Embedded - 443: Vexing Machines

Episode Date: February 17, 2023

Chris and Elecia talk about photons, comets, patterns, other flying objects, and cameras. Chris uses PixInsight for processing  and has an Ioptron Sky Tracker. Apologies to our southern hemisphere listeners because Polaris is not visible there. There are (of course) other ways to align and even in the northern hemisphere more modern trackers don’t necessarily need Polaris. Star Exterminator: who cares what it does it has an awesome name. Though it does what it says (on photos, no real stars were harmed in the making of this podcast). Jupyter Notebooks on a Circuit Python board. Elecia’s Yoshimura sine pattern generating Python colab. Also, Rigidly foldable origami gadgets and tessellations is an excellent article about Miura-ori and other rigidly foldable patterns. You can see her patterns over on Instagram. (You can see some of Chris’ photos on his Instagram.) Transcript

Hello and welcome to Embedded. This week it is just going to be Christopher and myself chatting with each other. Hello. Hello. Hello. You started by asking me if I was enjoying the every other week cadence. And I have to admit, I am, although the fact that we're talking to each other means that I managed to spend two weeks without inviting anybody.

So I did start some invitations today, so we'll get some more guests. Okay. Well, that'll be in two weeks. In the meantime, you've been doing a lot of astronomy because I hear there was some sort of celestial event. We're just going to dive right into topics. Just get to go. Oh, I'm sorry. How are you? I'm fine. I'm good. How's astronomy?

Astronomy is interesting. Do you want me to tell the whole story of astronomy from the beginning i mean doesn't that take like 12 billion years it's very quick you start with gravity and then and then uh there's the big but no uh yes i've been doing a little astronomy so that's a hobby i've had for a long time but i haven't kind of put it by the wayside for a while. I did a lot like around 2000 to 2004, something like that. And when I say astronomy, it's not so much observing stuff, it's taking pictures and doing astrophotography. And as you imagine, back then computers and cameras were a little bit different. So I think the camera I had back then was, it was pretty advanced.

It was a cooled CCD camera, which is what you're supposed to use for astronomy if you're being, you know, super, super cool about it. But it had a parallel port interface and stuff. And yeah, anyway. So there's a comet and... What's the comet's name? Comet's name is C slash 2022 ZTF parentheses E3.

I finally got that right after getting it wrong on social media a bunch. Why aren't we giving them good names anymore? I think it's because it's too hard to kind of not have a standard. So I think that sometimes they have good names and the boring names, but the boring name tells you when it was discovered and sort of who discovered it. So the 2022 is when it was discovered and ZTF is, I think,

Zwicky Transient Facility, which is a telescope facility that found it. So anyway, this is a comet that recently became visible and was somewhat, I wouldn't say spectacular, but a good visible target that for a while was good in binoculars and small telescopes, and now is pretty much invisible unless you've got a big telescope or a camera. But you took your SLR out. Yeah. So I've been fighting with thinking I was going to start astronomy again.

And you have a giant telescope. Yeah. So I have a giant telescope. Well, I have an eight-inch telescope. Physically giant. Sorry. I don't really know the telescopeness of it. It's a good telescope.

It's reasonably sized. I mean, there's many bigger ones for amateur astronomy,. It's a good telescope. It's reasonably sized. I mean, there's many bigger ones for amateur astronomy, but it's a good middle-sized telescope. But it's pretty heavy, and the mount for it, while it's a good mount, is extremely heavy. And so I think, all told, when I have counterweights and stuff, if I'm going to move it outside and use it, it's 75 pounds of stuff that I've got to move. And by the time I've got it in place, I just... And aligned. And aligned. I just want to go to bed and not touch anything. So the con was out there, and I was like, well, I've got this other stuff.

