AVSP_Episode793

0:00You're listening to a very spatial podcast episode 793 September 13th, 2026. Welcome to a very spatial podcast. I'm Jesse. I'm Sue. I'm Barb. And this is Frank. And this week we're going to be talking to another person that was at the, as we you see this year, but first, of course, we have lots of news, including Atlas AI. So what we talked about world labs, whenever they first organized, and I think that was like two, maybe three years ago now, and their focus, of course, isn't so much about large language models, it's about spatial models, and trying to understand how to use AI for geographic information for spatial intelligence as they call it. And Atlas, I think, is their second full model release.

1:06And it's basically, it's what you can use to create scenes. And so I'm, you know, one of those things from a geographic perspective, yes, I want things in a small area to make sense, but I also want things and large areas to make sense. So it's kind of this focus on small area for the most part, or using imagery for reconstruction. So it's a few different things that allows you to do. So yeah, we're getting into that space more and more in AI now. They have a little bit frightening verbiage, I think, on their announcement. It's the reasons that they want to do this is to understand how worlds appear and behave and evolve. That's their language that's perfectly fine. But the end with, and help robots, robots, plan actions. No, I don't, I don't want the robots to have a tool for planning actions. I saw the matrix. It's a bad idea. I think for me, of course, with a lot of the, the, Gaussian splats and stuff that we're, we're playing around with more and more as we get different

2:09tools, access to different tools, this suggests what people have been talking about, because I haven't had a chance to, well, we haven't requested access. So we haven't used it. So, but from what I've heard, it's generating really nice models with fewer images, which is always a weird thing with Gaussian splats, you know, that, that tipping point of having a good number of images to get a good quality render versus having too many pictures and it kind of creating weird things. But yeah. The really interesting bit is the regression part that we're talking about is that they're using the small number of images and doing to oversimplify a regression type of model to, to infer things that should logically be there based upon a whole lot of data that they probably collected and analyzed, which is kind of cool. But I agree with you, the less, less is more thing. Just rubs me wrong because everything else we do in the geospatial realm is more pixels is more,

3:12less pixels is not more. And so it is a little, it's interesting. And all the videos they show aren't necessarily rendering real spaces when they are, but they're not necessarily rendering real geographies, I guess, for a better term. And I am kind of curious as to when you do the differential comparing a real place to what they come up with, what's the regression catch and surprisingly, what is the regression leave out for like a better term? Yeah, I mean, you've got that counterpoint that some of this is based on real world data captures. Others are generating scenes. So you've got that world well knowledge built in in the model, but you've also got that ability to create representations using the models and different ways. And you know, even getting into the 4D aspects, because whenever we talk about Gaussian splats, they're often abbreviated as 3DGS. But if you take pictures, multiple pictures at over time, you can actually get motion in these as well. And

4:18if you go over and watch any of the YouTube videos from a cameramer's name, but he's the guy who is kind of behind Radiance Fields, the website, you know, he's been creating 4DGS videos. So video is based on 4D Gaussian splat data sets. So you can kind of see where that's going as well. And there's a picture of a strawberry dropping. And that's an example from this on the announcement page. I can see where for world building and modeling, this would increase efficiency for you to do your work and then go back and check and update, adapt, and change things. And I wonder how much would you have to do for correcting after the fact. So you know, but looking at the videos, they look amazing. But again, how much of that is the video and how much of that is the actual tool? Because anytime I see really awesome videos, you know, I go, well, this is the best that you've got to, you know, with what you're doing. But they're amazing.

5:22Next up in the news, NASA and IBM have partnered together to make a lunar AI model. This is in support of going back to the moon and possibly making a more permanent or certainly more permanent base on the moon. The problem is, is there's an awful lot of the moon that we haven't really completely mapped out. So this AI model is an attempt to kind of fill in the gaps in the information, particularly in places around the poles where you're talking about ice may exist and it can be used as resources if we have a more permanent base on the moon. It's pretty cool, I think, and kind of exciting that they're using this technology to kind of somewhat fill in the gaps. Having said that, there's a quote in here by one of the leads for this that says, experience travelers know to get the lay of the land before you're setting out for foreign destination. That sounds great, but also in almost all those cases, the foreign destination, people live there, and it's well documented, and there's a lot of information you can pull into. This is a little

