From CAD to Commerce: How Better Product Visuals Reduce Returns and Strengthen Trading Partner Relationships

[0:00] Johannes Crepon: My name is Johannes Crepon, one of the founders of PDM Automotive. We provide a data platform for the automotive aftermarket, always focused on helping suppliers become more efficient with their trading partners. If you’re a parts manufacturer or supplier in the automotive space, you’re selling to retailers, distributors and often individual shops and jobbers, and PDM’s vision is to make that connection and relationship as efficient as possible. That starts with one of our core value propositions, catalog and content management, and a big part of content management is digital assets. So I’m really excited to have Will on this webinar today. Will, do you want to give a quick background on yourself and on Glossi before we get into the slides?

[0:56] Will: Of course. Hi, everybody, I’m Will, the CEO and founder of Glossi. Our mission is helping brands and manufacturers go from CAD to all the downstream assets they need to sell. We combine the best of real-time 3D and AI in the browser, so brands can take their CAD models directly from SolidWorks, or wherever they design their 3D assets, into a browser-based platform and create photos and videos to very specific brand standards, at scale. Johannes Crepon: Awesome. Let’s go through a couple of slides, and I’m probably going to cover topics that are no news to anyone on the call. As we were preparing with the Glossi team, we asked: what’s one of the core reasons customers adopt the platform? There are many reasons, like great images, but a big one is avoiding returns. I assume everyone on the call knows how big a challenge returns are in e-commerce in general and in automotive. In automotive we talk a lot about fitment data and how important it is for reducing returns, but it’s just as important to have rich content describing the product, and digital assets are a big part of that.

[2:44] Some stats out there say we’re averaging around a 20% return rate. That’s really high. If you send out 100 parts, you should expect 20 to come back. And according to some of the sources cited here, 22% of those returns happen because the photo was inaccurate. These are high-level numbers, and everyone probably agrees 20% isn’t great. If you break it down, it’s not just that a sale gets returned and you refund the money; there’s a cost associated with it. These are very conservative numbers in my opinion, but we did an example calculation.

[3:41] If you ship 100 parts a day, that adds up to 36,000 parts a year. At a 19.4% return rate, you’re looking at about 7,000 parts returned, and that all has to be absorbed. The cost of a return isn’t just the refund. There’s processing and labor, which I’ve put at a very conservative $13, and shipping at $10, which as we all know doesn’t go very far; ship something larger across the country and it’s much more. And a lot of what comes back isn’t as perfect as when it left the warehouse, so you may have to rebox it or write it down because someone installed it.

[4:38] So $30 per return is a very low number, but even in this example it adds up to over $200,000 in cost. That really drains your margin. So how do images help reduce returns? We’re excited to share examples from a case study with a shared customer, MagnaFlow. We’ll talk about that after the demo, but some of these images are of their products. A big part of what makes the right product more likely to be purchased, and less likely to be returned, is the level of detail.

[5:37] Look at this product: it’s not just the main exhaust system but all the details. Every bracket, every nut and bolt, every reducer is captured in the image. If you’ve ever photographed a large exhaust system manually, you know how much work it is. It’s a bulky product, and a high-quality photo takes a lot of time. You also risk missing pieces, maybe you didn’t take every bracket and clamp out of the bag they’re packed in. If you can take your CAD drawing and turn it into an actual image, it’s much easier to keep every component represented and to keep the image updated. If the product evolves in three or six months, a bracket changes, or you add or remove pieces because they’re no longer relevant.

[6:51] Keeping up with all of that is a lot of work. And when we talk about efficiency in creating images, it’s not just one great image. How about six great images? That helps the customer really understand what’s in the box and all the individual components, which reduces returns. It’s not only about reducing returns but about selling more. If you sell X today and don’t want 20% back, how about selling more than X and getting even less back? That’s a big win: fewer returns and more sales in the first place. I don’t think anyone would argue that better images reduce returns and increase sales.

