The Potential of CGI in E-Commerce

Computer-generated product images: The demand for CGI is growing

Instead of classically produced photos, many e-commerce companies, including Amazon, Ikea, Obi, Otto, and Wayfair, are already extensively using digitally generated images and visualizations. Restrictions due to the coronavirus crisis recently hampered typical product photography, which is why computer-assisted image generation gained relevance once again. About a year ago, we discussed the potential of computer-generated images for online shops with Tobias Nientiedt, Head of Otto Content Lab, and David Wischniewski, co-founder and CEO of RenderThat, and have now followed up on how the technology has developed recently.

Computer Generated Imagery – From the film industry to product images

The technology behind a virtual image is called CGI. The abbreviation stands for “Computer Generated Imagery” and, roughly speaking, means that the images or visualizations are created on a computer. Originally, the technical term comes from film production and refers to computer animation in filmmaking: “The technology has its origins in the seventies. In the first Jurassic Park film and the first Star Wars movies, CGI was already used to depict what otherwise could not have been shown,” explains Tobias Nientiedt, CGI expert at Otto.

The automotive industry was the first to discover that CGI could also be used for product images. The basis for image production via CGI is the design and construction data of the vehicles, so-called CAD data, which the car manufacturers already had. Another factor was decisive, according to Nientiedt: “Especially in the automotive industry, a lot of money was spent on creating images – for example, to close off airspaces especially for photo shoots so that no one could see models that had not yet been presented. Creating product images via CGI is simpler, faster, and more cost-efficient.”

The technology has evolved to the point where content for online shops can now be produced. Otto, for example, made its first attempts with CGI in 2016.

After the first test images were produced, it quickly became clear: The company was not yet fully exploiting the added value of this technology by simply producing these images. For the past two and a half years, Tobias Nientiedt has therefore been at Otto, driving the topic forward as an expert. By the end of the 2021/22 financial year, Otto aims to depict at least 70 percent of its entire furniture assortment and 35 percent of all articles in the home textiles sector using CGI.

To this end, the company invested 2.6 million euros in an internal CGI database last year and has since further expanded it. “We are in the process of building and expanding the database – and since it will ideally always be growing, it will probably never be completely finished,” says the Otto CGI professional. The Hamburg-based online retailer has created the foundations for scalable CGI production and now aims to gradually increase it. “Our focus is still on the Home & Living assortment, but there are also initial discussions about depicting other assortments.”

“In reality, it is not possible to photograph all variants.”

What makes CGI so interesting for e-commerce? “With CGI, any product can be digitized and placed in any desired environment. The digital twin of a bed, for example, can easily be placed in a youth or parents' room and adapted according to the needs of the target group. Precisely this flexibility leads to many companies already preferring CGI to traditional photography,” says David Wischniewski, CEO of RenderThat, an IT service provider specializing in CGI.

Especially when many different material or shape variants of a piece of furniture are to be displayed, classical photography becomes complicated. Nientiedt gives an example: “Suppose we offer 100 kitchens. Each of these kitchens comes with eight worktops, eight carcass colors, eight fronts, and eight handles. To depict all the resulting variants and present them to the customer as product photos in the shop, more than 600,000 photos would have to be created. In reality, that's not possible – with CGI, it is.”

Otto’s online shop offers more than 100,000 pieces of furniture and home accessories. The group’s declared goal has long been to evolve from a retailer to a platform focusing on Home & Living. In this context, Otto is gaining more and more partners who offer their goods on otto.de and recently also introduced an automated connection for retailers. With more partners, the number of products and product variants in the online shop is rapidly increasing – and accordingly, so is the need to visualize them.

Digital artists instead of classical photographers – how an image is created with CGI

Digital product images are created by CGI artists. These specialists construct furniture like a classical furniture maker – but with the help of very complex 3D programs and render engines. At the same time, they must also bring expertise in photography and image editing. The creation of a CGI furniture piece usually begins with a cube in a 3D construction program. The CGI artist then deforms it in several steps – until it ultimately becomes a hundred percent digital likeness of the real furniture piece, the so-called digital twin.