6:22different because by definition, it's not definition by their mission and it's stuff that is things that we don't know really well. So I am again, like the last news on it, I am curious exactly where that differential ends up being between what AI inferred versus what we end up measuring in the end. And I think this is a case where scale really, really matters because you get it right from a distance, a certain scale, but then your lunar land just coming down and it turns out that the landing site that looked like a little bit of a shadow, you're coming down right by a cliff or something. And so that's that question, right? Is yes, on some level, it's useful, maybe? And I think it's cool, by the way, I do think it's cool, but that is really a question when you're talking at the scale at which operations would actually occur. And I thought this would be very useful for partnering with the NASA Rose grants that have been around a very long time for this type of mapping and exploration of planets not a earth as they they term them, not

7:24planets, but bodies not a earth. So it's pretty exciting. To me, this would be an amazing tool to use to help guide or suggest places we should look harder, which hopefully that's how they're going to use this model is to say, okay, this is what we think is probably there so we can use our better equipment to focus on that instead of trying to do, you know, scan all of the moon at the super high resolution. But you're right, Sue, you know, having 24 contours is amazing, but that's totally different than having one foot contours, right? There's a lot of stuff that can happen between those isolines. So I think it's a tool that guide where to look harder, it'd be a really, really powerful tool. More importantly, you guess all of you can go out and download this from Hugging Face and check out what NASA and IBM have put together and plan your own moon mission, because, you know, you have the launch capacity and all that to do that, but hey, it's still fun to go out and play with different models. And so that's another one of those that's a spatial model.

8:26Now, speaking of Hugging Face, Nvidia apparently is going to acquire Hugging Face. And so if you're familiar with Hugging Face, if you've been doing anything with AI or any, you know, any of the things that Hugging Face has been involved with over the years, it is now the, essentially go to portal for open models and those type of things where you can easily go to it. You know, if you're using any of the local AI tools, most of them have a link in the Hugging Face for you to be able to pull down individual models, Ezri at the user conference. I was talking about, you know, how their models were being, a lot of their models were being put on Hugging Face for everybody to have access to. And so, but yeah, with that acquisition, my peer-lil here is GitHub and Microsoft's acquisition of that. So I don't know, what are your all thoughts on this? Okay, first of all, when you look at the link in the show notes, you're going to look at the number that they used to acquire Hugging Face and you're going to have to, if you're like me, you're going to look at it like three times ago, yes, there are that many zeros in that number. That's a

9:30really big number. And is it be at the end of that? Yeah, yeah, that's a really big number. And that says more about the capability of Nvidia in this age to look at that and go, yeah, that makes sense. You know, that that the size of that company, that's an aside, but well, that that can take us into a whole conversation about the bubble. But yeah, that's yeah, that's an aside. Now, and given that it's not in the least bit surprising that Nvidia would acquire something like this. And it's actually a little shocking that Nvidia isn't gobbling up companies like crazy like this at an even greater rate than they are. I agree with you, this is kind of a Microsoft acquiring thing question. One of the things we don't know particularly well about Nvidia right now, particularly when it's something like this, is exactly what they're going to do with it and how they're going to treat it. They don't have a strong history in Microsoft tends to buy things and incorporate them and, and you know, absorb them. That's just a pretty normal thing. Google does

10:31similar things, right? They buy it and they go, okay, this is now ours and we'll use the technology or we'll incorporate them slap a Google label on it. It's kind of interesting that it's going to, I don't know if I want to be too word interesting. The concern I think is how's Nvidia going to going to treat an asset like this? And that's to be seen. Yeah, I'm sticking with the the GitHub parallel. The GitHub does still exist. It's still a standalone thing. Yes, it is integrated now into visual studio and those type of things, but you don't have to use Visual Studio to use GitHub. In fact, tons of people probably more people are not using Visual Studio who are posting things there than are, but I think I think it's that kind of thing that they understand that to have their tools be more broadly accessed, you've got to make sure that this repository that allows people to use local models or, you know, use them at local models at enterprise levels, that people can still