[7:47] So we all agree that’s obvious. The question is how you actually manage it. What’s unique about the automotive industry is that we’re not talking about five gadgets; we’re talking about thousands of SKUs. On the low end, 1,000 products is a small catalog in automotive. The average on the supplier side of our customer base is more like 3,500, and on the retailer and distributor side you’re talking about millions of SKUs. How do you keep six or ten images per SKU up to date and maintain them? This is where PDM supports you: one platform that makes it easy to manage all those assets, keep them current and keep them available to your customers.

[8:47] I’m speaking from the supplier lens. If you have 1,000, 2,000 or 4,000 SKUs, you want to keep that data up to date continuously, because every day someone hits your catalog via API, file download or manual lookup. They take that information to represent your product on their own website or in their own catalog, whether in an internal system or online. Keeping it current is super important, and that’s where technology comes in. The box on the right mentions the PIES® standard, the industry standard that supports communicating digital assets. Within PIES® there are plenty of options to share digital assets efficiently, whether images or installation instructions.

[9:43] We’ll also look at 360° spins, and Will, without putting you on the spot, you can also show a 3D rendering you can rotate around. That’s always great to see. Having all of these assets in one place and keeping them updated there is crucial when you share that information downstream across your customer base and channels. That’s where PDM supports you. But PDM houses and distributes the information; the assets have to be created first. That’s why I’m excited to have Will here. Handing over to you, Will: this is your slide before the demo. I’m really excited about what you described, going from a CAD drawing, which nearly always exists because you have to build the product in the first place.

[10:46] And turning that drawing into a commerce-ready image or set of images. Over to you for the demo. Will: Really appreciate it. Big picture, this is a process everyone deals with, and our mission is eliminating the gap between the original CAD or manufacturing file and all the imagery and video you need to communicate that product downstream. That can be an exact package for the PIES® standard, or adapted to your own brand and lighting for your e-commerce experience. In this demo we’ll go through a standard flow: dropping an asset in, applying materials you can track in the platform, and scaling that recipe across every product you bring in.

[11:53] You can adapt the lighting, detail and materials, as you’ll see, and every shoot you create in Glossi can be turned into an automation. Any asset you drop into a folder, Glossi will automatically produce the visuals and send them to a data management system like a PIM, or connect directly to your website. Let me jump into the demo. Here’s the Glossi platform. If you’re familiar with 3D rendering tools, in Autodesk or tools like KeyShot, they’re pretty heavy on desktop machines. The minute you start rendering, your fan goes off, and it requires a lot of technical expertise even for a basic image.

[12:54] A lot of what we’ve done is bring all of that into the browser, so it’s accessible to everybody. It acts as a 3D DAM as much as a production system, and it makes producing assets really easy. We use Unreal Engine in the background, which will make more sense when we get into the studio. Quick tour: here’s the library where all your models end up. You can quickly add a model by uploading a 3D file and view it in 3D. This becomes a library of all your 3D assets. You can also set up templates, designed with your team to your brand standards, and manage materials, props and even decals for your product.

[13:58] Everything you need to bring your 3D asset to life. Now the fun part, a project, to show how easy it is to go from a 3D model to a final image pack. To set the scene, especially with the MagnaFlow case study, this process traditionally takes 30 minutes to an hour per product, even after replacing photography with a 3D process. This will take a couple of minutes, with better quality and more control. And to Johannes’s earlier point, this isn’t only about creating a one-off asset. It’s about managing an entire library of products, updating them continually, or even producing product packs through the same system.

[14:57] Here’s our studio. It’s simple to use, with real-time, high-fidelity output of every part of your product. We’re in photo mode, and I can capture any angle and apply specific brand standards like you would in a photo studio: focal length, aperture, camera position, exposure. If I like the shot, I add that angle, then add one from the top, and everything is saved as I shoot and direct. This model came in directly from SolidWorks, so it has some color on it, but the goal is to make it photoreal. I can go into materials and apply a new material set just by clicking a part.

