Many 3D print designers love building from scratch. However, when a project comes across your desk that requires fit or creation with something that already exists, working your way back can offer some pretty great insights into the best 3D manufacturing process. In this episode, Tom Hazzard and Tracy Hazzard discuss the role of 3D reverse engineering in the digital manufacturing workflow with Dillon Patel, a mechanical R&D engineer working for HP in the 3D scanning and 3D printing fields. Dillon shares the importance of using reverse engineering and how they reverse engineer over at HP in the additive manufacturing process, touching on how they use 3D scanners and the tools they have that help with the accuracy of their digital manufacturing workflow. Commenting on intellectual property issues and offering advice on how to approach a product that needs to be perfectly matched to another part, Dillon provides insights into advances and helpful tools to utilize 3d reverse engineering in your design or digital manufacturing process.
Watch the episode here:
Listen to the podcast here:
The Role of 3D Reverse Engineering in the Digital Manufacturing Workflow with Dillon Patel
We’re going to talk about reverse-engineering, which I have to say is something I rather enjoy.
I’m having fun with this collaboration with HP because I am learning some new things about different parts of businesses, workflow processes, and stuff that we don’t always touch unless the necessity comes up on a specific project. It’s usually not something where we do in our daily jobs. Reverse-engineering is the topic. That can be a dirty word. For some people that can be a dirty phrase. I said this before. I’ll say this in the interview, you will hear that. We’re creative, so why would we want to reverse-engineer? The opportunity comes up more often than you would think.
It can be a necessity sometimes even as a part of creating something new. You’ll hear in this interview and we’ll talk a little bit more after the interview about the opportunities and why reverse-engineering is here to stay.
Not just that but this is the thing we all want to pay attention to as we head into the interview. This is a skillset that you may want to understand and master because should a project come across your desk as a design firm or agency or anything like that, you may be surprised by how hard it is. We’re going to talk with Dillon Patel. He’s a Mechanical R&D Engineer working at HP. He works in the software organization part of the 3D print group in the 3D scanning and 3D printing fields. His background is focused on new product development and rapid prototyping for consumer and commercial products. He’s a grad of the University of Michigan. He came as an intern to HP and then went on to work for them. He worked on the Sprout, 3D cameras, and the Z. We’re excited to have Dillon here.
Let’s go to the interview and we’ll be back to talk about it a little more after that.
Dillon, thanks for joining us.
Thanks for having me. I’m excited to be here.
We don’t often talk reverse engineering because we’re designers. We start from scratch and all of that but we know a lot about it.
I’ve had some experience reverse-engineering some things and scanning them into 3D models. It was a very difficult process. I’m excited to talk with you about how it’s being done at HP and how you are working on that as a part of the workflow.
We’ve done some reverse-engineering workflows here whether it’s for acquisition working with some clients to figure out how to archive some of their parts or going from an end-to-end where we take apart and reverse engineer it, print it and iterate through that.
That’s what most people don’t understand here is that sometimes we have manufactured products, especially the ones that we’ve been making for a very long time that don’t have real drawings to them anymore. There’s a tool somewhere but there’s not real CAD drawing or things that we could turn into a 3D print if we needed to. They don’t exist and sometimes those drawings don’t print well. We have more work to do there. Is that the kind of thing that you’re working on?
It’s an iterative workflow like you mentioned. We can go from that model and scan it with whatever acquisition method and then try to reverse engineer it. Some of the software aren’t typically suited for certain things so you have to iterate through and figure out how to capture that geometry properly. Make sure it’s useful to the end-customer or to yourself and that part is accomplishing what it needs to.
Are you now able to scan 3D scan parts and then manipulate that geometry more easily than at least what I’ve experienced in the past? I’ve done some 3D scanning and then it produces an STL file which is very hard to work with. There’s very little you can do to manipulate that and confirm its geometry. I’m hoping things have advanced since then. What can you share with us about that?
It’s very difficult to go from a scan directly to either a CAD model or directly print it. There’s always these steps that you have to go through in between to make sure that you captured your geometry properly and manipulate it in certain software whether that’s SOLIDWORKS or Geomagic, which is a scan to CAD or reverse engineering software that we’ve used to capture that geometry properly and make it usable. If your scanner is a low resolution, your print might not be that good, but that could be okay because your 3D printer resolution isn’t up to those accuracies that you need.
