 Rhaid i am ddim yn fawr i gafodd Andrew Godwyn yn ystod o gael Llyda ar y tymiadau a maen nhw'n gweithio ar gael. Rhaid i am ddim yn gweithio ar gyfer. Mae'n Andrew, rydyn ni i'r gwahanol yn gweithio ar gyfer Python a y web, ond mae'r hobeidydd yn gweithio ar gyfer mae'n gweithio ar y cyflogon i gael â'r gweithio ar y cyflogon. Dyma'r gweithio ar ddi. i'r wneud, yn y bwysig, mae'n gwaith eich moddau a'r ffyrdd dwylo. One of my first projects was this. I apologise by the way this is like a weird low res thing with the screen. But imagine it looks much prettier. But these are small metal milled national parks of the US. So I moved to America around four years ago and one of my goals was I want to go and see every one of the national parks. There are 59 of them and I wanted to do a thing Wnaeth cymryddi'n ardal y dyma. I'r rhaglion er fydd dim yn dweud mewn gwahanol, Dyna'n dweud sydd wedi'i meddwl 3D a allwn i'r parwgau'r ddech chi'n brifetig i'r meddwl? Wyddon i gael ein ddweud eich meddwl i'r cyntaf. Mae arifiad pwn i ddechrau'r cymhp achievedig. Ac rydyn ni'n dweud y dyma eich yr ynw Ierysaeth Francisko, ddiogel eu Lundewg, yw eich mwyaf. Ond ydyn ni'n ym 1 m ddiolog a o bobl 30 o 40 cm o'n holl, a rydyn ni'n gweithio i'r tylu bwysig. Rwyf i'r wneud hynny, mae gennym i fyny o'r arferio'r prop sydd yn ymddangos. Gweithio ar y cerddur, sy'n gweithio gweld i'r cyfnod yn cyfnod o'r argyrch yn ei fodw. Gweithio'r argyrch yn ymgyrch. Mae'r rhan o'r tyfnod yn ymgyrch a'r rhan o'r leisarwch ar ei dynig o'r werth. Rwy'n gweithio i'r argyrch oherwydd dwi'n gweithio'r argyrch a'r argyrch. Help mewn cyhoedd, y bydd yng Ngwneud yno i ddweud i ddweud i ddweud i ddweud i ddweudio. Gweddai yma? Gwyddech chi'n ddweud i ddweud, gweithio chi'n ddweud i ddweud i ddweud i ddweud y cael amser? So godd ynogi. Gyda gweithio, mae'n ddweud o'r amser. Fyddech chi'n ddweud i ddweud, mae'n gallu maes ei ddweudio allan. Mae gweithio arall wedi gofio yna yn enkhedlaid. Mae yna gweithio chi'n ddweudio eu ddweud. mae gennow'r model digital i oedafur, a wnaeth i ffwrdd o ddweud arweithio crisis, mae'r wyrddau cynllun yn griffodol o ddweud o'r model. Mae genno ffwrdd agoedd diwethaf model oedafur, gallwn bywysol o ddweud i ddweud a gydag rhan o'r modelau deunydd, gweld a'r dweud o ddweud i gydag o leistau, a ddweud ar y ddweud o ddweud i'r ddechrau. Mae'r ddweud i greu ciolio cei ar allanod rhaid, ond y model ymlaen, rhan o gwahanol ar y model ymlaen ar y model yma, ond ydych chi'n gweithreitio'r bobl yn ei dynnu. Ac mae'r model yma yn gweithreitio'r model, rhan o hynny, yw ymchwil yn y gweithreitio'n gweithreitio. Mae'n ystod y gwahau i'r ddau cyfrifiadau, ac mae'n ddweudio'r ddweudio. Mae yma'r ddweudio ar y holl hwnnw, mae'n ddweudio ar gyfer y gallu ddweudio ar y holl hwnnw. Y ddechrau ar gyfer y cyd-dweithio cyllidegwyr yn gweithio'r cerddau ar y cyd-dweithio. Dydyn ni'n ddweithio cyd-dweithio ar y gyrdd yng Nghymru, a'r llwyffydd i'r ddweithio'r ddweithio'r cyd-dweithio. Mae'n ddweithio'n gweithio yn gymaint, mae'n ddweithio gyda'r ddweithio'r cyd-dweithio ar y cyd-dweithio. Mae'n ddweithio'r cyd-dweithio ar gyd-dweithio'r cyd-dweithio. Yn gyfyn ychydig, Ac mae'n gwybod a'r edrych yn ôl i'r cyflwyno. Ond ydych chi'n gweld y mynd i? Mae gennym, yw y gallwn y ddweud y ddweud argynnu ddweud, mae'n cyd-dweud y ffair yw, y febri yw y 2000, iddo fe, wedi bod yn ei ddweud ar y ddweud yn cyflwyno cymdeithas, ac mae'r ddweud yn cyflwyno ddweud yw ddweud, y ddweud ar y ddweud yw ddweud yn y cyflwyno, yn y ddweud ar y ddweud yw ddweud ar y cyflwyno, oedd y gallwn yn ei chyfwyr, ac mae'n cymryd wedi bod yn ddweud i'r ystafell hwnnw, rydyn ni'n ddweud, yn ddweud i'r ystafell hwnnw, a'r rhaid i'r ystafell hwnnw i'r ddweud i'r ddweud i'r datblygu'r cyffredinol. Rwy'n ddweud i'r ddweud, ac mae'n meddwl ar ystafell hwnnw, ac mae'r ddweud i'r ddweud. Ond, y dywed i'r ddweud, mae'r ddweud erioed yn gwybod. Mae'r ddweud yn gweithio rhesol. Mae pawb yw amgoedd o dweudio'r ingredients, ac maeherwydd yn eich gweithio y cyfle maer. A mae'b amsgwmpio'r frysgau mewn i wneud y ffordd oedd ymy乎, a mae'n gweithio dim ffrindio gyda gweithwyr fawr rymdau i chi'r unig. Mae angen i ddyn ni â'r antennae erbyn yn ysgol. Mae'n 10-m pasteurio ar gyfer y lliden. Mae'n mynd i dda i ddyniaeth o'r ffordd. Rydw i'r LIDAR. LIDAR yw rhoi radar, but with light. It's a very inventive acronym. Basically, you send a laser beam at your target, you receive it back again and you measure how long the radar beam took to go to your target. And then you put that laser on a thing that spins really fast and you have a thing that can scan the world around it at a fast pace. This here is a robot, but you see it on self-driving cars. Often they're on bottom of planes. And in general, what you use LiDAR for is more close in