[16:03] Here I have some metals. I click on chrome and immediately update that material set to match a specific metal. Beyond the presets, I have complete control over the exact color. You can save materials to the library and auto-apply them to any asset you bring in. Let’s make this one a brushed metallic, bring the roughness down, change the color a little, and apply the same chrome to this other yellow part. It’s all in real time; what I see is what I get. My angles are still saved. I can adjust colors, use a pure white backdrop, or put it in a scene with shadows, but we’ll keep white since that’s the standard.

[17:13] I can apply different lighting presets, or adjust individual lights. I can build out the scene further: capture this front shot, a top shot and a three-quarter angle. Getting the images is as easy as pressing export. I select my angles, render, and they show up in my library or any tool I’m connected to. Along with photography, video is a bigger and bigger need across every e-commerce experience. Video usually feels expensive and out of reach; our tool makes it as easy as creating photos.

[18:16] I go to video mode and quickly add new clips, review them and adjust them in real time. We’ll build a 360° spin from scratch: pick a still camera move, push the distance out, set a 45° angle, then animate the rotation from 0 to 360. In that time I have a full 360° spin from every angle, and I can duplicate the clip for a closer view of the details. What I see is what I get, always editable in the browser, and easy for anybody.

[19:17] The final output looks studio grade, and exporting is simple. I select that clip and these images and export at 1080p, or 720p, which is standard, or up to 4K if you’re doing a big commercial or content for your website or social channels. I press render, and all my clips show up in my library, exportable to any other platform. That’s the quick demo. I can always go deeper, but hopefully that shows how easy this can be.

[20:21] So much of creative tooling is that kind of control over materials, lighting and everything else, to show your product exactly how you want. But the real value of Glossi is that it’s not only a visual photography platform; it’s an automated production platform. Once your settings are defined by your team, which you can do in-house or we can help set up during onboarding, you can send any asset you drop in through a preset recipe and get 4K images and video back, to your exact standards. Johannes Crepon: That’s what really caught my attention. Companies already have the CAD files from engineering.

[21:23] And as they bring a product to market, once those recipes and automations are set, nobody even has to use the interface. It just flows from CAD to e-commerce, or into their DAM or cataloging system. Will: Exactly. If anyone here is familiar with 3D rendering, doing this the traditional way on a desktop, I’d hit render and wait a couple of hours for great assets. In under two minutes I have all these images ready to review, comment on, edit with AI or send elsewhere, all from a single platform.

[22:25] Johannes Crepon: We loaded a few images, so let me share my screen. For anyone familiar with the PDM platform, in the digital asset section this is how the images would arrive in PDM. These came from a preset recipe; they’re the images we saw in the presentation. You have your high-quality primary image, you can choose which image is the primary photo, and you also have your 360° spin, which you can view within PDM. These are now ready to be distributed to your trading partners, whether that’s a big retailer, a marketplace or anywhere else.

[23:22] That’s what I’m excited about: the automation. There’s always a challenge in automotive with great technology: it’s great to do it once, but how do you do it a thousand or ten thousand times at scale? That’s where this opportunity comes in. Will: I can quickly show how the automations are set up, since that’s in the platform too. This is all live on our site today. Here’s the automation tab, where you can set up a simple workflow triggered by anything.

[24:20] Here’s an untitled automation canvas, with a job tracker showing how many models have gone through and what’s in progress. I can add a trigger: when a new model finishes processing, or a new model lands in Google Drive or OneDrive, kick off a material swap and a render with a selected template, like this soft white studio, create images at a certain resolution to a specific standard, then export to OneDrive or apply an AI edit. Once it’s set, I turn it on, connect my accounts, and it runs in the background, truly automated but always manageable here. Johannes Crepon: That’s huge.