In a next step, materials are applied to the surfaces of the digitally constructed 3D furniture piece. “The materials should behave in the virtual space exactly as they do in reality,” says Nientiedt. For this purpose, the materials are not captured via a flatbed scan, but also scanned in depth. This allows the three-dimensional surface structure to be digitized. As a result, materials such as upholstery, wood, or metal ultimately look very realistic.

Subsequently, the environment in which the respective piece of furniture is staged is created. “You can imagine it like a typical photo shoot, just virtually: There is a room where the furniture is placed. In it, there are, for example, light sources and shadow reflections. You can also adjust aperture or ISO settings like on a camera.” Finally, this image is calculated with render engines, meaning a graphic is generated from all the construction data. The finishing touches are then added with image editing programs, e.g., Photoshop.

For more complex cabinets or kitchen fronts, visualizations of each individual part are created, which can then even be automatically assembled as needed. “This saves an enormous amount of time, and colors or materials that are newly added to the assortment or are no longer available can be easily added or deleted,” explains RenderThat CEO Wischniewski about the process.

CGI – the basis for Augmented Reality

For customers, the difference between a real photo and a product image created using CGI is ultimately imperceptible or only barely visible. “Our goal is to create images digitally with the same image quality. A high-quality image created on the computer is indistinguishable from a photo – only the production path is different,” says Nientiedt.

Wischniewski confirms this: “With CGI, the image quality can be precisely tailored to the needs,” but CGI can do even more, as he explains: “Further use of digital twins in Augmented Reality or Virtual Reality apps is easily possible. With Augmented Reality (AR), it is possible to place virtual elements in the real world.” This was also a significant reason for OTTO to tackle the topic comprehensively in-house: “The technology allows the customer to place various pieces of furniture in their own home before purchasing,” says Nientiedt.

Trying out furniture at home before buying – AR makes it possible

The idea that customers can imaginarily place furniture in their own home with technical aids is not new. Ikea, for example, already had a function in its “Catalog App” in 2013, which allowed selected products to be virtually placed in a previously selected environment: Customers had to place the Ikea catalog in the room where the furniture was theoretically supposed to stand. Using the smartphone camera, the area was then scanned, and the desired piece of furniture appeared on the smartphone display in the captured environment.

In autumn 2017, Ikea then launched a successor, the Ikea Place App. It is based on AR technology and already depicts several thousand products. “Ikea wants to reach and interact with even more people – no matter where they are. This is made possible, among other things, by Augmented Reality,” an Ikea spokeswoman explained to OnlinehändlerNews in mid-2019 regarding why the furniture store chose an AR app. “Many people are unsure about furnishing or redecorating their home. Especially when it comes to how a new piece of furniture fits with what they already have at home. We show inspiring ideas and solutions online and in our furniture stores. Nevertheless, it is difficult to imagine these in your own home. For example, buying a sofa is a big investment. Therefore, you want to be sure of your choice. With the help of the app, we want to make it easier for many people to visualize furnishing products at home.” By autumn 2019, 3,000 products could already be visualized in one’s own home with the app.

Otto uses a similar concept in its YourHome app. An AR function is also available there for several thousand products. Otto customers already appreciate this service. The probability of purchase is significantly higher if customers have previously placed the desired piece of furniture in their own living room using the AR application in the app. “The feedback is still pleasingly positive, we are satisfied. This is also why the number of 3D models we produce continues to grow monthly,” says Tobias Nientiedt about the app's current development. Amazon also enables AR viewing of products via its app – for example, a TV on the wall at home – and Google presents AR models in search, adds RenderThat CEO Wischniewski.

The bottom line is: comprehensive, authentic visual content offers great added value for customers and significantly and positively influences their purchasing decisions. Otto wants to achieve precisely this added value with CGI and other technologies in online furniture retail and considers this in the cost-benefit analysis. “As more and more content is needed, production costs increase. With CGI, this mass of content can be created efficiently,” says Nientiedt and adds: “If you break it down to a single product, the CGI representation is probably more expensive, as the basic model first has to be built in 3D. But given the mass of product photos that OTTO needs, CGI is significantly cheaper.”