mapping. Mapping of city scale or individual terrain scale. So, for example, this here is an excerpt from an aerial LiDAR survey of the San Francisco Bay Area. They took a plane, you stick a big LiDAR pod in the bottom of it, and the plane flies in strips over the place you want to map. And the LiDAR device literally scans the ground below as you go. It gets all this detailed information. And with that, you can build up a really accurate model down to often 50-centimeter accuracy of the place below you. In the raw data here, you actually see the individual crossings and individual parts of streets in the data, and this is the low-resolution version. You can, of course, also do it not just from the air. Here at the site itself, there was a wonderful LiDAR scan done of the campsite before everyone arrived a few weeks ago. And so this is from a van that just has LiDAR sensors on it and a few cameras. And it's really versatile technology. You can map things at different scales. You can literally just map one or two walls or the interior of a room as well. And the nice thing about LiDAR is it's very high resolution. If you can get down to centimeter or sometimes of the new stuff, even millimeter-level accuracy, the devices are very compact. A standard LiDAR scanner is around this big. All the laser inforometry stuff is much smaller than LiDAR is in general. It only works over short distances. The problem with light in general is it attenuates over long distances very quickly, but generally you're doing it only from a maximum of a few kilometers away, even if you're flying up and it's probably not that far. So in general, for most things, apart from orbital missions, it works pretty well. And one of the nice things about LiDAR in general is that even though it's very expensive to do, a lot of free data is available from different governmental departments. In particular, our very own environmental agency in other parts of the UK government have free data available. The environment agency has basically mapped most of the floodplains and coastal areas in the UK with free LiDAR, and I believe there's an ongoing project to map a lot of the rest of the country in LiDAR for free as well. And over in the US, the US Geological Survey has mapped most of America in a sort of low resolution and again a lot of floodplains in coastal areas and high resolution. LiDAR is often used for floodplain and modelling and analysis, is where you see it often more near cities and rivers and things. So we have this data, we have a basic idea where it comes from, but what do we do with it? Well, the problem with LiDAR is it's not a nice simple elevation model, like I showed you earlier. What you get back from LiDAR is a thing called a point cloud. Now a point cloud is literally a load of points individually in space with no links with each other. You might get what colour they are, you might get what sort of intensity the laser bounce off of them, but there's no link. If you zoom in, you just go through this swarm of points, there's no solidity to them at all, you can't work out what the surface is. And so we have to take this point cloud and turn it into something we can actually use. We feed this to a 3D printer or a milling machine, it's not going to do anything, it's just literally a point, these are dimensionless points in space, they have no material being. And so that's one of the problems. But the other problem is it's also not pure data. LiDAR, when you get it back, there's a couple of things extra. When the laser hits something, it might hit more than one thing. You get what's called multiple returns. And so you may in fact get, in any given point, a series of diminishing points behind each other. If you're flying over something, you in fact may get several different features below you in the same vertical space. If you're trying to make an elevation map, you might get, well, there's a top of a building, but also we went through the glass to the floor below it and got like a weak return from that. Or we can see through trees and things like that too. And generally it's just quite noisy. The data tends to be marred by reflections. Water and steel is especially bad. Also if you go anywhere near like a power plant and something hot, one of my favourite examples recently is looking at a bit of Didcot. And Didcot's pretty well mapped. Apart from over the power plant, there's a sort of giant mess, as it sort of starts going near all the giant steam emissions. There's a quick example here. This is the sea off of San Francisco again. There's just random spikes in the middle of the sea. I can reassure you there are no random spikes in the Pacific Ocean. I've checked myself. But the light are occasionally bound off a wave and things is a spike there. So you have to sort of