I have a little camera tracker, which is where you can attach a normal camera to. And all it does is rotate at the speed of the earth's rotation. So it keeps stars mostly in place. But if you don't align it perfectly, or there's errors in the gears and stuff, it'll bounce around some. So it's good for wide field stuff where the stars

aren't going to, you're not going to notice that kind of stuff. And the comet is a wide field target. So I just said, I'll just go out there and see what happens. I'll take my camera, put it on there. And I actually got some decent shots the first night and I was really like surprised. So I went out a couple more times and got a little better at it and got some good photos that I've posted in various places, Mastodon and mostly. Instagram. And I'll put at least one in the show notes. So it's got me more interested because it was easy. Like the mount and the tripod weighs maybe eight pounds maximum. And the camera is, you know, normal DSLR that weighs a pound. And so I can, you know, move stuff around and aligning it was easy. It was just easy.

The activation energy to get going on this. The friction, yeah. That's the trouble I have with hobbies is there's too much friction. I don't want to do them. So, yeah, I can go into the process a bit more if you want. I do, actually, because that's changed a lot in the last few decades. Sort of. I mean, it's all signal processing and averaging and trying to find good noise.

Has it gone in the direction of AI? A little bit. I can talk about that. But most of the processes are the same. It's just we have bigger computers. And bigger images. Well, yes. So some of the advantage of the bigger computers has been taken away by the bigger images.

So my old CCD camera I used to use, I think it was like... A super fancy CCD camera. I paid $1,500 for it. It was used. Yeah. I remember it was... It was one of the biggest purchases i ever made in my life at that point and it was 750 by 400 pixels wow just so many pixels and my new camera is

6 000 by i don't know 4 000 it's 35 32 megapixels it's a lot um and so you know you can do the math on that but the amount of processing that goes into those is a lot more. So, yeah. So the way this works, basically, I'll go through the pipeline a little bit. So in the old days with film cameras, you used to get high sensitivity film and you'd take a really long exposure, as long as you could, with your camera being stable on a mount, and developed that, and that was the picture. And there were different ways of developing it than with normal photos, but it was pretty much the same process. And there were different ways of light gathering,

which is part of the telescope. Yeah, yeah. With digital cameras, you can play a lot more games. You still want long... Checkers? Yes. Chess? Snake.

Okay. You still want really long exposures for a lot of objects, not for the moon or planets necessarily, but for deep sky objects or comets and things. You want really long exposures. There's not that many photons coming from these things. They're pretty faint. And while you can see them in telescopes,

they're often very faint, smudgy things that you can kind of see out of the corner of your eye if you're lucky. So you still want really long exposures, but you no longer have to take one long, big exposure. In fact, you generally don't want to because you'll saturate. If you go long enough, you start to saturate the sensor with bright stars and things. So what you do is you take an array of shorter exposures.

A series of shorter exposures. Yeah, and these can be anywhere from a second to a couple minutes to five minutes, depending on what you're doing or how stable your amount is and stuff. And once you've got that array of photos, and it can be like hundreds, you throw it into an astrophoto processing program that takes all those and aligns them because they won't be exactly aligned anymore because things move. And then once it's aligned them, it averages them all together. And that's the magic bit, because there's all sorts of sources of noise with taking pictures of anything or any sort of optical photometry kind of thing, right? So there's noise

from the sky. There's background kind of sky glow and light pollution. There's noise from the sensor itself from getting hot. So there's heat in the sensor, there's thermal noise that shows up. And there's noise when you read out the sensor for the whole, you know, digital signal processing, analog signal processing chain. And a lot of those are uncorrelated. They're just random. So if you average a whole bunch of images together, that all goes away. And so your signal-to-noise ratio goes way up. And for stuff like light pollution, you're, and I may get this wrong, but even your faint objects, the photons are all coming from them in a consistent way.