That’s always the crux. How good is my scanner? How good is my printer? Is that going to give me the result I want? Is the part going to fit if I’ve got to fit it into something else?
The tool that we most commonly use for reverse-engineering is Geomagic Design X. It is a pricey software. There are things that we’ve developed and tools that we’ve made from open source Python tools to scan something and then make sure that the accuracy is up to what our client specified. We had a client where we scanned a bunch of shoe lasts and we needed to reverse engineer them. As you said, they didn’t have the CAD for them. All these lasts were shaped by hand over the last 100 years. They’re around for a while. They wanted to digitize these lasts and have an archive. If anything happens, they can reproduce it. What we did was we used a scanner that wasn’t necessarily accurate but it got the approximate geometry correct. We then developed software tools to measure that virtually, and then compare those to physical measurements that we took.
You always have to compare. That’s an interesting part because we’ve had difficulties with calibrating parts, getting them right, and getting the measurements. Even though you created in CAD, but it doesn’t print the same. There’s that adjustment time that has to go through that. In your workflow, do you have tools that speed that up, or is it still this back and forth print-test, print-measure.
It’s a combination and there are some tools that can speed that up. With something like Geomagic, you can go in and design the actual CAD geometry around this part, extract curves, and then you can compare so you can see that heat up. That will help reduce the time from where you create the part. You print it and then you compare it whether it’s a physical inspection with the CMM or just eyeballing it. The iteration will take less time but it’s making sure that your printer is up to the accuracies that your model is specifying and going back and forth. You always want to test these prints and make sure that you’re getting what you want because it can take some time.
You said CMM. We always like to define acronyms here.
That is Coordinate Measuring Machine. That’s where you’re going and touching the specific points that you were looking for. That’s may be specified in a drawing or a critical dimension that your client needs. Making sure that that print is accurate. If it’s a go/no-go gauge, I’m going to go and measure in those little areas and then compare it to the model and what it’s going into.CMM or Coordinate Measuring Machine makes sure that the #3Dprint is accurate. @hp @zbyhp Click To Tweet
We talked about scanning as the first step in the process. I imagine there are considerations, key factors to choosing a great scanner for what you need to do. Is it always the same or is it different by what part you are trying to engineer?
It’s different. Choosing a 3D scanner is very difficult and you always have to take costs into consideration. For something like a cosmetic part or where you’re going in and you want to create an animation out of it, you might not need a very accurate model but you want good texture maps. Those photos have to be high in resolution and the color has to be accurate or accurately calibrated.
End-purposes have a different starting source.
I had experienced 3D scanning a ceramic part that was a part of a machine. The outside service of that ceramic part was important for its function. Not its appearance, just the function. The inside or the core of it is what I was changing. I was able to get the scan to be reasonably close enough but then I had to use Boolean operations to subtract from the inside of the ceramic part and make what I wanted to. It was very hard to manipulate that geometry. Is that still the way it is in detail? Can you take a scan model and make it more editable in a more conventional CAD software way as if you were creating new geometry?
You mentioned that you were scanning the outside of a part and you cared about the inside of that. In a perfect world where cost wasn’t a consideration, you could use something like CT scanning and get the geometry on the inside. For that workflow, it’s difficult to go from a rough scanned model directly into CAD. You always have to take into consideration those mesh files. They can be heavy, sometimes you don’t need that resolution, different software will handle them differently. In my experience, it’s very difficult to go directly from a scan to a CAD where everything is perfect. You always are going to have to inspect that model to make sure it’s providing what you want and then test it out.
That was always the difficult part. This particular part wasn’t the case, but sometimes that’s why we start from scratch. There’s almost no point in scanning. We might as well measure, draw, and get it going from there. Reverse engineering might be a dirty phrase for some people. It reeks of copying and doing things. There are a lot of positive uses and positive things about reverse-engineering. How do you address that when people hear, “You do reverse-engineering? What?”
You always have to consider what you’re doing and what the end-use case is. Some of the parts that we’ve worked on have been for fun and not for profit. If you’re taking someone’s IP when you’re reverse engineering and you always want to take that into consideration. Your end-use case is different for different uses. For example, if I need to reverse engineer a component to get the whole pattern off of it, because that whole pattern is hard to access, that’s not taking the IP, it’s adopting it and making it more usable for me.
