In this exclusive piece for WhichPLM, Martin Semsch shares the story behind Vizoo GmbH – a 3D material digitization company he helped co-found in 2013.
Digital Materials: the backbone to every good, and realistic, 3D visualization. Being the lead supplier in the fashion industry for material digitization hardware & software, Vizoo offers plenty of tools and algorithms to tackle the task.
In our industry, material creation is a particularly manual and tedious process. Digitizing a material sample is not as easy as simply taking one picture and dropping it into your 3D application. Let me give some more background here.
The human brain is a wonderful thing. By looking at a real sample, you immediately recognize how the material is constructed. In the example above, we assume that the material consists of small hemispheres, just by analyzing how the highlights and shadows behave. Unfortunately 3D applications are not quite as smart as we are – at least not yet. That is why we must provide the visual material properties in a more easily digestible format, in so called “texture maps”. Basically, instead of a single image, your application reads in several images, each describing a different visual property. For instance, there are textures for color, structure, reflectivity, transparency and more. Put together, your computer can recreate the physical accurate behavior of your swatch in virtual space!
From theory to practice, Vizoo’s history began at the close of 2011, in the tiny town of Herzogenaurach, Germany. The soon-to-be founders of Vizoo, Reinhard Meier and Martin Semsch, at that time were both working on the project of creating digital materials from physical swatches for Adidas, a pioneer in the field of 3D footwear and apparel. There was no fast and easy way to obtain the described textures. So instead, the duo took the only option left – the hard one – and started with a flatbed scanner and, sometimes alternatively, a DSLR camera set up to scan fabrics.
The attentive reader will realize that this only creates one image – not several texture maps, which we would need. Hence, what followed is a long, difficult and exhausting procedure through the digital jungle of Adobe Photoshop, Crazy Bump, NVidia Plugins and Custom Scripts to somehow recreate the visual property maps based on that one image. While at the early stages that workflow provides you with a good start, there are some key problems with it:
- Quality: Recreating the texture maps from a single input image will never be physically accurate, so results are prone to suffer.
- Expertise: The process itself and the tools it involves are hard to learn and take months to master (e.g. Photoshop).
- Time: Even for an expert, the digitization of a “difficult” material can take several hours.
- Reproducibility: Due to the huge user input required for the creation and subjectivity of the workflow, there is no consistency whatsoever in the final results. Depending on who digitizes the material, results will always look different.
Recognizing these heavy issues, together with an increased demand of materials, we soon realized that we needed to come up with a better and more automated process. Easier to use, faster, and with less user input. That was when the idea of xTex was born.
With a saw, a soldering iron, a hammer, a handful of screws and some basic programming skills we developed our very first prototype of today’s hardware. By taking more than a single image of the material and analyzing the data, we managed to extract accurate visual information, compared to the flatbed scanner. There was no Software interface built around the prototype yet – most of the post-processing was still done inside Photoshop – but the automatically created textures were already a big advantage. We had managed the neat trick of combining the acquisition of color, transparency, structure and reflectance of a material in a single and compact device.
Fast forward a couple of months, and after several iterations and countless test-routines, the project was a success. The now properly constructed prototype was sent to Asia for fabric scanning, and development of the scanner ended. Obviously, Adidas is a sportswear company, not a Hard & Software supplier, so there was no incentive to research further.
Nevertheless, our fingers were itching to do more; we saw big potential in the technology, and wanted to build software around it to take Photoshop out of the process and increase the hardware capabilities even further. Consequently, we began meeting after work, and often on weekends, to follow our vision. As both of us worked full time unfortunately, it soon became clear that we could not develop a sophisticated product in our free time. Hence, we took the decision to leave our current employer and, with the blessing of our colleagues at Adidas, founded Vizoo GmbH at the end of 2013.
After settling in our new roles as Managing Directors of a two-people business, moving into new offices in a start-up center in Erlangen, and roughly a year of constant work, we managed to hit the market late in 2014 with our now finished product: the xTex Scanner and xTex Software.
Our first client, and up to this day still a good customer, was Adidas. However, apart from a few other big brands, the rest of the fashion world did not have any kind of 3D visualization pipeline yet and were unfortunately uninterested in our product. Therefore, at the beginning we turned our head towards another industry which had been utilizing 3D for decades already: automotive. With the Autodesk VRED event in Berlin, followed by a sales tour to Detroit, and the “Big Three” as well as the Siggraph event in Los Angeles, we took the market by storm. And finally, in 2015 we gave the signal to begin in the Fashion industry, with the Apparel PI in New York, re-designing our software specifically for our new customers. Although 3D was just making traction in the fashion market, the feedback was overwhelming. As if everybody had just been waiting for a signal, there was now an initiative in nearly every company, or at least an interest, to integrate 3D technology in their pipeline, and therefore our technology. Not only is Vizoo already a well-known name in the field of virtual apparel, we even had to adjust our supply chain to meet the increasing demand.
Through countless discussions with, and openness of, our clients the necessity to integrate into the existing system landscapes soon became clear. Today we can proudly state that we are working together with well-established 3D-CAD suppliers like Browzwear, Clo3D and Assyst, and we are still actively extending the list of partners.
The products: xTex Software and Hardware
Where the xTex hardware device provides the detailed material information due to a complex interaction of camera and dozens of lighting sources, the xTex Software gives you all the necessary tools to process the data and create a seamless repeating fabric. We soon understood that Wovens, Knits, Leathers and Synthetics are all too different to be handled with one single solution. Thus, we invented different algorithms to meet the specific needs of Apparel and Footwear fabrics.
Compared to the previous tiling workflow, including Photoshop and other tools, we already managed to cut down the processing time by 80%. The newest addition to our tiling algorithm family even goes a step further – by analyzing the material and trying to identify a pattern, it manages to completely & automatically tile most repeating patterns, without the need to select anything by hand.
- Stitching: Automatically matches the borders of your scans, horizontally and/or vertically (belts, laces, wood)
- Synthesis: Can generate an endless sized tile by randomly repeating a smaller input texture (leather, plastics, organics)
- Pattern Recognition: Automatically finds repeat and can generate an endless sized tile (meshes, wovens, knits)
In most cases, no 3D expertise is needed if using the xTex software. Especially for users unfamiliar with the alternative complex tiling tools like Photoshop, this is a huge benefit.
And what does the future hold for Vizoo? Of course, we are striving for perfection by continuously improving our hardware and software capabilities, becoming faster, scanning bigger surfaces and even generating more accurate output.
After the successful integration with apparel simulation tools, the next logical step for us is achieving the same beneficial setup with the apparel PLM systems.
On the bottom line, we are striving towards becoming the industry standard for digital materials and are pushing to implement this digitization over the full material supply chain, starting with the digital swatch book from the fabric mill.
*Curious to learn more? Feel free to reach out and have a look at our website.