[25:16] Before we go back to the slides, if you have questions, the Q&A button should be visible at the top, so feel free to post them. We’ll share Will’s and the Glossi team’s contact information as well. Back to the presentation: we teased one of the companies you’ve been working with, which we’re both proud to share as a customer, MagnaFlow. Walk us through the use case. What was the before and after for MagnaFlow with Glossi? Will: Before Glossi, every product MagnaFlow created started in CAD.

[26:17] Then it went through a manual process, either photography that took days plus shipping costs, or 3D rendering in SolidWorks. At its best, the fastest workflow took 32 minutes per SKU, per colorway, to render images, and even more for 360° video. With a large catalog of different shapes and sizes, consistency is really difficult. MagnaFlow’s challenge was how to connect with the products they already had and create a consistent, scalable visual production workflow without managing one-offs. They went from a 32-minute minimum to a three-minute maximum per SKU through our pipeline, often zero minutes because it runs in the background as an automation, saving hundreds of hours of manual production, fixes and per-product tweaks that now run on autopilot from one digital twin or CAD file.

[27:42] The other piece is that motion and video content had been priced out entirely. It was traditionally too expensive and required a whole other team of technical people, and some of the more creative work their marketing team wanted, beyond the standard assets, was out of reach. So our tool not only collapsed standard product pack creation but also unlocked new ways to communicate their products across their website and social channels. That’s 10x faster SKU production, hundreds of hours saved, and all set up in just weeks.

[28:39] Johannes Crepon: Looking at those numbers, and that’s just the savings, before expanding beyond what they were doing so far. We also know that with search shifting to chat and AI interpreting product information, rich assets are a real benefit for discoverability. This slide sums up the business impact: it’s not just saving time and reducing returns, but increasing revenue and being future-proof for discoverability. Will: Exactly. And I’m sure every team has talked about how to take advantage of this new world of production and AI. But in practice, the way those tools work today, feeding in a single image, accuracy degrades over time and especially at scale.

[30:00] Images hallucinate, and your parts aren’t part of the training data an image model works from. Your products are very specific; it’s all about accuracy, size and engineering precision. The only way to provide that, and make it future-proof, is a single system that goes from the source of truth, the 3D engineering model, all the way to the final asset. Johannes Crepon: That’s such an important point in the world of AI. AI always works toward an average. So starting with something extremely specific, and nothing is more specific than a CAD drawing, and using that as the foundation every time rather than averaging toward it.

[31:00] I can see how that’s a big benefit over generic AI image generators, where hallucinations go rogue with your products. Will: Another trend we’ve found very interesting, especially for very specific products and returns, is AI content guidelines and the legality of AI content, as returns caused by false representation of products using AI have exploded. That’s another reason this is built to be future-proof and to scale with both engineering changes and content production. Johannes Crepon: Plus, there are already regulations on the way, often starting in California, requiring images to be an exact representation of the product you’re selling.

[31:59] Fast-forward five years, if that applies to every product, there are a lot of images left to be taken in this industry. Will: Yes. Johannes Crepon: We do have a few questions, and I want to make sure we address them. I’m jumping to the last slide so everyone can grab Will’s email address. Glossi and Will are offering two things: a free five-SKU workflow validation, so you can experience it with a handful of SKUs at no cost, and, which surprised me, they’ll waive the onboarding fee within the next 60 days. Really excited for everyone who can take advantage of that.

[32:57] Let’s go to the questions, in the order we received them. Will, can you see them? Will: I’m not sure I can see the Q&A, but I’ll listen. Johannes Crepon: I’ll read them out. The first asks you to clarify when the conversion happens from CAD to the polygonal mesh that’s rendered in Glossi with materials, and asks about NURBS and spline surfaces. Will: We’re built to accept files from any 3D tool, and we have our own pipeline, simplified to make the experience as easy as possible, with UV remapping and all sorts of optimizations of the model that comes in, to make it pixel perfect and easy to apply materials.