get rid of that stuff. And in general we have to do more than that. We have to take our point cloud. We've had to thin it. We have to take the point cloud and go, well, there's individual data here to much higher accuracy than we need. I'm not printing a tree in this kind of size. I'm printing a much sort of zoomed out model. And so we have to take all those points and reduce the resolution. Often what I'll do is if I'm doing a service model, I'll say, okay, for each small like 10 centimeter grid, find the highest point and just give me that. And that way you can thin it out to a nice grid. You then have to take those points and turn them into an elevation model. Often this is easy as looking in that grid square, taking the highest point and just writing it into a file. These two first steps can be done with a variety of pieces of software. Unfortunately none of them are entirely free. I use what's called LANs tools, which is free for small amounts of data and evaluation use, but if you're to use it commercially you're going to pay a licensing fee. If you're lucky and your department is good, and this includes the UK government who are good at this, they will just ship you ready-made DEMs from the LiDAR data so you can skip the first two steps. You can also have some smoother results. This is done with, in my case, some special code that sort of tries to average out the different stuff, but you have different techniques here. You can try clipping, you can try reducing stuff. And one problem I had was when I got data back from a LiDAR scan of London, it was when Crossrail was being dug initially, about 2014 or so. And the data was like, oh, there's loads of points that are like minus 40 metres below sea level. And of course it had seen the giant pit they'd put the TBMs in. And so I had to, rather than print a model that had this thing with a big hole in it, I had to tell it to smooth it and bring it up to sea level. Like nothing should be below sea level. We'll ignore the giant hole on the ground for now. So that's kind of the easy part. Taking the data, getting the data, there's one place to do this. And it's not easy, I'm being a little bit overzealous here, but we have GIS tools, we have ways to do this. It's tutorials online. If you want to find elevation data for a part of the world, you're at least going to have a rough resolution of data. You might get mountain level rather than city level. But then we come to fabrication. Now, one of the things I was maybe a little bit too, shall we say, overzealous about initially was, oh, it'd be really easy to make these. How hard can it be? Well, let's go through the three basic techniques I've been using here. 3D printing, which is what makes the city models, things like these tiles here. Laser cutting, which is what makes those sort of wave-like shapes, like this one here. And then milling, which is what makes the small metallic shapes as well. So let's start with 3D printing. 3D printing is thankfully these days pretty cheap, very really, like quite accessible for a lot of people, and quite easy to get stuff into. The basic requirement is to have a 3D model. That's kind of it. Once you've got a good 3D model, the slicer will take care of the rest and give you a good toolpath and let your printer do most of the work for you. So the basic thing here is to take that data and turn it into a 3D model. This is achieved, in my case, through a wonderful Python script. It is open source, links at the end, if you're interested. But this takes the data, it makes it into 15 by 15 centimeter models as STL files. Now this data is quite detailed. It tends to be on the order of 25 to 50 megabytes per tile. Some 3D printers do not like this and will just sit there and complain for about 10 minutes and then fail. So there's options for making it less and less detailed so it sort of fits in the printer. But in my case, I took these big tiles, I put them on one of my 3D printers at home. This is a Cetus Mark III and left it to print. So 3D printers, thankfully, are unattended. They're fully automated. Now the problem is this is a very detailed model. It's got a lot of fine detail and like tiny buildings. It's got to like print a little bit, lift the print head, travel over print a little bit again and keep doing that. This means the tile of this size, which is about this big, is between four and 12 hours of continuous printing. You want to sort of put it on and then just go to bed, wake up in the morning and hope you don't find a big spaghetti mess of plastic in the morning, which unfortunately has happened sometimes. And then the problem is you're printing tiles. You've got to assemble those tiles into a bigger hole. Now ideally that looks like this, but this is after hours and hours of meticulous work, sanding