In that case, the whole pattern wasn’t proprietary. Is it something that was needed for a proper fit?
In this case, we didn’t have the CAD and we needed something to mount to this whole pattern but it was hard to access in a difficult place. That’s where 3D scanning came in. We didn’t fully reverse engineer the whole part, just the specific locations.
I could see that being very useful with modifying or enhancing products especially the more outer shell of a product or appearance, but there’s a certain structure underneath with attachment points and things that you have to match up with.
It’s not just that, but a lot of people like to do this when we’re creating creative accessories, things that we want to add on decorative knobs, or other things like that. We have things we have to match up to. Sometimes it’s much easier to develop the geometry if we have the original part, to begin with or if we have something to start from. That makes it easier. I can see that as being critically important. The workflow for reverse engineering for how you’re going to work through the process is fragmented. The way that I would describe it is it’s broken up into pieces and it’s hacked together with which tools you used. I’d love to know how you sit back, analyze, and say, “I’ve got this part I want to create. What am I going to put together?” Do you think about that or do you always go to the same tools?
I always try to approach a part differently. In different cases, I’ll have different tools that are available to me and it’s a different end-use case. If it’s a more cosmetic part, I might not gravitate to a high accuracy scanned part. It’s the texture map that I’m looking for. For example, one of the parts that I worked on was this Porsche hub cap. It came from an employee who had three of them from his Porsche that was very rare a car. He wanted to scan this part, print it, and recreate it. It was this electroplated part that popped up on my desk. When I looked at it, I was like, “This is going to be very hard to scan because it’s shiny.”
Material and surface quality have a lot to do with that.
Optical scanners, especially the structured light scanner that I was using, doesn’t do well with shiny objects because the reflections cause a lot of issues. In this case, I took the geometry so the logo was important for this cosmetic part. Everything else was a simple bracket where it popped into that wheel center and we could take caliper measurements and recreate it based on that. I blended the tools where it’s either using calipers or measuring it virtually with common SPECT, taking the actual logo, placing that onto the CAD part, and blending both of those tools. In this case, we used the scanning software Magic to go into the logo. We touch it up, print it a couple of times, and iterate through that to get to that final part.
The logo itself was the aesthetic part that you needed to scan. It must have been a relief. It was a three-dimensional representation of a Porsche logo.
It was 3D with very fine features. We tried a couple of scanners. Some of them didn’t work at all. We couldn’t resolve that fine texts but we were able to get a couple of good scans of that. One of the ways we get around shiny objects is by spraying it so it diffuses the light. They sell 3D scan spray but we use Tinactin.
We’ve heard that once before with a very early desktop 3D scanner that they were using a foot powder to do that. I was wondering, as you were talking about it if you might have painted it with a gray primer or something. If it’s a precious Porsche part, how would you clean that up?
Let’s not do that. That’s an issue because Tom has a taste in antique cars as well. He has a Volkswagen Karmann Ghia and he’s had to do similar things where you’re trying to find parts and you can’t. There’s no way to do it. The only thing to do is make sure that you have something to start from to figure out what went in the spot and go from there. Sometimes you have to create from scratch but scan something that it’s going to go into. You’re pointing out there in this whole process that there isn’t something that is an end-to-end use product. Someone out there can buy a scanner, use the software, and they’ve got everything because it’s different based on what you’re doing each time. That makes it so much harder.Choosing a #3Dscanner is very difficult, and you always have to take costs into consideration. @hp @zbyhp Click To Tweet
It comes with practice. You start to get a lay of the land with the tools but it can be very intimidating for someone who wants to reverse engineer a small car part for something that they can’t find. There are some good resources online to explore and figure out a cheap way to scan it or a service bureau who can help you through that process but it gets expensive.
That’s what I was going to say. This is where service bureaus can benefit from having multiple scanner options in their bureau. It gives you a chance to go. You may have the standard one that you do for 90% of the projects that come across your desk. For that 10%, you will have a bureau you can count on, you can go into and you can try their different types. That’s where they can play a big role for you, for a lot of engineering firms, and other companies that do these types of things.