11:32have this place to share them. So that's why I'm thinking of it like the GitHub comparison. And in their announcement, they think with that way when they said that they will continue, well, it will continue to remain an open platform and Nvidia compute will not be required to build on or deploy. So that is what they're saying. Yeah, and I think in this case, it has to be, when you look at their chart of downloads and you see who's on there and where they rank towards the top of 500, you know, 500 downloads and more, you know, they go by month from 2025 that it's just showing the power of people and the open community, you know, that's what this is, was built on. This is what people are seeing value in. And I don't think you can can separate the two, especially when you look there, they're in that graph above that I think that's meta, that symbol and above Microsoft. So at the risk of slamming Jesse too hard, it's a little uncomfortable for me being the pessimist in this particular scenario. I hope I am utterly wrong, but I do have fears that we don't

12:38Nvidia so large right now. And like Jesse said, it gets in the questions of the bubble and on and stuff, but let's assume that let's not touch the bubble, but it's so large and they make all their money being the shovel manufacturer and less the doing thing. And so I have concerns and I hope I'm really wrong. Well, we can also talk about how we've seen John Deere change over time, you know, how it's gone from being the shovel maker literally to, you know, the data company. So it's I, yeah, I get your concerns, but I was just reaching out for your now where I had seen a similar acquisition in that, you know, the GitHub one is just the one that makes the most sense as a similar output that if all goes well, we'll still have access to it. Now there aren't issues will get up. Don't give me wrong. I agree with you, but I also kind of think that Microsoft hasn't quite figured what to do with GitHub yet. So I think the other shoe is ready to drop at some

13:40point. But again, that's the pessimist. Google deep mine has released the weather next three and updated version of its AI powered weather model. It claims it's going to be a lot more detail than its forecasts and the number of variables that it covers. What's interesting to me is a lot of the focuses on things like large solar farms. So not necessarily an individual personal weather, but for what we know that these models are used for things like knowing the weather predicting it sooner for its impact on larger areas. The really interesting thing here is that it's using AI's is more of an analysis tool rather than waiting on the actual data as sort of predicted. And the really cool thing here is they're going from a 25 kilometer, a square kilometer model to a five kilometers per model, which is pretty impressive that they're increasing the resolution that much in the in the predictiveness. I'm assuming you have to assume that they said there's

14:40a six hour window between the data actually being released and when they come and need it. So with that six hour lag, hopefully that just makes all these models that miss more powerful as they can change course in their prediction as they get real time data. And right now you can only get weather next three behind whether for Google search Gemini and Google Maps. And I assume the API out yeah, the API is Google Earth Engine. So you know, that's where it's going to be deployed right now. But here's something I did not know that the original weather next. So this one is weather next three is actually been open source test of last month. So if you want to play around with with those model, the older version, you can't do that. And the new one you can also play around with in Google weather lab. So some tools if you want to work with it and you're not necessarily using where it's deployed. And like Frank was saying, you know, talking about going from that 25 meter pixel down to the five meter pixel, sorry, kilometer, 25 kilometer pixel down to 25 kilometer pixel.

15:44It's huge in terms of, you know, that resolution change. And the fact that you can that they're generating hourly forecasts versus their previous six hour forecasts is also, you know, a jump in the temporal resolution as well. That this is, I was recounting to the crew before we started recording that my experience with weather yesterday, because we had some flash flooding events on the Pittsburgh area down into West Virginia. And the temporal thing, I think, is huge knowing I'm getting a much better estimate of when these events are going to happen, a lot should prepare and plan a lot sort of stuff like that. I happen to be driving from the roughly the Pittsburgh area back towards West Virginia. And I was trying to plan my trip so that I would be ahead of it. And I was, I planned poorly and I get caught in it to some extent. Luckily, I was going faster than the weather. So I was able to get ahead of it. But you know, that was based upon information that was six or eight years old, eight years, six or eight hours old at the time, because that's