the size of the tiles down, making sure they're perfectly square, aligning them and trying to glue them perfectly. And in the end it works out pretty well. If you look very closely, you can see the seams between the different tiles. My 3D printer's not exactly making perfectly straight edges, but it generally works quite well. And again, here's the London one. This is a bit more like East Endersey. You can sort of zoom in with your head if you like and sing the theme tune. If you look closely here, maybe point out on this projector, you can see visible seam lines in this one. These tiles aren't as good. In this case, my printer was not behaving perfectly right. And so I have to try to do a second part of these and sand them down to perfectly square. And so it's a relatively easy method and what I like is it's mostly unattended. You run the script, it makes a file, you press print and then you just go and do something else for 12 hours. There is a little bit of manual work involved, that sort of lining up portion and making sure it's all corrects and the gluing stage is a bit tense because like, well, once the glue's on, you've got 20 seconds or so of work time for its sets. And if you get it wrong, you can peel it off and have it looks all right again. So it's a good start. It's not particularly durable. The ones you've seen here are printed using PLA. If you know 3D printing, you'll know that's a cheap but not relatively durable plastic. I want to move up to ABS and other things too. There's a variety of plastics you 3D print with. But in general, 3D printer stuff is a little bit brittle and it's a little bit hard to deal with. So let's move on to something that's much more durable, which is milling. So if you're not familiar, milling is sort of the opposite of 3D printing. 3D printing is additive. You start with nothing and then it adds plastic to build the model up. Milling is different. It's subtractive. You start with a big block of metal, in my case aluminium, because aluminium is cheap and easy to work with. And then you have a mill which progressively carves out the shape with multiple passes. In sort of a similar way you build up layers in 3D printer, you carve out layers with a mill, smaller and smaller and smaller to build out a shape. Here you can see it's about halfway down a copy of Mount Shasta in California. Now there's a couple of problems of milling. First and foremost is you've got a certain shape of milling bit. Often it's circular. You can have a flat end or a round end. This one here is a round end in one. But if you want to do small detail, you need a small bit. In fact you need to be very, very small. But the smaller it is, the slower it goes. And so you want to do a big, rough pass with your big, nice big milling bit first and then a sort of medium pass with a half size one and then a final pass with a small size one. And in between these you've got to have the machine come up and let you take the bit out and put the new one in and know the depth properly and it is a lot of manual work. Whereas 3D printing is unattended, milling is not. It's kind of unattended. You can kind of sit there while it does each pass. It's about an hour or so per pass. But between those you've got to get up, change the tool out and do all that kind of stuff. There are machines that have automated tool changing I would probably give some sort of limb to have access to one but they're very expensive. And so the one in my local maker space in Oakland, in California does not have any of that. It's very much a you get to put new bits in and do it all manually kind of machine. But the nice thing is the final result looks really good. You get these sort of very small detailed models almost intricate in their design of various different national parks. And things like the mountains and the terrain come out really well. These are the six I've made so far. Given how long they are to make it requires me to have an entire free evening just go and sit next to the machine and just listen to a podcast while it was away and hope it doesn't break anything. I have broken about 10 milling bits at least making these six of them because as you go smaller and smaller they break really easily. So I'd be carving out a valley and go and just break off. The other problem with milling is it's not quite as simple to just feed a model into the machine. You have to take the 3D model feed it into a CNC program. I use Fusion from Autodesk and then sit there and manually tweak the tool pass and tell it the strategies to use. You say oh I want you to do this kind of flat pass first and then a pass with this milling bit. And it's a good 30 minutes or so each one of these of me sitting down and planning out how the machine is going to work. Most of the routing is done