A lot of these scanners are very expensive. The software that comes with them isn’t necessarily sold by that same company but is as expensive if not more. You always have to understand the trade-off between the scanner and the software. When I approach a part, I always try to frame it in the sense that what am I trying to get out of it? What are the accuracies I need? What are the tolerances? What’s that end-goal? Try to use the tools available to me or the client and figure out how to do this in the most cost-effective way.
It brings to mind, Tom, that project that Grace Sims did with the doorbell. She had a doorbell and she couldn’t get a replacement for her doorbell on her house.
It was a plastic portion of the doorbell. There was some metal portion and then a plastic portion. All the houses in the neighborhood used the same kind of doorbell.
There was no replacement part available.
Getting a whole new doorbell at your common Lowe’s or Home Depot didn’t fit the holes and the space that was cut out of the molding in the door. She went through the process of finding someone locally to help her reverse engineer the plastic portion of the doorbell and print it so that it would be restored. It was a lot harder than she thought it was going to be.
That was the big a-ha. Many people think, “This is easy. I’ll scan it and I can print a new one.” That’s not the easy part. Those of us who’ve done it or have that engineering background of having gone through this process knows how hard reverse-engineering can be. Do you have any advice for anyone who might be going out there as to some approaches they take on how they analyze a product that comes across? What should they look for and then how should they match that up with something else?
I would try to take advantage of the free tools that you have available to you. If you have to scan, Goldman SPECT is great because it’s a free software where you can go in and make sure that your point cloud is accurate to that model. Don’t be afraid to use calipers. I know they’re not an accurate tool but they’re great for a rough measurement. If your point cloud is distorted, you can go in and say, “That one didn’t work well. Maybe I should try a different scanner or take a stab at recreating the geometry yourself.”
I like that approach because so often, we clean everything up and then we measure instead of measuring right at that point of like it’s not accurate from the beginning, “Maybe I should try that again.”
It’s nice to go scan a part and inspect it roughly to make sure that you’re getting what you need in those critical areas. You can go into different software and start manipulating it, spending all that time cleaning it, eventually bringing it into CAD software and manipulating it there. You want to make sure you’re getting those right dimensions off the bat or be aware of that in the long run so you’re not making an incorrect model.
Get it right from the beginning and less work later. You were talking about the shine and texture of the material, but the color does matter too. Why is that?
With optical scanning, it’s hard to scan darker objects because for structured light, you’re shining a black and white striped pattern onto this object. You’re not going to get a lot of transmissions back. The cameras aren’t going to get to see those dark objects because it’s not going to come up in those pixels. There are ways to get around that. You can scan objects at different exposures which is HDR where you scan it for the brighter areas for the white part. For the dark part, you turn your camera’s exposure a little bit higher and you can resolve those darker textures. For something like a translucent object, that’s where you have to go in and spray it, the same thing with a shiny object. Those are the three features that are tough to scan.
I want to talk a little bit because we’ve been exploring that Z by the HP model of things. That in and of itself being able to resolve the textures and other things afterward in a different way so you get your geometry right and then you get your texture right. Is that a workflow that you follow sometimes?
I haven’t personally done that. You have to go in and define that UV mapping. It’s taking the image and then mapping it to the 3D texture. I’ve always used the processes within the scanner and the skin software but that is something you can explore if that’s what you’re looking for.
You’re going to have more texture options in the future, which I love the idea of that. Sometimes we have parts that don’t have that material quality that is so hard to scan. I remember when we scanned my hair. We did a scan of ourselves and my hair was a disaster because it’s dark and textured.
The resulting scan, you end up looking almost like a solid plastic molded version of yourself rather than the fine qualities of hair. There are limitations in trying to do this but or a lot of hard products, your Porsche hub as an example, there are lots of things that can be done. The good news is it can be done. It certainly can if you approach it right and you’re careful about how you do it. The other important thing to note is this isn’t easy and it takes a combination of skills and tools to get reasonably close. Would you say that’s a fair description?