16:49the last time I checked the weather. But, you know, having a more of an update from my phone or something like that, telling me, Hey, you know, in the next hour, it's going to get really, really nasty would be so nice. Next up in the news, a couple of pieces about something that generally is just for us in geography, cartographers, us in GIS, and that is map projections on a smaller level France cited to endorse a different projection. So this is all kind of pushing back against Mercator and its distortions of size. And so France decided to stop for their foreign ministry, stop using Mercator in favor of Eckert for, which is probably like, you're like, what even is that? So that's a projection that's actually been around since the early part of the 20th century. But I believe it's an equal area of projection and it's sort of pseudo cylindrical. Anyway, lots of technical stuff. But just to say that it's a recognition that the way we have to take a 3D Earth and put it on a 2D introduces distortion. And so every map that you see that has any kind

17:53of distance to it, right, is going to have some distortion. And so map projections are our way to control that distortion to pick one or more of the properties and make them as true as possible and understand that we'll have to distort other things. So France went with Eckert. The vote was very recent. And so therefore, in ministry will stop using Mercator in favor of this projection. In the other vote, the United Nations, which was prompted by a resolution from Togo, African country, voted to encourage the use of non-mercator maps in order to recognize the size of Africa and relative size around the world. You can kind of imagine why an African country would lead the resolution as to wanting Africa to be recognized as not the size of Greenland. But there you go. So yeah, what projection they do is did they suggest even though they're not requiring it, this is their suggestion as an option. Equal Earth, which is a pretty new one actually. But like

18:59projections like Robinson, it actually distorts all of the properties, but does so in a pleasing way. And it's important to remember that there are no perfect projections. They're, they have benefits and minuses depending on what you're trying to do and you kind of have to match the projection to what you're trying to accomplish. So Mercator projection was about navigation that is arguably less dominant now than it was 300 years ago. So it makes sense for us to think about this in other contexts and other purposes to think about our projections. However, that being said, I think it's good that we're moving away from Mercator projection. I personally have a little bit of concerns about the fact that everybody's going to be grabbing their own particular version of a projection. The good news is the software kind of is smart enough now that it wasn't like it was 20 years ago. Now you can just, it'll just go look, I see them with projections. I'll re-project everything. Don't worry about it. It didn't use to work that way, but it is going to, I think it's going to lead

20:02a little bit of confusion. It's kind of like having different size meters. It's just like, sometimes we may end up having to come with an international standard at some point, which theoretically the UN is trying to do that. But as we saw in the news, France is doing something slightly different. I would say in some ways it was artificially pushed standard for Mercator through the Google Maps. I believe that they use Mercator, correct? Yes, it is. Webricator. Yeah. And most likely we're not going to be seeing Mercator go away. I don't assume that Apple Maps or Google Maps or even ArcGSL Line are going to enforce a move away from it. But it is what it is on that side. But whenever you're creating maps, again, and one of the things in the resolution, yeah, resolution that Togo put out there is that it's as much about educating people about projections and about the fact that none of them are correct. And that, you know, what each one does is just a little bit different way of showing our joy. And I think, you know,

21:10a lot of people are missing that push for the education aspect. And if you go over to places like the map room, Jonathan Crowe has been doing follow ups on what people have been saying about this after the fact, both from various government organizations about why they either abstained or voted against. US was the only one who voted against everybody. There were six, I think, abstentions. And so, you know, we're finding out why people either did like or didn't like this motion as well. Yeah. And I think, well, two things really quick one. I think it's important to mention it because I know my students, right? In the education setting misunderstood what was going on because I actually had two students in a G. H. S. intro class asked me why the UN was interested in a map that would make Africa bigger. I had to explain that that's not what's going on. It's a difference representation. But I do think that, you know, a lot of this is only really if you have maps where you're talking about a regional or even a world setting. So in a, in a

22:15smaller area, first of all, whatever projection you usually just going to continue on with whatever your workflow was. But I will say that that the maps I remember growing up right where in the atlases and things were different projections. And there they were much more interested in preserving shape often. So for example, Alper's Equal Area was often the one used from North America because of its ability to represent shape. So in any event, I do think that the fact that it got out into a broader median discussion among non-carto-crovers was kind of interesting. It got picked up. And yes, many, many, many, many clarifications and questions. But it is true, right? Maps, maps are power in the sense of how they visually represent things. So in terms of geography awareness, I think it has done a good job. But like Sioux said, it doesn't go a lot of times deep enough to explain it. But I will say I have never seen that clip from the West Wing about projections as much as I have after this news came out. So I think every educator everywhere was excited to put that out there. Yeah, we