for you but the strategy, the high level idea is yours and you sit there and go well can I shave half an hour off the time by changing the strategy a little bit? So it's a lot more involved in general which is why they're going a little bit slower. They are a bit faster than the 3D printed ones because they're a lot smaller obviously. If you were going to mill a one meter thing it would take almost forever I imagine on the machines I have access to and is attended work. You can't leave a CNC mill alone. They're giant spinning wheels of death. They're kind of dangerous to get near. So you want to make sure they're attended and not going to sort of like ram themselves straight into the side of the piece and just spray aluminium everywhere. The small features also are hard to mill. I have an example in my bag you can come and see it later if you want and it shows very well like there's a little bit between two mountains that you just couldn't get to. There wasn't a bit small enough to mill that part out. So you lose a bit of features in those deep valleys which is a problem for anything with mountains in it. And finally CNC mills are one of the most expensive things I'm going to refer to in this presentation. The entry-level model is like 10,000 pounds and also you have to have some light industrial space to put it in and also you have to have like some air extraction in that space and also you then need all the tools as well. I would love to have one. I do not have the money or the space to do so and so thankfully maker spaces kind of solve this problem but even then a lot of maker spaces don't have the budget or the expertise to run a CNC mill as well. It's kind of rare by itself. So that's often kind of the big blocker to doing this as a project. I really hope in the next few years we get a more of a revolution in desktop or smaller CNC mills. There's a few of them out now but we could make a lot bigger steps in having smaller more accessible milling for this kind of stuff. But finally everyone's favourite laser cutting. I think pretty much every maker space from here to Sydney has a laser cutter. They're incredibly popular and incredibly easy to use. I love the things. That's one of the reasons I then took to the laser cutter for this third project because it's something that I can literally just plan out an SVG file, send it to the laser. It's about 15 minutes to cut all the pieces out and then a little bit of gluing. This here is an example of one of those sort of profiles that's being cut by the laser. Now I have a piece of software that simply takes the terrain data you want. You slice lines out of it in two dimensions and then for each of those lines you take the sample of heights and turn them into a side-on view as a solid object. That ends you up with about 30 or 40 of those individual pieces. Obviously I would cut a giant piece of plastic. This is an illustration of one of them. You get this big piece of plastic, you pop all the pieces out, you glue them on to a nice little ridged bed and you end up with stuff like this. This is my first attempt. This is the wonderful Scotland. It's not great. I used clear acrylic. I didn't realise that because it's clear you can't see it. It's kind of an obvious thing. It looks lovely when you shine a light from the back of it but unfortunately it's very hard to photograph. Another problem is Scotland is almost too big. When you're zoomed out that far the height of the mountains is not that high relative to the height the size of the landmass. I try to exaggerate the height and say, oh, make the mountains way higher but it doesn't quite look right. I'm still working this. I want a nice Scotland on the wall but it's taking a bit of work. What is easy is Hawaii because it's giant and surrounded by sea. So various islands of Hawaii you can do this too pretty easily because they are very, very high like 10 to 12,000 feet each. You can drive from sea level to 10,000 feet in an hour. It's ridiculous. And also they're quite small. They're only about 10, 15 miles on each side. And so you get that nice ratio of height to depth. I also think things like the South Island of New Zealand could be good for this. There's a few other options as well. There are a few problems with this though. The sort of presentation of this has a few flaws. First of all, you can't see it if you're looking straight at it because it has to be viewed from the side which actually means you want to do them side on so you can see them as you walk around the room. If you do them vertically then people of a certain height will never see the artwork. If you're short at all, it's great but if you're exactly at an average height you're like, why is there just a random grid on the wall? It doesn't make no sense. Other problems is I tried