And the light pollution is maybe less consistent. And so you can start to drive, when you average them together, you can start to drive the base level noise down when you stretch the image out in contrast. Oh, I didn't get that. I mean, because the stacking of the images, it made sense to me in a dithering sort of way

where you take multiple too fast samples on an ADC. Sure. And then you average them together to get another extra half bit. That's another thing you can do called drizzle. But yeah, maybe I'm explaining that wrong. So it's more about the long exposure because it's more about the long exposure without saturating. So if you take a really long exposure, the sky glow might saturate. If you take a lot of short exposures, the sky glow won't. And so the longer you go, the sky glows the same in every image. The more sequence images you have, not the longer you go is ambiguous. two photolines for whatever. So that's always going to start to dominate. And then you can, in post-processing,

kind of contrast stretch out the background radiation as long as it's not saturating the sensor. So, and we'll fussy on some of that. So that's what you do. And then you do some, there's a lot of other stuff to kind of, when the images come out of a camera from photography, you usually don't see anything. Like if you just opened a picture from Photoshop, you might have one or two bright stars. And that's because when you're in a linear transfer function between the sensor data and your screen, the astrophoto images are in a very narrow band of values

for your sensor. So that needs to be stretched out so you can actually see it. So there's usually like an exponential or an S-curve or something you apply to take the very narrow dynamic range that you've actually captured and stretch it out so you can see it. It's kind of weird. I mean, relating back to ADCs, it makes a ton of sense. So yeah, there's a ton of software packages that do this stuff.

I use one called PixInsight, which is kind of the gold standard right now. It's sort of expensive. There's a lot of free ones out there that do a lot of similar stuff. Serol's a new one people like. S-I-R-I-L. There's Deep Sky Stacker and a few others. It's fun.

It's kind of like fishing in a way. You're just out there doing nothing and then eventually you might have caught a picture that is of something that everybody else has taken a picture of, but it's your picture. It's fun to look at the star charts and see what you accidentally picked up.

I was doing that with the comet photos a little bit. Oh, there's a little kind of smudgy nebula there in one spot, or there's some dark clouds of interstellar dust and things that block stars, and that's always kind of interesting to see if you expose long enough. Yeah, I was surprised to see those in your images. It looked like you had black clouds, and I was like, that doesn't really make sense. The sky should be full of stars, but it isn't, actually.

It's full of stars and dust. Yeah. So yeah, I've been enjoying it. It's hard, and there's stuff I don't understand. Like you said, things are advanced, so the software has some things I don't understand I need to kind of study up on. But now the post-processing is hard instead of the activation energy. Oh, because I can take out the little camera? Yeah, and so you can choose to spend more time on it later. Oh, yeah. It's much more comfortable to sit on a couch with a laptop than stand in the freezing cold.

Well, even with the bigger telescope, it was an effort to do that. It's like, where is the effort? Is it at the starting point or at the ending point? Yeah, yeah. I've been thinking a lot about effort and how to do things without making yourself do them. Yeah, I'm thinking about that too.

And the effort thing is right, because for this, it's like the sky moves and night is only happening sometimes and clouds happen sometimes. So just a kind of urgency, got to get it set up, got to get it exposing while there's a window while the star, the object I want to see is not blocked by the house or the trees or not pointing. So there's a lot of that stuff.

It's like, got to get it, got to get it. And then when you're post-processing, it's like, well, I can spend three days on this. Oh, you asked about AI. Yeah. I mean, technically, you could say, I already know what everything is going to look like, so I'll just sharp it up so it looks like the canonical image, which seems not good. I mean, you can always do that with everything right now,

but there are some AI programs. Well, except that there are still 1.5 billion people on Mars. I don't know, talk about that. It's 2.5. Yeah, well, I mean, for all image processing, there's a bunch of AI stuff now that does noise removal or sharpening or, you know, just general enhancement stuff. And those are becoming more popular and they work pretty well. For astronomy in particular, I've used a generic photo denoiser on them. So that, that, that works fine.

Uh, there's been some specific AI ones, um, for astronomy. Uh, I used one of them on the comet image when I was trying to play around. When you're doing comet images, there's a bunch of games you play where you want to enhance the comet a lot separately from the star's background and then recombine them. And I didn't get it to work. I was trying to bring out more detail in the comet doing that. But one way you do that is you remove the stars from the comet.

There's a bunch of ways that you can accomplish that. One is to do what's called plate solving, where the computer figures out exactly where you're looking. And since it has a star map, it just subtracts all your stars. It says, oh, you're in this right ascension map, it just subtracts all your stars. Okay. It says, oh, you're in this right ascension and declination. Here are all the stars I see. I know what they are.