[34:00] If you apply materials in another tool and export, those materials are imported and can even be saved to your library in Glossi. Most of the time, though, people aren’t applying materials in CAD, because the tools that support it, like SolidWorks and Autodesk, don’t produce great materials. So you bring the model in, it’s re-UV-mapped, and you place your materials. Part of the automation is applying materials automatically to specifically named parts, which saves a lot of time, whether setting up a single asset or multiple colorways per asset.

[34:58] You can say any part with “steel” or “aluminum” in its name gets the right material automatically, so it’s ready to render the minute you drag it in. Johannes Crepon: There are two more questions on a similar theme, so I’ll read both. One: what formats do you accept? The ones mentioned are IGES and STEP, among others. The second: for the digital twin process, are UVs being created, and are the digital twins authored in USD with variant sets? I thought I was geeky about product data, but this is above my head. Will: Great questions, because that’s really the workflow. First, we accept 40 different file types.

[35:59] From the earliest days of building this product, we knew it’s only helpful if we don’t add another step to anybody’s 3D process. People work in the tool they work in, export, and it shows up exactly as it needs to in Glossi. Part of that is accepting a ton of file types; the other part is that UV remapping happens automatically when you bring in an asset, and if you have special UVs you don’t want remapped, you can turn that off as part of the pipeline and the initial SKU validation we set up. Second, yes: every model you bring in can have variants. Those can be part configurations, where you move parts around to shoot a template with an exploded view, a part hidden, or certain parts of the whole product.

[37:07] All of those variants are saved and exportable as USD or GLB. Johannes Crepon: Awesome. We’re running out of time, but Will’s email address is on the slide, so send follow-up questions there. Sorry for speaking for you, but I’m sure you’ll be happy to get them. Will: I’d love to learn from anyone and answer any other questions, and set up time to talk about your specific workflows and needs. Johannes Crepon: I think we covered the audience questions. One I wanted to understand better: aren’t there a ton of different processes for how people do this today?

[38:12] How do you adapt to all of them, when someone says “We only shoot the primary image,” or “We use CAD for secondary images”? What’s your experience across your customer base? Will: Everyone starts from a different place. Normally there are two starting points. Either we do all photography in a studio: we create products, wait for samples, ship them to the studio and shoot them all. Or we already have a 3D rendering workflow, but it’s run by a couple of often-overwhelmed people, without much automation or control for updating assets once they’re created, or easy standardization across one-off projects.

[39:14] So we focus on adapting to the workflow of any company that manufactures anything, in three ways. First, accepting all sorts of file types and making background optimizations and automations easy for preparing 3D assets to look their best in the studio. Second, what you saw in the studio: precise camera control to specific creative standards, whether a retailer’s or your own brand’s for e-commerce, including lighting, materials, camera angles and motion. Third, the workflow and automation side, so no team has to learn a new tool or run things manually in their process.

[40:16] You can set up a folder where you probably already keep 3D assets, or a platform like PDM, and make the output automatic. We also think of it as a mental shift from creative as a one-off process to a live studio that’s always there. People talk about it less, but as customers adopt the tool, it always comes up: “We have all these shots per product, but I’m running an email campaign or a sales presentation for a specific customer and want to show a specific angle.” That becomes as easy as logging into Glossi, capturing a new shot and exporting it, instead of a handoff to another team or a new request to a photographer, agency or freelancer.

[41:20] Johannes Crepon: I’ve dabbled in CAD tools for personal projects, and there’s a learning curve. I’ve seen your tool three times now, and it already feels very manageable. That’s a big benefit: you don’t have to be an engineer or a photography specialist, so a much broader audience can benefit. We’re really at time now. Will, thanks so much for joining me today. I hope this was valuable for the audience. Glossi is putting out a couple of great offers, so please take advantage of them and connect. Thanks again, Will. Will: Thank you so much for having me.

[42:17] Johannes Crepon: Thanks, everyone. Will: Thanks for the time, everyone.