it with things that were sort of cut out portions of landscapes but it kind of needs things that are surrounded by flat edges. Islands obviously perfect for this but also if you've got sort of a big mount in the middle of a nice flat area there's a few of those around the world that would work pretty well. And of course it's limited detail. This is much more of an artistic interpretation piece most other ones. It's very much going to show you the rough outline of how it looks. It's not really going to show you fine detail and the script takes a lot of liberties in exaggerating peaks in fact it multiplies everything to a power of two to really push up the mountains and bring down the valleys and you see those details much more clearly. That's kind of the secret of all of these. Everything I've showed you the London model all the city models everything has the height exaggerated because if you just look at a normal scale model you're like, oh that's really flat, it's boring. You have to have that exaggeration to make it seem interesting to have it see how you see it when you look at the window of an airplane for example. So that's a brief rundown of the stuff. There's a few things next I want to work on. These projects all kind of in progress along with my other like 3,000 hobbies they all get like time shared but one of the key things is I want to do more milling there are 53 national parks remaining. I do not have 53 spare evenings in the near future and so I need to work out a way to do a less human intensive version of that. Whether it's doing them quicker or whether it's somehow finding a milling machine with an automatic tool changer and like bargaining with someone for that or sending away for even I'm not sure. For laser cutting I want to work on the feature highlighting and I said it's getting Scotland working great it's a case of working at how do you make a nice defined coastline but still bring the mountains up to be really obvious. I want to see the great glen in perfect detail while still seeing the shape of the landmass and that could be really hard to do when the edges can be quite flat compared to the mountains around it. With point clouds and lidar one of the big things we'll do there is 3D directly from the point cloud. What I've shown you here is taking the point cloud making it into a 2D height map and then making that into a model but that what that means is you lose undercuts you lose like buildings that have things underneath them and I'll you know in London like the gherkin would render is just the top half and straight sides and so you need to do a bit more for that. There's some commercial solutions but I'm a cheapscape and so trying to find a way to make a script that actually pauses point clouds directly to surfaces is my next interest and then maybe I can get a lidar scanner. When I said the CNC mill was the second most expensive thing lidar scanners are the most expensive thing. The entry models are like 20,000 pounds. I do not have that anywhere. There are handheld ones but it's one of the things like I'm hoping like as we get them and like they used like as a crucial component of self-driving cars they come down in price a bit. If you're curious of more of this some of the code to make 3D models is up on my GitHub. There's a few blog articles with some of the process of this as well and there is a couple of videos of me making the process and going through the big city models on my YouTube channel as well. It's about 20 minutes of me mostly faffing around gluing things badly. Please don't hate me for my bad gluing. And that's basically it. Thank you very much. Yes. We can do questions. We have the giant throwing box of questioning. There's a hand up over here. So do I just I guess I just talk in the box then? Yes. To hold it very close and talk loudly. That's awesome. Thank you very much for the talk. It is really good actually. I had a question about your actually more of an idea. I really like the idea of laser cutting the edges and then stacking them. That's really cool. And I also understand you've got a problem for looking directly at it. You wouldn't be able to see. So what about maybe building your model and then putting it into a vacuum former? So this way it would actually feel put a very thin sheet of plastic between all of your layers basically and cannot fill it. So this way you'd have a completely filled model with shapes so wherever you look at it you'll you'll still get a light refracting from it and it would really pop I think. Yeah that's a good idea actually. Another one on the same thing is I've also tried casting those shapes into and like to get a negative of them basically. That's another way of doing that too. Okay. And yeah like there's a lot of interpretations of that like this is only