They're gone. But there's also another program. That's kind of magic. Yeah. Another program I used, I actually tried a trial of it, which is AI, called Star Exterminator. And it's a neural network.

Star Eliminator? Star Exterminator? I hope it's Star Exterminator. That'sterminator? I hope it's Star Exterminator. That's so awesome. I think it's Star Exterminator. Anyway, you throw your stuff at that and it removes all the stars using a neural network and just leaves the comet. And then you can do your processing that way. So there's a bunch of stuff like that. And then there's specific astronomy denoisers and sharpeners.

But, you know, people, I don't think people realize how much processing goes into all astronomy images. Like, you can't take a picture of the sky and then get it out to your camera and show it to somebody. Even if you contrast stretch it as your only thing, it'll look nice sometimes. I mean, you can do that for the moon, but that's about it. Yeah, and you can do it for, I mean, you can take a long exposure and just contrast stretch it. And, you know, most, a lot of the time, that's what film used to do, right? But it's a lot harder to do that and get a satisfying result, especially if you, you know, need to expose 30 total minutes or something. And even stuff like Hubble and the James Webb Telescope, there's tons of processing that goes into,

post-processing that goes into those kinds of images. So it's a lot of work. And so there's always a little bit of an artificialness to the image you're producing. But I think kind of my goal and other people's goals is just not add data that doesn't exist. So if there's values coming from the sensor that you want to enhance or bring up and others that you want to move down, I think that's totally fine because you're trying to see details that are hard to see.

How much of the processing pipeline do you not understand? Of the one I used? Yeah. There's, so, I'm trying to think the pipeline I use. So obviously get the camera images and there's a format conversion. So I kind of know that. There's the registration process where it aligns them. I think I understood that at one time. I'm not sure I do anymore because I think it's a lot smarter.

Back in the old days, you used to take like a master image and then you'd mark several stars by hand. Fiducial stars. Yeah, that you wanted to align to. And then it would do a bunch of stuff and try to pick the high quality images based on some focus factors and things. I kind of understood that. And then I guess I probably used cross correlation and other things to kind of do that. Now you just throw a pile of images at this and say, please go, thank you. And so I don't know how that works exactly.

That's a lot smarter and nicer because you don't have to go mark stars or throw out images. It just does all that stuff. You know, you can parameterize it all and change stuff to like, oh, here's what I want you to use for quality factor and stuff. But for the most part, the defaults just work. So unless you've got a funny image. So I kind of understand that, but not really. And then there's just some other things now that there's dynamic. I saw this stuff.

Like when you take an image of the sky, there's often a gradient, either because your lens is vignetted, where that means the dark corners of your lens might be a little, the corners of your lens might be a little darker than the center. And that happens with all lenses. And most people don't care when they're taking daylight images or brightly lit images, but when you're taking faint images, that tends to show up.

And there's also gradients in the sky from light pollution or just all kinds of things. And so you want to remove the gradient. Otherwise, you've got kind of a bright glow in the middle of your image, and then dark corners. And so there's stuff that analyzes that and just divides it out. So I guess I kind of could come up with how that works.

But there's other things like automated color adjustments and things. I guess they're mostly arithmetic, but I don't really know how they work. But yeah. If you're looking to get into this hobby, you can get into it for a lot of money or a ton of money. Nah. It's actually, with a DSLR and a Star Tracker, you can do an awful lot, which is kind of exciting. It is, because a lot of people have the

yeah dlsrs and and it doesn't have to be your bigger one it could be the older version yeah no no a lot i mean any modern where modern is since 2010 2008 uh cameras got a decent enough sensor for this stuff and like i said a lot of it comes down to just taking a lot more images and averaging them. So if you've got a less good sensor, okay, take more images. So you were doing this in our front yard, and people do drive by on the street. Yeah. Did the program just kind of drop those? No, I didn't see that affected much.

I'm not sure. I would have to look through the images and see. Probably I should have rejected more images than I did, but I wasn't really thinking about it. And the images you got had way more stars than I could see. Although, realistically, I don't know that there are seven Pleiades, so how many I can see is not really the question. Because you were stacking and getting the fainter stars.

