideration mentioned above, that each scene must be rendered separately on two screens at once.
First, we’ll create an HD model. We might sculpt that completely from scratch using ZBrush, or we might start from photogrammetry data. Ultimately, we have a ‘clean’ model, comprising a fairly high number of polygons.
Next, we’ll create a low-poly version of that same model, carrying out a manual retopology in ZBrush or 3ds Max. The final low-poly version corresponds exactly to our HD version.
Then we’ll unfold the low-poly object’s UVs, essentially laying out the 3D object as a flat 2D shape, for texturing.
We’ll create our textures in Substance 3D Designer or Substance 3D Painter, either from our photogrammetry data or completely from scratch. We’ll typically create a ‘clean’ tiling material in Designer, and carry out any painting in Painter – even when we’re working from photogrammetry data, we might repaint various rocks, or add dirt, and so on.
Then, with Painter, we project the HD mesh onto the low-poly object. This allows us to apply the normal maps, and any sculpted details we’d created previously, while keeping the file size of the object very light.
Filling a Historical Gap
Pierre: In cases where precise historical data just isn’t available, the historians we work with are keen that we avoid representing any parts of the pyramid that require an element of guesswork – even when we’ve studied those areas in great, great detail, and those guesses are probably accurate. Their suggestion would typically be that we show the areas where we lack some data in magenta, or another bright color. But of course that’s not very interesting for the more everyday visitor to the experience. So some parts of the pyramid are recreated based on our best estimations. We always research such points extensively, however; we never make such decisions lightly.
For instance, towards the end of the experience you see the funerary temple, and witness the mummification process. Really, we aren’t one hundred percent certain that the purification and mummification process occurred in this small room within the temple – almost no part of this temple exists today, so it’s difficult to be absolutely sure about any part of it. Here again, we depend on the opinions of the historical experts, like Peter; in this case, Peter told us that the mummification could have occurred here. And this is how he presents it when teaching; he’ll say that it’s possible that the mummification occurred in this room, but that we aren’t sure. The alternative, if we were to only show scenes and places that require no guesswork at all, would be to omit this scene entirely. And that would be unfortunate.
Guillaume: In these cases, when we’re recreating areas for which we’re missing information, the first step is to get as many reference photos of that area as possible, to allow our artists to make informed creative choices. The pyramid is a popular tourist destination, and so much of the exterior of the pyramid has been extensively photographed. But that isn’t true for every part of it – the summit is a good example here; this area isn’t accessible to the public, so in recreating the summit we were forced to work from a very small number of black and white photos, and from aerial photographs taken from far away. That isn’t ideal.
In these situations, the artist makes creative decisions to ensure the scene is credible overall, even if not 100% factually precise. We know, for instance, that there’s a lot of graffiti at the summit of the pyramid – some of the graffiti dates from a couple of centuries ago, other parts are just fifty or so years old. But information on what has precisely been written up there is so scarce that we haven’t been able to recreate the graffiti exactly. In this particular case, we’ve allowed ourselves to have a little fun, while maintaining the overall credibility of the summit scene of the experience. The vast majority of the graffiti at the summit is very close to what truly exists – but if you look very closely, you can spot a few of the team members’ names in there, as well.
Getting references for the interior of the pyramid, and for the scenes that take place in ancient history, is also tricky. The Egyptians painted their statues, for instance, and it’s a challenge to try to determine the colors of a statue when you might have just a photo of the statue’s base, with just a trace of one layer of paint on it.
The goddess Bastet.
The historians that we work with are an immense help, here. They’re often able to point us towards other, similar sites that are better preserved, allowing us to see things like more accurate colors, or more intact pottery and other artifacts.
Pierre: Actually, pottery is one of the most precious assets allowing archaeologists to put a precise date on discoveries. And this is something that we paid close attention to, and on which we worked closely with Peter; all of the pottery within the pyramid of Khufu is historically appropriate, to within a period of about fifty years, no more than this.
These small details are extremely important to me. It’s when you recreate the right pottery, say, and place it on the right table that you start to bring back some of the memories of how the ancient Egyptians lived. A vital part of our work is to show the everyday, very human moments that comprised this incredible civilization – and even to evoke how we, today, might connect with them.
Balancing Historical Accuracy and Visual Appeal
Guillaume: Balancing the need for historical accuracy in the experience with the need to create something that’s appealing to the general public can be complicated. In creating the experience, when I’m working with the Artistic Director of the project, Maxence Fournier, our goal is typically to make something that looks outstanding for people in the VR experience. But when we go too far in that direction, we often risk veering away from the historical truth.
When creating the interior of the pyramid, for instance, we started to position well-crafted blocks around the edges of the pyramid – detailed, good-quality blocks from the Tourah quarry. But the historians connected to the project stepped in, telling us that the these blocks were only used in the center of the pyramid, and for its casing stones ; the edges of the pyramid were built from more common, lower-quality blocks. So we went back and redid those sections. In my opinion, the lower-quality blocks don’t look as impressive – but that’s the reality of the pyramid of Khufu.
In the scene where the visitor travels back in time, and views the funeral scene, we knew that the interior of the temple had been painted, but today nobody in the world knows exactly what colors had been used. The historians connected to the project advised us to just leave the stone unpainted in that scene, to avoid inserting incorrect, and invented, information. But that isn’t as visually interesting for visitors. Ultimately, we made a compromise of sorts, making our best guesses with the information available. As mentioned above, we had a pretty good idea how the statues had been painted, so we painted those, and we left the rest of the temple as unpainted stone – though we added some fine net curtains, as decoration.
The Challenges of Creating the VR Experience
Pierre: For this project we needed to faithfully recreate the interior of the pyramid as a space that people would physically move through. We had to respect the constraints imposed by the historical reality of the pyramid while ensuring that visitors have an immersive experience. In VR, it would be immersion-breaking if people walked through walls, for example. So figuring out the ideal route through the pyramid for visitors was a bit of a puzzle.
In total, there are 17 different scenes in the experience, with around 60 different stages for the visitors to go through. Ideally, we don’t want groups of visitors to the experience to be too close to one another, either – the concept of the experience is that it’s a VIP visit to the pyramid, with Mona, your personal guide. If you encounter too many other visitors, that can also break the immersion.
To find the best route through the experience, we developed a tool that allows us to simulate the various different paths, and draw a map of how to best guide groups of visitors through the experience. This involved a lot of trial and error. It took us about a full week to establish the path through the experience that we currently use, and that may continue to evolve as the project progresses.
Guillaume: Another big challenge of the project, I feel, was simply finding that core conceptual element of creating something that was original, and visually attractive, and that communicated what we needed to say about the building, and the people who inhabited it.
The scene when the visitor first explores the inner core of the pyramid was particularly difficult for us. We didn’t have a lot of visual references to work from here, so it was hard to imagine how that section of the pyramid would look – down to the level of details such as how the specific blocks of stone had been cut and placed together. And the visitor moves around a lot in this section, which adds another level of complication. It took months to put that scene together.
The Legacy of Khufu
Pierre: Peter once told me, ‘Look at this CD, or at this USB key; in a hundred years, or two hundred years, its data will be gone. These people in Egypt carved their messages in stone thousands of years ago, and we’re still examining them. That’s the medium to use, if you want to transmit information.’
And that’s true. If you want your story to last, carve it into stone.
All artwork courtesy of Emissive/Excurio. The Horizon of Khufu, a VR immersive expedition, is currently open at Lyon Confluence, and will remain there until March 2023.
