00:00.00 archpodnet Welcome back to the archeotech podcast episode one ninety four and we are talking about photographgrammetry being used in a and say a new way but definitely a more robust way than we've probably seen in the past at least in our the journal articles that we've read so. Paul let's get to the meat of this. Um, you really dove into some of their um data and methods and and things like that. What? what are they doing. 00:24.23 Paul Okay, well the meat of it. You're getting hungry too. Ah yeah, I'm starved. Okay, so basically they start out with a Dtm of the site and from that they create ah a solid three d model. 00:28.70 archpodnet Ah I'm absolutely hungry. Dinner's right around the corner. Ah. 00:39.50 Paul So they you know it's not just the surface but it's also extru to downward some arbitrary depth. So it goes down deeper than the um than the surrounding field I'm not entirely sure why they start with this actually because it doesn't reflect anything else. Ah or it's not obviously reflected in any other part of the article I. 00:39.65 archpodnet Um. 00:45.89 archpodnet Yeah. 00:57.42 Paul Think though, it's so that they can have you know a baseline that they can then essentially re-excavate the ah the units that they're digging and then they can use that for comparison. 01:06.20 archpodnet E. 01:12.86 Paul For comparisons between various set trenches between the different excavation units. They're excavating in a 5 by five meter grid like I said before with bulks between them. So it's almost ah you know the wheel or Kenyon Box um and then. Every time they're excavating very slowly and stratographically so they're not doing arbitrary depths though I think you could adapt this this system to um to an excavation strategy that uses arbitrary depths but they're really relying on the skill of their excavators to recognize the strata and the various contexts as they're going down. 01:41.26 archpodnet Um, go which is tough. 01:47.20 Paul So again, this is this is tough and you know kudos to them for being able to do that again that that figure showing the the complexity of all the context just in one layer um is is phenomenal. So. 01:50.22 archpodnet Yeah. 01:57.11 archpodnet Yeah I I have excavated stratographically before but it was nothing this complex right? Like we used we did that in the southeast where there where you go down a meter and a half and there's 3 strats right? and it's all sand so you know. 02:08.94 Paul Yeah, yeah. 02:14.46 archpodnet Not that big of a deal to be honest. Ah. 02:14.65 Paul ah ah yeah we excavate something like this in in Ed Lagosh um and with great results now that that everybody's kind of getting the hang of it and again this is I mentioned earlier but sadap piummenti this is the way that she insists. 02:34.90 archpodnet Um. 02:34.20 Paul That you know the trenches that are under her field directorship I don't know what they're her overview are are doug. Um, and it's It's really quite a skill and it's a skill that I personally don't have but to watch people that can do it. 02:51.98 archpodnet Yeah, yeah. 02:52.46 Paul And the kinds of detail that they can pull out is amazing. So they're starting with that. Um, once they get to a new stratum. They overlay a 1 by one meter string grid to help with the photography and the illustration and this is something I believe in the way they talk about this is something that they've been doing for a long time that they did it from. 03:04.51 archpodnet Oh. 03:11.82 Paul Before they were doing photographgrammetry on this project and that's something that we've seen before on other you know I've worked with string grids either you know nail them in you know into the ballks and draw across or even sometimes where you have a ah frame with the strings across it that you lay down over right. 03:19.71 archpodnet Sure. 03:26.80 archpodnet Yeah visos. Yeah. 03:31.74 Paul So that's something that people have been using for decades anyhow because of the method that they're doing ah, they're still doing that and so they they dig in their 5 by fives when they get to a new stratum. They overlay the string grid. They. Do all these photographs now they do extra photographs because they want to turn to photographgrammetry. They still do the illustration. You know me I'm big on people still illustrating. Um, but now they've got that that grid that they can use for aligning and scaling their. 04:03.27 archpodnet Right. 04:04.77 Paul Their Sfm they're structured for motion photogrammetry models right? So they they kind of stumbled into ah ground control points into gcps with this grid which is I think you know brilliant and fortuitous and and cool and at so many levels. Ah. 04:13.93 archpodnet Yeah. 04:22.73 Paul And so they're using that and they're using what I think they said they were using metahape they go into a lot of detail about the settings and about the 2.5 d and the sparse point cloud and all that stuff that frankly I don't care about um I don't think it's useful. Um. 04:35.20 archpodnet Yeah. 04:41.30 Paul Ah, because there's so many different ways of getting very good photometric models now you know it might have been useful a few years ago it almost certainly was important and useful when they started the project but right now it doesn't doesn't do much for me. Anyhow, they get that that model and then they move into blender of all places. 