I would agree. The biggest industry for reverse engineering and we’ve talked about this a little bit is auto market parts, where the secondary market doesn’t have access to the CAD for a car but they create these mounts of the cars. If it’s a roof rack, they don’t have that geometry but they’re recreating it using 3D scanning. Since our industry is big, they have access to these tools and a lot of them are very expensive. They have FaroArm, different laser scanners, structured light to get in four different areas, maybe the paint is shiny. They have ways of getting around it but that market is big so they have access to very expensive tools. For someone who’s going in reverse engineering a doorbell, you don’t have that same accuracy. Maybe you have an iPad, you want to take a couple of viewpoints, use Photogrammetry and stitch them together. That’s when you have to go in and say, “Maybe my model is not accurate.” I can go in, take some calipers, measure it and get the right dimensions off of it, and then use the scan for that cosmetic part.#3Dscan a part and inspect it thoroughly to make sure that you're getting what you need in those critical #3Dprint areas. @hp @zbyhp Click To Tweet
That’s what I personally would do. Not being part of a huge corporation that has a lot of expensive tools, I have my digital calipers. I would measure things out and I have a desktop scanner if I need to scan something but it’s tricky. I imagine when you did the Porsche hub, you created the geometry that was behind the decorative part, creating it conventionally, but then when you had to join it with the scan, you had to do that in a mesh sense, a Boolean operation to join them, is that right?
That’s exactly how we did that.
You have large part issues and small parts. It’s a lot harder to do a large part like a car or the roof of a car. You have to break it all up in pieces and piece it back together to get with what you’re looking for. We have the same problem with furniture.
You can buy a very expensive scanner with this huge scan volume or go in, painstakingly move the scanner and try to stitch things together. That’s when you have to make that trade-off and say, “I’ve got to get some help with this.” Have a service bureau who has access to this tool to help you out and get the geometry right.
I don’t think that after a conversation with you, that reverse-engineering is for me. I live in the creative side of things. I still get how often I want something that I’m creating that attaches onto something or does something, then I need someone who can do that for me. I can see that I would not have the patience for that, personally.
It depends on what you’re trying to do.
I would make Tom do it.
I like creating new things too but at the same time, I like getting more life out of the existing things. When you have an antique perhaps or a rare car or it’s not that old, but it’s no longer manufactured but you want to get more life and use out of it, then reverse-engineering and making your own parts is a necessity. I’m in favor of it.
Dillon, thank you so much for coming on the show. We appreciate your time and insights into the workflows and the other tools of reverse-engineering.
Thanks for having me.
The Role of 3D Reverse Engineering in the Digital Manufacturing Workflow — Final Thoughts
I thoroughly enjoyed that because I felt like I personally had an experience that I could relate to Dillon and what he’s been working on. I was somewhat surprised that the process for reverse-engineering something where you’re going to scan something to get the critical part that you could not create from scratch, certainly not easily. In any timeframe, you could justify or expense you could justify. With conventional CAD work and that combining them, it hasn’t changed that much in terms of the workflow and the process while the tools, the software, the scanner, or some of the things you would use certainly have improved. They always improve year-over-year but the process is very much the same.
It’s still pieced together, relevant and needs to be specific to what you’re creating or what you’re reverse engineering in this particular case. HP has some resources on their website. Dillon mentioned an aside and I’ll give you a little bit of a scoop on that because it is public information. They’re working on a fit station, a laser scanning process that is going to be used to take measurements of your feet, take scans of your feet, and other things. They’re going to be doing some in-store scanning through some of the processes of what they’re working on. That’s future coming out too.
I didn’t see that case study or any information on the fit station in the resource center, but I guarantee you, it will be coming up soon. I keep learning new things each time. I’m excited for you to keep going through this. We’re about halfway through our series here and there are many great episodes to come and so much great information about different types of processes, tools, projects that are going on applications and real case studies in the workplace.
Hopefully, you subscribed and stay tuned for the next episode of this special series. We will be back with another great episode.
Get Even More!
- HP Structured Light Scanner
- HP Resource Center
- 3D Scanning White Paper
- Dillon Patel – LinkedIn
- Z by HP
More About HP:
Capture and Create with Z by HP, Inspiring you for your next creative breakthrough with the Z portfolio designed and built to improve the way you create. Discover the latest Z Book to help you with your latest creative project.
Experience your design with HP Multi Jet Fusion technology and solutions reinvent design and manufacturing, unlocking the full potential of 3D printing and bringing down the barriers of 3D printing adoption across industries through materials innovation. For more details about Multi Jet Fusion technology click here.