Yeah. How dark does the sky need to be for someone to do this? Not dark at all. Because the software can take care of some of the sky. It's just exposure time kind of heals all sins of the sky. The photons are still getting to you from everything, from the faintest objects. They're still getting to you. They're not blocked.

It's just that there's other photons competing with them. And as long as you don't saturate, you can get both. Yeah. So I have seen some of the most spectacular, ridiculous, Hubble-like deep sky images from people, amateur astronomers who are in Chicago. But to get the spinny thing, the making sure the stars don't go in a circle thing.

Yeah, the star tracker. The star tracker. Yeah. You do need to be able to find Polaris. At least you'll find Polaris, but it's a pretty bright star. And the way you align usually is through a small telescope that's embedded in the star tracker And the way you align usually is through a small telescope that's embedded in the star tracker or the mount.

And that'll help you, even if it's too faint and you get it in the ballpark, you'll be able to see it in the... If you can't see Polaris, then yes, probably, you probably shouldn't be doing any astronomy. But it might also be day. It might be daytime. In which case, you can look at the sun.

With a filter, a proper... Never case, you can look at the sun. With a, no, with a filter, a proper, never mind. Don't look at the sun. Do not look at the sun. Yeah, no, there's a scale called the Bortle scale of light pollution. It goes from one to, I think. Yeah, I remember you were telling me about the Wortle scale. Yes, the Wortle scale. I was doing the Wortle, and he was telling me about the Bortle scale, and I did not understand.

The Bortle scale goes from like one to nine, I think, or maybe one to eight. And it's just a measure of how light-polluted your sky is, and there's kind of a rubric to kind of evaluate. So Bortle 1 is the least light-polluted sky, and that's where you'd be like in the middle of the Atlantic or something. And you said you can see shadows from starlight. From like the center of the galaxy casts a shadow. That's how dark it is. I've never been someplace that dark. Jupiter and Venus are bright enough to ruin your dark adaptation, night vision.

So that's how dark that is. It's very few people have encountered that. But then it goes down. Two is probably, you know, mountains, remote desert. Three, similar kind of thing. And then three is like rural. And then four is a suburban rural transition. And we're kind of in four.

I believe that. Based all the maps. If people didn't drive by our house. I think we're in four if we just go a mile one direction away from the neighborhood. Yeah. And, you know, most suburban people are probably somewhere between five and six. I think in San Jose we were in six or seven. But you got some good shots there.

And I got some good shots in San Jose because it just works. It's really scary. Yeah, you sit there like, with that camera, I could get stuff out, take a longer exposure and immediately see something. And it was just weird to point at an area of the sky where I couldn't see anything in the telescope. Like, I could see some brighter reference stars to know where I was on my star chart, and the object would be invisible and just sit there and take exposures and have this invisible object appear.

It was very cool. But there's all sorts of new stuff. Like they have, you know, this whole principle of averaging stuff, people have built it into all-in-one kind of telescopes for observing. So you can go out with this telescope and point it and just look on your iPad or iPhone, and it'll kind of accumulate while you watch so you can see things and observe that way. So they're, they're coming out with kind of

neat, neat new products that are, you know, more for just fun using, using this kind of imaging stuff and for serious astrophotography. So I recently built a pinhole camera. Can I use that for astrophotography? Maybe. You know what you could use it for? If you ever see where somebody exposes for an entire year the sky and so they get this figure eight

of the sun, all the sun's positions. I don't know if that would work with a pinhole. I don't know if that's multiple exposures or not. I don't know if that would work with a pinhole. I don't know if that's multiple exposures or not. I don't know what would happen. I mean, you'd have to expose for a long time, which means you might get star trails.

It would be interesting. I think, yeah, it depends on the film, too. If the film isn't very fast fast it might be so faint that you don't get much so in paper engineering class i made a camera a pinhole camera and the instructor sent us some special paper that is very light blocking and And of course, she helped us through the instructions of making the camera. And she also sent some film and instructions for making a developer at home using caffeine and baking soda. Don't believe that for a second. But I haven't tried the pinhole camera yet.