a few weeks old. The laser cutting is like it's the most recent thing it's the least developed I'd say. Yeah that's a good idea I think. Because I think you could potentially just stick it in and just press it down and that might work. Yeah it's the wonder of vacuum form you just pull the lever and it works. Yeah exactly. Maybe another thing is expanding foam. Do you think that could work? I haven't got any of that around it's kind of messy though. It's kind of messy though. Yeah that's right you need to do a lot of cutting. Yeah but yeah this is a good idea so thank you. Thank you very much. Yeah. Any more? Oh over there do you want to throw over there? Wheeeee Oh too much. Does it still work? Yeah it's working. Yeah. Have you seen the Polish map of Scotland? The what sorry? Okay so. No it's the answer but yes. So about an hour south of Edinburgh near Peebles there is the what claims to be the world's largest terrain relief map called the Polish map of Scotland or Map of Scotland and it's made of concrete and it's 10x vertically like exaggerated basically but it's probably as big as this tent if not bigger. It's next to a like a country home but it's really cool. That sounds amazing. That's it. Okay that's going on my on my list of things to go see. Thank you very much. Over here. One more? Yep. One more question okay. It's just about your milling. Yes. Whether you've considered milling perhaps wax or much softer materials which you could potentially then lost wax cast or something like that. Yeah I think in terms of increasing the speed I was looking at literally milling machining wax casting that and then casting it into a resin. The problem is I quite like having aluminium as the result and I don't really have the ability to you know forge or cast metals and so it definitely would be a way of increasing speed and making like nice resin versions. I'm not sure I can get the same like aluminium like machine effect out of it. Is it specifically aluminium or you're just wanting a harder metal? Well I mean the softer metal the faster it is right. Bronze or aluminium is quicker. You can mill iron I guess. Bronze is something you could sort of potentially cast. Yeah but yeah. I haven't done the milling for a couple of months now. I lost access to the machine a while ago and just got back access to a different machine. So I need to start on that new machine basically. Thank you. Cool. Oh there's one more question in the back. Oh no come on let's go. Go long. I'll go long. There we go. Okay yeah there we go. Away from people even better. And the scan that you did of this site. Yes. That looked like it was done optically rather than LiDAR as in was that a camera? Of course it's not my scan. Somebody else's scan. I've blanked on the name temporarily. Yes. I've done some work using Open DroneMap and it looked very similar to that and one of the advantages of that you could just take video or photo images and it stitches them. Right. And I wondered if that was done the same way. No so that scan was done using a professional LiDAR van with a LiDAR rangefinder I believe. You can do what's called photogrammetry which is the point we're talking about where you're like you take lots of pictures and then you use the just differences in the pictures to work out the shape of the terrain and that's much cheaper because you're using the camera basically. It's also much less accurate and doing the stitching into 3D models takes forever. I have some photogrammetry of Crater Lake in Oregon and it took like 3 hours to render the model out of it so it's possible but it's difficult. Indeed I did a scan of our local hair space and that was a 12 hour render. Right exactly. So in terms of like throughput and accuracy like LiDAR is the thing and also it's being improved a lot too but then also maybe photogrammetries as well. Thank you. Oh thank you. I don't show that on the question. Okay. Yeah. Yes. Is that the scan? The person to the scan is right here. He's amazing. Please ask him questions. All right. Actually I've got one question. Oh you've got a question? Okay. Oh sure. Go for it. So actually could you make your own LiDAR with like an old laser printer or something like that mechanism for less than 20 grand or something? The laser is not the hard part. It's the receiving of the laser and the hard part. Shining laser is very easy. Accurately measuring the time of flight much harder. There is decent entry level like single laser stuff it's the case of like taking that mounting on an accurate gimbal and mapping and stuff like that basically. And I'm probably going to get like a the most basic laser range finder I can try and like try to a sort of cheap hacky version of it I think is my next project. All right. Thank you very much. Thank you very much. Thank you.