Will online retail soon be impossible without CGI and Augmented Reality?

When asked whether online retailers need CGI to remain competitive, David Wischniewski already had a clear answer about a year ago: “Absolutely. Many companies contact us because traditional methods like photo shoots are too time-consuming, expensive, and inflexible for them. Furthermore, photographing is not practical, especially for larger quantities of images.” According to his current assessment, CGI is also becoming increasingly important. “In a dynamic market, both online and offline retailers are experiencing increasing cost and time pressure. At the same time, customers today want a more immersive product experience than ever before. They expect comprehensive presentations in a realistic environment. 3D visualizations using CGI perfectly serve these increased demands,” explains the RenderThat CEO.

And Nientiedt's assessment was also clear at the time: “We are firmly convinced that in a few years, only retailers who offer Augmented Reality will sell furniture online, because furniture will no longer be bought any other way. CGI is the basis for AR, so there’s no getting around it.” He also observes increased interest in the technology. “The demand has grown even larger and will probably continue to rise; the topic and the technology are becoming increasingly relevant. There are more and more new services based on this technology.”

Effort and costs for CGI content

Producing CGI content yourself is not so easy for smaller retailers. Setting up a corresponding production landscape involves effort and significant investments. This is precisely the basis of RenderThat's business model. “For the mattress manufacturer Recticel Schlafkomfort (Schlaraffia), for example, we designed a process that allows us to create digital twins of mattresses already during product development. This way, image material that can be used for trade fairs or marketing measures is created even before their completion,” explains Wischniewski. “The costs depend heavily on the scope of the visualization. Simple product visualizations are relatively inexpensive, costing a few hundred euros per image.”

For larger companies, RenderThat goes a step further and establishes automated CGI solutions. “Once the pipeline is integrated, it reduces the costs for product visualization to the cent range.” Otto allows other retailers and providers in the furniture industry to benefit from its experience. The Hamburg-based company recently re-invested in its CGI production service to establish it for platform partners as well. They also offer the service to other companies.

Is classical product photography dying out?

However, by no means all retailers are even involved with this topic or actually want to switch. “Often, workflows with photographers and other partners have been established for years, while switching to CGI would, of course, initially involve additional effort,” Wischniewski knows from experience.

CGI is also not suitable for all industries; there are hurdles, for example, in the fashion sector. While virtual product images can be produced based on digital patterns that some designers already use, “fashion thrives on products being seen on a person,” says Nientiedt. It is possible to create virtual images of people, but “these images often lack vivacity.” According to the RenderThat CEO, this is a matter of time, because currently: “With the help of automated rendering processes and artificial intelligence (AI), ‘CGI outfits’ can be displayed on people – simply and photorealistically. A person is first photographed in neutral underwear. Then it's the AI's turn: it automatically projects a variety of clothing items onto the model and achieves astonishing results. In the medium term, even the issue of photographing models will disappear, because these can also be created automatically and photorealistically.”

For electrical products like cameras or mobile phones, CGI imaging has been less relevant so far, as technical data is usually more important than their presentation in different spatial contexts. If the effort for image generation with CGI is still particularly high or significantly higher than the actual benefit, classical photography could remain a suitable means of product visualization.

However, a detailed comparison of the two disciplines, 3D visualization and product photography, reveals many advantages of the new 3D-generated images.

Demand for CGI production is growing

However, almost any product is fundamentally suitable for CGI production, says Wischniewski: “So far, we have not had to cancel a single project due to excessive complexity.” And more and more retailers are now relying on CGI content. “The first quarter of 2020 was already the best quarter in our company's history at the beginning of February. The second quarter will also be similarly successful – despite Corona,” he replies to the question of how the demand for these technologies observed by them on the market has recently changed. Overall, the trend is clearly upward: “We are currently establishing our fourth and soon also our fifth location.”

Similar observations are made by Otto's CGI expert: “We are very satisfied with the interest and demand; the feedback is positive. In times of Corona, fully digital image production through CGI is becoming increasingly relevant. This is also noticeable in the CGI market.”

david wischniewski