04:50.53 archpodnet Right. 05:00.23 Paul Ah, blender is typically conceived of as a a 3 d modeling and an animation program. But it's open source. It's extremely powerful. It's deep. It's hard to kind of get your brain around initially and once it clicks people love it. Ah. 05:11.72 archpodnet Um. 05:19.61 Paul So it's interesting to me and I think good in their whole workflow that that's where they're doing the majority of their work. In fact, they have a digression about doing some work that they were trying to do in Arcgis because you know hey it's in yeah gis and it's going to be better that way and you know all good or all the time and it turns out that. They've run into all sorts of problems with it. It doesn't render things crash things don't work so they go back to blender where they can do it pretty easily you know so and again back to the thing about the all the details they do explain all the little details of how they do things in blender. In this case I think it's more important though because this is where. 05:40.30 archpodnet Yeah. Nice, Nice. Yeah. 05:58.30 Paul You start to understand I start to understand how they are turning these photogometric models into a real 3 d space of all these contexts interlocking you know as if you were to take a chunk of soil out and have it. 06:12.13 archpodnet Ah. 06:15.51 Paul And that's something that you and I have discussed a lot in the past um about having documentation that could permit somebody in the future to reinterpret you know my master's paper was on reinterpreting some of Woolley's excavations at the royal sometier cemetery of ore. Um. 06:25.38 archpodnet Red. 06:33.90 Paul But when we talk about archeology as a science. One of the big things of science is reproducibility which we don't have if you're excavating right? You can't re-excavate something and we teach you know, archaeology 1 to 1 students that right away is that. Yeah, you'll never be able to re-excavate. So What do you have to do you have to document really? Well what they're doing here is documenting really well very organic shapes that have a chronology inherent to them and have human activities inherent to that Chronology. Ah and they're capturing that. Um, and so anyhow they get each level modeled and then they extrude it downwards so that it would intersect the next lower level. It doesn't have it can be any arbitrary depth just has to be deep enough to hit the next level down. 07:10.83 archpodnet Um. 07:21.27 archpodnet Which is cool that that's the part I think is is really neat because I haven't heard people talking about really doing that before. Yeah. 07:27.61 Paul Right? No and that's where and then it gets even smarter they're doing in blender. So then they take that photometric model from the next level down and they slice the bottom side of that top model with the next lower level down and they keep on doing that I think they said they had some like 80 different models. 07:37.31 archpodnet M. 07:46.49 Paul Ah, units oh no models I can't recall um of these things but then it gets more complex because that'd be fine if you just had a simple stratum that went from you know one balk to the other balk North Southeast west 07:47.62 archpodnet Um, yeah. 08:01.23 Paul And maybe it's a little bumpy or a little sloped or whatever and then you've got another stratum directly underneath. But what they have are like I said earlier tons of little pits and post holes and use areas and all sorts of stuff I'm not sure exactly how they're defining their strata because I can't quite see it in the um in their illustrations. 08:18.77 archpodnet Um, sure. Yeah yeah. 08:20.64 Paul Doesn't matter at this point I Trust what they're doing Yeah ah but then to get all those little context what they do is they remember I said they were doing illustrations they make shapefiles of those illustrations scaled and everything in yeah gis they import those shapefiles. 08:35.33 archpodnet Um. 08:40.38 Paul Um, with a plugin that didn't know existed for importing shapefiles into blender. Um, another open source project and they've got it linked in there in the article they import those shapefiles and then they slice those context out of the solids that they just made in blender. 08:54.90 archpodnet Um. 08:57.97 archpodnet Um, that's awesome. Yeah. 08:59.00 Paul With those shapefiles and they call it like a cookie cutter which was funny because when I I was reading I was like oh that's kind of like a cookie cutter and then they they say cookie cutter here. We are at food again. Ah. 09:14.86 Paul Ah, so 1 thing that's unclear to me is that these shapes are organic and if you have a cookie cutter you would end up with kind of cylindrical shapes right? or straight straight walled and that's not what they show in their final result they everywhere else they talk in lots of detail about how they do it. 09:22.40 archpodnet Yeah. 09:33.97 Paul And here they kind of gloss over it. Um, something about voxels and smoothing. But I don't know how you don't end up with either cylinders or something that looks like wedding cakes or upside down wedding cakes like you see. Um, you'd be familiar with this Chris like you see of um airspace right? yeah. 09:36.94 archpodnet Um. 09:50.12 archpodnet Yeah, yeah for sure. 09:52.76 Paul The upside down air ah wedding cake model of of different kind of classes of air space. Um I don't know how they get from that to something that that's nice and smooth that shows like the roundness of a pit. Um, yeah. 10:04.63 archpodnet Yeah I mean it sounds like there's a lot of interpolation going on as well. You know when they create these models and I guess it really just depends on. You know they're digging stratographically. But when you're coming down and you encounter a soil change right? You encounter something different. 