CGTrader Interview: Tristan Bethe And The Long Road To 3D Scanning

Tristan Bethe is working on 3D scanning, but he had a wild ride before he got to be where is he now. What was his journey towards doing something we only imagined in sci-fi? Find out in the interview!

How did you get into the 3D art sphere?

Between ‘94 and ‘98 I was attending a school for graphic design here in the Netherlands. I liked to draw and this seemed to be the matching education some choice. I did some self assessment and I came up with the idea to specialize a bit by combining it with another hobby of mine: computers! I learned Photoshop and immediately started to think in layers. That was great. I could split my drawing into a background, characters and other elements. Today,I see that it only made me postpone learning I actually needed in any artistic profession. But at least I could fiddle with software and plugins for a few years, which is way easier to learn.

Luckily, I could intern at one of the few 3D animation companies iwhere I could play around with 3Ds Max and learn things like why glass objects looked so fake (no raytracing yet). Together with a classmate, we actually managed to create an animation on live action footage about paper items that had enough of being used by humans and decided to rebel. It was all hopelessly over ambitious, of course: animation on (analog) video, recording dialog and trying to match the lips, rigging and skinning.

Did a few years of creating CD-ROMS and Internet sites. Together with 2 friends, we set up a site called imageafter.com that was giving away free images and textures. Doing that I really got hooked on creating digital assets.Then Splutterfish’s Brazil was released and introduced one of those features I didn’t know I was missing: global illumination! A technological fix to let images look nice again! It was enough to get me back into 3. While working full-time in 3D illustration it started to dawn on me what was missing from my images. All those things that cannot be solved by plugins. It was really hard for me to look beyond the technical capabilities of a piece of software. Creating realism was still my passion and that was getting easier by more powerful render engines. Spent 3 years making illustrations for the cosmetics industry and that image with the robo hoppers in quiet times.

How did you start your 3D scanning business?

After the before-mentioned jobs I started working at a new company where I did a lot of architectural visualization. I remember a manager asking us "guys, why is it so hard to get good looking 3D people into the visualization?" None were good enough. So like lots of companies we tried to talk the clients out of using people at all. Then it clicked with me: this is a great problem to solve! And exactly the kind of digital asset business I was looking to offer after I missed the boat with imageafter.com. It could solve an existing problem and the products would be valuable if done right. The idea was to create assets that an artist could use with minimal tweaking. I invested in a ridiculously expensive 3D scanner that cost around $40k. Although it was sold as a full body scanner, the quality was almost less than what you can now achieve with an iPhone or a Kinect.

To get some quality out of the thing I bought a Next Engine laser scanner just for head scans. It involved seating the person on a chair and making 8 scans of about a minute per scan. But the texture quality of both scanners was pretty bad, so I also took DSLR photos during the scan; I projected those photos on the mesh in Mudbox and painted them in by hand. After doing about 14 models in a few years on my spare time I realized this was taking too long and the poses were pretty static. In the meantime I discovered photogrammetry through the work of Lee Perry-Smith, but I was in a bit of denial about it because of the huge sum of money that I had paid for my current scanner. I created a sort of scanner that used photogrammetry. Mark Florquin, also an active member on CGTrader, was of great help. He showed me how to scan using only 6 or 8 cameras and a rotating table. I was amazed by how much better the quality was with just a fraction of the costs.

We do 3D scanning mostly for creative companies. We decided to focus all of our attention on creating really eye-catching 3D people and trying to improve anything the arch-viz people needed. We consulted with brilliant artists like Bertrand Benoit, my personal favorite in the 3D architecture field. Hair was high on his list of things that needed improvement. At the the time I had no idea how to do hair in a somewhat timely manner. We hired an artist, an old colleague of mine, Kevin Beckers ( we do all the production together noew). Kevin is a guy who always tries out the newest software and he enabled us to do 3D hair. We changed the way we do scans from then on. We first scan the model with his/her natural hair so we have a reference of how the hair falls. Then, we redo the shots with the hair in a net. This enables us to easily make a bald version of the person and it enables us to grow the digital hair. The hair is then converted to polystrips with opacity maps so it can be used in all 3D apps and is not dependent on any plugins.

What made you believe that 3D scanning is the future?

I do not think 3D scanning is the future. That would suggest that it replaces whatever we are doing now. I really see it as an extra tool 3D artists now have in their toolbox. With software like Agisoft’s Photoscan or Capturing Reality’s RealityCapture every artist with a good DSLR can now sample the world around them. Made a cute forest gnome? Why not sample some forest floor or moss for the background or tree branch to let him sit on? It is about time we see realistic pillows on couches, tasty looking food on tables and, of course, stylish 3D people in our (architectural) 3D scenes.

What do you, as a 3D designer think, about 3D scanning?

I think it is a great development! I am no modeling purist, I am just concerned with the end result for my clients, and that helps. Having done both now 3D modeling and scanning, I can tell you that it is far less simple than outsiders think it is! It is more like modeling with training wheels. There's a lot of cleaning with messy 3D scans and doing a re-topo so you can actually use the scan. There is always occlusion in a 3D scan - parts that cannot be seen by any camera and that still require traditional modeling skills to fix. Like the backside of a hand that is turned towards the side of a leg.

What is the most challenging scan you have ever done?

We scanned a few Dutch artists for an award show. One of them included a whole band (Rondé) and one a horse… (Typhoon). The horse was a modeled 3D asset we bought and posed. A good example of how to combine best of both worlds! The same goes for the instruments of the band. They did bring theirs; however, geometrically shaped objects lend themselves better for modeling - otherwise you spend more time cleaning up. We’d pose the band with their instruments and then removed those just before the scan. Afterwards, we bought the 3D models of the instruments. This enabled us to be done in short but intense 2 weeks.

What do you think about the future of 3D scanning in relation to VR?

Scanning or modeling are both great ways for creating 3D assets for sure. It is absolutely certain that the future needs more3D assets due to VR and AR uses. We plan to add rigged and skinned models once humanalloy.com is a bit further - then the step to VR is relatively easy. We did some test projects in Unreal Engine; AR is also interesting, however, it possibly even less mainstream at the moment. For now we have our own AR business cards… just because... well, we have a 3D scanner, so why not?

What do you think about the future of VR?

I wonder where we are in the hype cycle. Are we almost at the ‘Peak of Inflated Expectations’? Passed it? However I do think it will take a while until we find the right and productive use for the technology. Personally, after trying the HTC Vive I am a believer again and cannot wait to use it for gaming!

What would you like to scan the most?

For now, I want to keep improving the poses and realism of the 3D models we sell on CGTrader We are getting great feedback on what is useful and what not, so there is lots of room for improvement there. I love to scan stuff that actually makes other people's work shine. For now I am hugely enjoying scanning 3D people. And we would love to expand so we could deliver much more models in a shorter amount of time than we do now, adding more formats and rigged 3D people as well. Your stats are very helpful in determining where to go next!