The Uncanny Valley can be a frightful place when you’re creating photorealistic 3D models of people. But with a talented team and a lot of cameras, you can create digital humans from all walks of life.
Digital Reality Lab took us behind the scenes of their studio in Bulgaria where they specialize in character 3D scanning services for VFX, video games, e-commerce, and more. But how many cameras does it take to get an accurate pose scanned? And what does that camera rig look like? (Spoiler alert: It’s big.) Keep reading to find out.
Can you tell us when Digital Reality Lab was founded, how big your team is, and what made you want to specialize in 3D scanning services?
Coming from a 3D artist background, we have always been fascinated by digital humans and the high fidelity needed to surpass the Uncanny Valley. With that passion, we created Digital Reality Lab in 2020 to work towards achieving accurate photorealism and helping other businesses benefit from it. Since then, we’ve grown and are currently a team of ten talented artists and managers.
Your TurboSquid store page primarily has scans of people from almost all walks of life. How do you choose what to add next to your store selection?
Our products on TurboSquid represent a small number of scans we’ve acquired during the last three years. We have a database with more than 25,000 models, and we choose the most interesting and practical ones to put into our store selection.
Can you take us through the planning and production process from scanning a human body to having a usable 3D model that you can sell on TurboSquid? What 3D modeling and rendering software do you prefer to use?
Everything begins with casting the right models for a scan. Usually, we create castings based on requirements around ethnicity, age, gender, and body type. Then, review all candidates and choose which models match the best for the project. Once the models are selected, we schedule a date for them to be scanned.
Then they come to our studio housing our full-body photogrammetry rig and we review any legal documents required for the project. We practice poses, expressions, and clothing before we proceed with the scanning.
Once we have those covered, we hop into the scanning rig and capture 113 photos for each pose, expression, or clothing we do. From then on, we process the images, load them into RealityCapture, and generate a 3D scan of the person.
This is where our artists take over and do cleanup in ZBrush to make sure any imperfections from the scan are removed, and in cases where a highly detailed closeup is required, we add extra detail from the photos.
Once we have those, we create an easy-to-use topology and transfer all the textures from the initial scan to the new topology, which has a nice and clean UV layout for editing. Lastly, our artists create the PBR textures in Adobe Substance 3D and output high-quality mesh and textures that can be loaded into any software.
Recently, we have been using Unreal Engine to evaluate our models, but we also like Marmoset Toolbag.
Are there any challenges finding certain body types to scan with Digital Reality Lab being located in Bulgaria? If so, how do you solve those problems?
Being based in Europe poses a challenge for us in terms of finding a diverse range of ethnicities and ages. For example, one of the most difficult castings we ever had to do was for people of African descent who are 60+ years old. That demographic is rare in Bulgaria, and only some people want to be scanned.
Our solution to this problem has already been designed and is currently in the works — make the scanning rig mobile. This will allow us to easily travel anywhere in Europe and make the casting process more accessible as we can cast more people.
What kind of creative challenges does your team generally encounter in their day-to-day job of scanning and turning the scans into 3D models?
The main challenge is capturing the person’s essence and showing it digitally. We strive to make the model as realistic as possible by capturing every human’s unique imperfections and transferring them to the digital 3D model.
Are there any industries and trends you watch to predict what you’ll scan next?
Yes, we closely follow AI and computer vision trends to see how major companies think and what directions they want to develop in. This gives an idea of what might be needed and what services and products we can provide them.
Can you talk about any of the media properties you’ve contributed to such as in movies and/or video games? Are there any projects you’re particularly proud of having worked on?
I can’t talk about movies and games. Still, I can share a project that we worked on with Datagen, where we developed a casting pipeline specifically for them to help them acquire an extensive amount of high-quality scans of humans that they will use for AI development.
The scale of the project challenged us to think outside the box and do a lot of casting to find the right people. Conversely, we had to develop many protocols for streamlining a large volume of consistent, organized, and accurate data to the real world.
Can you describe your experiences working with other professionals in game development, archviz, and/or animation?
We specialize in working with software companies that develop 3D tools and applications. However, we do have experience working with archviz and creating entourage for those scenes. Those required us to scan a lot of people in seasonal clothing and various poses of interaction—people sitting on a bench, riding a bike, handshaking, waving, pointing at something—that would look realistic when put in the scene.
Do you generally do the animation rigging for video game developers?
We do provide animation and rigging for video games or VFX productions. We have a dedicated team for grooming, rigging, and animation. All of those projects we take as custom requests as each company has its own workflow that we follow.
Additionally, we have been experimenting with rigging our full library of A-posed characters which is about 1500+ models. Hopefully, we can have a product to show you early next year.
How is Digital Reality Lab currently using and/or experimenting with AI? Are there any other technologies or techniques you’re considering adding to your workflow?
We are testing many AI tools that come to the scene, and we are trying to use them in a way that will allow us to scale up and produce better solutions for clients. For example, we have been working on a pipeline that would generate data of the same high quality as a real scan but completely synthetic. There are a lot of astonishing AI developments out there, and we are just starting to see the impact of those on the industry, which makes us very excited about the future.
Have you learned any good takeaways from selling your models on TurboSquid?
Yes — people love bundles 😀
Joke aside, we learned that many people who buy 3D models want to do some work on them later, so they want well-organized, easy-to-edit, high-quality models. We aim to provide the best quality for a reasonable price that everyone can afford.
Do you have any advice or recommendations for people who would like to specialize in work like your team does at Digital Reality Lab?
I advise artists to follow their passion and work on it for fun. Every time I work on something I find fun or fascinating, later in life, I get the chance to work on an actual commercial project involving that type of work because I had fun with a personal project before.
So, whether you need pre-made models with consistent quality, a diverse range of ages and ethnicities, or just need to save some time on a project with a quick turnaround, photorealistic 3D models like the ones Digital Reality Lab creates can be a no-nonsense solution.
Find all of Digital Reality Lab’s models for sale on TurboSquid and learn more about the company at digitalrealitylab.com.
Cover image by Digital Reality Lab.
Looking for more photorealistic 3D models?