10:17.55 Paul M. 10:20.40 archpodnet And then you start digging down around it. Maybe you leave that for a little while and then you know then you go into it or maybe they're they're excavating it everything down uniformly but then mapping as they go but the as they go part is the one that will determine that shape right? because what is the what is when do they stop and measure. 10:33.57 Paul Yeah. Um, I mean yeah so I do think looking again at that figure 4 the the ortho photo of it I think that each one of those little different you know, ah hundred different contexts that you see. 10:40.13 archpodnet That's what will help determine the shape. 10:50.42 archpodnet Um. 10:54.75 archpodnet Yeah. 10:56.40 Paul With different colors and textures on the right I think each one of those gets fully removed and once it's all fully removed down to some base layer. That's what they call a stratum I'm not entirely sure. Um, but that would mean that that that roundness of the um of the context the organic. 11:06.50 archpodnet Okay. 11:16.15 Paul Shapes rather than being you know these cylinders sliced through cookie cutter style would be reflected then captured in the photogrammetry. That's what I think is going on but I'm not sure and again, that's of all the places that they had more detail than I cared about. 11:25.80 archpodnet Ah. 11:34.46 Paul This is a one that really really I wanted to know and I don't have quite enough detail there and I think it's just because there's some Baseline assumptions about the way that they're excavating that I don't necessarily understand I Would you know have to have a conversation and have to be on site and see exactly what they're doing to to fully. 11:38.40 archpodnet Um, yeah. 11:48.91 archpodnet Yeah. For sure. Yeah, it makes sense. Okay, well I mean this is this is super cool and it seems like something that I mean obviously you need knowledge of how to put all this stuff together. But the the collection of the data. 11:54.33 Paul Get that. 12:11.50 archpodnet Is something that could be done by almost any archaeological team on the planet right now. Um, you're already doing the steps necessary for the most part to be able to get here. You just need the equipment to be able and the knowledge to be able to take the photographs do the right drawings take the right points and and that sort of thing. So. 12:13.97 Paul M. 12:29.70 archpodnet Just modifying your data collection methods or augmenting them a little bit and you could actually do this and it would it would not strike me as odd at all if 1 of these grad students or professors or whoever that is doing this you know ends up starting a company someday where they're just like. Here's the kit here's all the stuff you need that we need to make your models for you collect all the data send us back the equipment and we'll send you all the models when we're done right? but but people are still. We're still in the phase where people are still just like figuring this out as they go. They're reading articles like this or listening to podcasts they're they're figuring out. 12:51.20 Paul Um, yeah. 13:02.51 archpodnet Okay, well this one said they use this this one said they use this I think I'm going to try this and then this and they're like you said they're not really, they're they're kind of using a recipe but there is no recipe. It's just like here's what I want and it's been done a hundred different ways. So I'm hoping we get to the point soon where there's really a. 13:11.18 Paul M. 13:20.65 archpodnet Really is a ah solid way to do this I mean obviously there'd be some variation depending on you know, certain variables but there there really should be just a handful of ways to start From. You know, nothing and get to these models that we want and and make it replicatable. So Hopefully we can get there soon. But it's pretty tech-heavy. Pretty high knowledge you need like to get into blender and and even Arcgis and and ah ah, whatever the hell they call Shape. Ah Meta is it Meta shape I think I Always forget what is?? Yeah yeah, right, right? So anyway, it's promising. 13:49.33 Paul I Think they started with meta shape. But um, yeah, that's the current name of the the software. 