23:16we have a clip for this. Well, the problem is, of course, that the Peter's projection actually feeds into some of the issues with Mercator. And so that's the reason to push for Robinson equal is to get away from that rectangular shape. And I hate to say it, but I think we can kind of blame Ezra a little bit of this for the professional use of these rectangular. UTM. I don't know. Most of us got supposed to Mercator. Again, that's not Ezra's. No, it's not, but I want to say that. But their use, their, their enforcement of putting base maps in those base maps being in Web Mercator just moved us in the professional setting from just making maps and using data that you know, we went through and chose a projection to just kind of going along with that projection. Now that was bad. Yeah, it's just as long as you know what you're doing. And so many people are doing GIS now that don't understand projections and you know, how it impacts people's perception

24:19of the world. Yeah, I'm you're absolutely right. The the reason the West Virginia chose a lot of its map storage and the projection it did was because people are going to throw it on Ezra maps at some level. And because of ArcGS online and those platforms it was well, let's just make it work of that because that's what makes it very easy for us to launch products and tools and things like that. Even though it isn't the best analytical projection for a lot of the things that people want to use more localized like state resources or even county resources to do. But it was well, it's easier to put into Ezra. So you I think you're absolutely right. I think Barbus, right, they kind of just followed Google, but they really should have been smart enough and knowledgeable enough to have a conversation to go, does this make the most most sense? Well, but I mean, as a as a coder, you have to admit, right, that making that that rectangular shape is is way easier than writing the code for any projection besides that. Yeah, yes, absolutely. But anyway, that's almost

25:22a whole topic in its own right. Just a quick note that Trimble has released SketchUp Pro's Civil Contractor. So this is kind of a mashup of their course traditional SketchUp product, but also their sites product so that our site works product and ortho works as well so that, you know, you can use something like SketchUp to then plan and then implement site work out on construction sites, you know, to go out and collect data about what needs to happen or what is there and then show what needs to be cut down and where site materials need to move from point to point be in those type of things. So yeah, just a slight different professional use case of SketchUp than we've seen in the past. And that's it for the news. you

26:28in the web corner, I was introduced by a photographer today that white, he's a poet who's using maps with his poetry when I went and did the the free tour of the site, which is beautiful. I was looking at it and a lot of it is interactive mapping, including things exploring poetry is written about his his home and other things. It's it's a very beautiful site and it's a really good example of integrating in one area of the humanities like poetry into the technology that's available. It is a paid site because of the the amount of poetry and work being put into it, but I have a lot of times people that are creative don't think about the ways they can use spatial technologies in the modern world to to reimagine what they're doing. A lot of this reminds me of stuff that was going on with the or has continued on with the Lake District of poetry around that and trying to get that into some sort of digital humanities spatial context deep mapping as one example, but

27:29you know it reminds me of that which is really cool stuff. Nick and Laura just stepped off the stage to give us a lovely episode of the Environmental Professionals Radio, EPR podcast and so I wanted to just ask them about their experience here at the UC. Well, Jesse, this is both of our first time at the UC. I used to do actually did GIS work in my environmental career which is 10 years now in the past and I had always wanted to come so this was this has been a really great treat for me. Yeah and like we've worked with Sunny Fleming for like five years now so it's been a long time coming we were excited to do it and I'm really happy to be here. It's huge in a good way so it's been really fun. I think one of the things that you know you were saying that you were talking to three great people who were focusing on conservation.

28:33One of the ideas that came out that everybody kind of picked up on was this idea of bravery and us as a community whether we're talking about sustainability or conservation or resilience, just talking and getting our voices out there a little bit more so you can talk about how EPR kind of is helping to do that. Yeah, I mean you know one of the main reasons why we do this show is building community. We want to make sure people are accessible and you do that in many different ways. You talk about people's work but you also talk about who they are making people accessible. Sometimes you don't do something that the person is working on but you like their personality and you want to know is there a way we can connect in other ways and we give them options. We talk about what their hobbies are. We like doing that because it's a showcase of who they are as a person and we always ask where can people get in touch with you. That's how we get those bold voices out. That's what we do. Yeah, we just we just like to open the door with the conversation. I always think that this I'm a career coach as well for environmental career seekers and it's daunting in