13:57.89 archpodnet More stuff like this is what we need more people figuring this stuff out I Love it. 13:59.77 Paul Yes, and before we go I'm just going to um to point out that they grafted all this onto their existing workflow because they were excavating a certain way. Ah, they were documenting in a certain way. This became an add-on the the other add on to it a little bit of equipment a little bit of software. They give some prices they give some time that takes these things um a fair amount of Know-how and an individual with that Know-how that can then devote a lot of their time in the field to doing that. 14:16.91 archpodnet Um. Yeah. 14:32.78 Paul And that's something that Marco and I talked about on last episode. You know how much time he spends processing but also then managing his data right? And that's something else they mentioned in this article that also boofs it is that they they tell their their dmp their data management plan at the end of the article. Um. 14:34.12 archpodnet Yeah. 14:43.68 archpodnet Um. 14:52.53 Paul Toward the end of the article explain what they do how they name the files how they store them and that's something that matters to me because we're in the midst of grant writing right now for the logage project and that's a big part of what we have to document is what our dmp is and so far we. We've been very good about keeping lots of data but it's not structured. It's not organized in certain ways and we need to formalize that. Um, so for me, it was another little piece of oh here's how some people are doing it just like my interview with Marcos here's how 1 person who I know does good work is doing it. 15:14.29 archpodnet Right. 15:27.44 Paul You know, a lot of things that I can just kind of mull over again back to that watching the chef. It's you know it's watching Julia Child Jacques Pein whatever on their cooking show and maybe not following along step for step but getting a sense of you know what. The the triggers are you know what? what are they trying to get out of the heat or the oil or the salt or the whatever it is that they're working with at the moment for an end result and. Regardless of all the the finicky little details that might go into this article or go into that that recipe. It's the the overall how you get to something that is good and acceptable at the end that I think they they managed to thread nicely in the article and um. And yeah, so I am a little bit richer for having read it. 16:17.43 archpodnet Yeah, absolutely now I I love it and like I said it's it's really cool that you might be able to use some of this at at Logosh and you know at least their data management plan and you know take something from it and I hope other people can too. So. 16:31.14 Paul M. 16:34.97 archpodnet I don't think I have much more to say on this you Paul. 16:35.68 Paul Oh no I've said plenty about it. 16:40.65 archpodnet Ah, nice, nice. Well you know what? it's interesting that we had and this doesn't count for the drinking game. You people out there. It's interesting that we had an entire discussion about photogrammetry and didn't once say the word drone I'm just saying ah. 16:53.19 Paul Ah, though to their credit. They do mention drones in the article. Ah they do so they have to taste your drink. 16:56.44 archpodnet Right? right? They do? Yeah but again, that's that's a but exactly. But that's one that's another one of those things that I just love about where we're at right now in early Twenty Twenty three is that. 17:09.30 Paul Move. 17:11.61 archpodnet We didn't We didn't have to mention Drones. We don't have to mention how do you take Photographs. We don't have to mention those things. It's just yeah I Actually don't I didn't have a chance to read this fully and and into the really detailed sections of these but I would imagine they're using not only you know the aerial drone photography but Also. Ah, photographs of units close in and things like that to kind of bring all this together into a thing and and that's just that's just the standard suite of photographs that you take, but that's the thing is it's kind of standard now right? If you're going to do this. You know that that's another piece of equipment and skillset that you need to have and it'll be enable to be able to really get it done. 17:33.55 Paul Yep. 17:46.29 Paul Right? actually and that's where I'll just end my last comment is that kind of standard that this is the next little piece of the puzzle this is where it pushes Beyond just making your photometric model with Meta shape or drone deploy or. Open drone map or whatever you're using, Um, which a lot of people are doing now to everybody's credit and to everybody's benefit but pushing it beyond that to okay, let's try to actually reconstruct a site but. 18:12.79 archpodnet Um. 18:19.70 archpodnet Yeah, yeah, yeah for sure for sure So that that's another thing I Really like about it too is that reconstruction aspect of it because archeology is and especially ah excavation is inherently destructive. We are never going to see that site again in the same way. 18:26.80 Paul E. 18:36.45 archpodnet That it was that being said I might be able to take something like this at someday and just play with the entire thing in my oculus quest too right? Just jump into it move some some levels around see what it looked like originally you know, push everything back with my hands and just dive down a couple of levels and see what it looks like there. 18:48.20 Paul M. 18:54.25 archpodnet Pull this shape out pull this context out you know, just like you know iron man. Ah you know the Iron Map movie where he's manipulating all that stuff like with his hands just something like that and I see that we're we're getting really close to having the data to be able to do that. We may have the tools to do that yet. But we're creating the products that we're going to need. To be able to do that with so I love it all right? Well with that I think we'll say goodbye and we will see you guys in a couple weeks. 19:16.63 Paul Um, yeah, absolutely. 19:23.53 Paul Bye Chris.