29:35this field because most people in this industry have some sort of degree. If they don't have a degree they at least have some body of work that's very impressive and so when you're coming to this as a new person it can be very intimidating and so we like to break people down into like who are you. What is the work that you're doing in more layman's terms. We don't have a very technical podcast. It's very easy to follow and casual. We leave the news to more hard hitting teams. So with one of the things that that did come up was you know we have this whole new continuously body of people coming into these industries and you know what are your thoughts especially as a career coach then professionals in the in the area of getting those people to understand all the different opportunities they have you know a lot of times career counselors and others are pushing them to do something very specific whenever they're coming out of high school. How do you help them understand where these starting degrees whether it be ecology or geography or whatever it is can take

30:40them to all these different places like we were hearing on the stage. Yeah that is a great question and it is definitely a challenge. The universities you know they their job is to get them the degrees and the basic knowledge that they need to do the work but it isn't necessarily their job to get them the job. So I think students have them misunderstood misconception that once I'm graduated then it makes me automatically qualified for anything I want to do which is not the case and so that's why it's so important early on to start getting involved and community then becomes important. Being brave and having conversations with people maybe you've never met LinkedIn has such a bad rap but it is an untapped resource for most people of where I could explore who's doing what and actually reach out to them. So I encourage people to be brave and listen to a podcast reach out to someone it's really really often a lot of work that you put into doing a podcast I'm sure you know this and getting very very little recognition or reward or anything so if someone

31:45listens and I know Nick appreciates this someone listens and they reach out to you and say I love do what you said about whatever show me a little flattery then ask me a correct question go for it you know. Yeah and honestly I would say to I think people people really do underestimate their value as well I mean we heard signing a talk about how she had a new person develop a map and in a creative way that she'd never seen before and when you come out of high school or college and you're doing your first job your aperture is wide open and that's a great thing it means you are seeing far more of an area than most people and as you get older in your career you kind of focus in on your expertise but there's value in both of those concepts it's not just one. Thank you for hosting a great episode today hopefully people will head over and listen to your your live show and thanks for talking me. Well thanks for talking to us. Thank you.

32:55Honda Dance Corner as always we encourage you to check out events around the world including over and boisey heading to Archon 2026 which is going to be 12th to the 15th of October. This conference focuses on reality capture so if you're in that space then this is somewhere you might want to check out next month. The Tolkien Conference Switzerland is going to be March 13th in Lussain Switzerland with the theme of maps and landscape in Tolkien's Middle Earth. That sounds amazing. Can I just say that although the fact that it's Tolkien Conference Switzerland which implies there's Tolkien Conference many other places. Yeah that's probably all right. Yeah of course if you'd like us to add your event to the podcast sent us an email to podcast at rayspacial.com. If you'd like to reach us individually I can be reached at suit at various spatial.com. I can be reached at barbedatvaryspacial.com. You can reach me at franketrayspacial.com contacting me at kind of spatially and can find all of our contact information over at rayspacial.com slash contacts. As always we're the folks in very spatial. Thanks for listening.

33:57I will see you in a couple weeks. Spend our time chasing nothing at all. I wasn't always chasing your desert where a free drop was left

35:03untasted. Even the last towns I gave you was wasted. If you're a cigarette and I'm a smoke that can't quit. I'm tired of being rolled over but for some reason I live with him. I know that you know that I should leave. But I can't come to terms with being the first one to the first one to fight before. Spend our time chasing nothing at all. I wasn't always chasing your desert where a free drop was left untasted.

36:17Even the last towns I gave you was wasted. Spend our time chasing nothing at all. I can't help but fight I'm the feeling of being used. It's not new to me and it's not new to you. I can't help but fight I'm the feeling of being used. It's not new to me and it's not new to you. Why do we fight before? Spend our time chasing nothing at all. I wasn't always chasing your desert where a free drop was left untasted.

37:25Even the last towns I gave you was wasted. If you're a cigarette and I'm a smoke that can't quit. Spend our time chasing nothing at all. I wasn't always chasing your desert where a free drop was left untasted. Even the last towns I gave you was wasted.