For me, it was more about the paper. And one of the special things about the pinhole camera is the focus is general. There is no... It has a very wide depth of field. Wide depth of field, that's what I said. Yeah. So I should do something that uses that, but since he's been doing astrophotography,

I keep wanting to take the pinhole out there and leave it on the roof for a couple hours. I mean, we could try it. It might blow away. That's the thing is, you've got to make sure it doesn't move. I've got to make sure it doesn't get wet. That too. It's made of paper.

Yeah, so I'm not going to become a pinhole photographer or one of those people who takes pictures using the pinhole by opening their mouth. I don't know. I don't know. I don't think that was a good idea, but there were two different artists. I know. I think they put the camera in their mouth. They used their lips as the pinhole. Maybe that one did, but I saw another one where somebody put a pinhole camera in there

because they could see the back of their teeth, so it couldn't have been the... Oh, yeah, no. This was, you couldn't see the back of their teeth. So it couldn't have been the... Oh, yeah, no. This was, you couldn't see the back of their teeth. You could just see who they were. I don't know. Anyway. So how is paper engineering? It's good.

I really, really like not being in charge. I mean, I really like... Of the class. Of class, yes. I like that it's a cohort. The other students are really interesting and have different perspectives. And I know with my class, people have said that they were sad when the class ended because suddenly their Saturdays didn't have that to look forward to. And I think that's going to be true of this. I kind of wish I could just

continue the class, even if the instructor didn't do it. I feel like I have so much more to learn from the other people. I was amused that one of the people in the class who is a children's book author, pretty successful, started to have a cohort-based class herself, in part because she took this one and enjoyed it and didn't want to go anywhere to teach the class. And so she's having her own children's book writing class, which filled up in a very short amount of time. So, I mean, I'm really enjoying that part. And the paper engineering is interesting. We're getting into pop-ups, which isn't something that is super cool to me. I mean, I think the books that have them are super cool, but it's not something I like to build.

It's interesting what things you can learn from a book versus the things that you need more people around. I actually picked up a pop-up book today because I was interested, but I could do all of the exercises in the book and I could probably actually do more pop-ups than I can do after week nine in a 10-week class, but I wouldn't have learned more about Bruno Minari and pinhole cameras and Canadian children's authors and Italian bookmakers. There's so much more, even though I would generally rather read a book. But one of the things from there that I wanted to kind of mention is the idea of rigid, flat, foldable structures.

This is one of the areas that is a crossover between origami and space. And whenever they talk about, you should see this giant cool space telescope, it's made of origami, what they're talking about are the patterns that can be folded so that they end up totally flat, kind of aggregating the material in a different form and then being able to spread it out into a flat structure like solar panels. One of the things we did in my paper engineering class was to go over these as tessellations, which I've done a lot of tessellations.

I was not excited about that. But once we started to talk about the rigid flat foldable, I ended up down the rat hole and got very into the different types of that. Miuri-ori folds are probably the most well-known. But Yoshimura folds is one that I have kind of fallen in love with. They tend to make arches. So, okay, I'm going to do origami on the podcast. I didn't bring down a piece of paper. I was tempted, but... It's all right.

I just got done talking about photography on the podcast. You take a piece of paper, a square, and you fold a diagonal. And then you fold the two corners into the center. You make a diagonal in the other way, fold the two corners into the center. And now you fold one direction. So those are all mountain folds, say. So they're all pointing one direction.

Turn it over, fold mountain folds in the other direction so that you have eight lines, which is in half, in half, in half. Okay. And now if you just make all of those folds. Like a pizza. Not like a pizza. Then in half and half and half. Eight lines. No, no, no, sorry, eight horizontal lines. Oh, see? I was thinking radio.

No, no, no, horizontal lines. And so now you have two sets of diagonal lines and one set of horizontal lines that's eight horizontal lines and four diagonals in each direction. So four, four and eight, so 12 lines. And now you just fold these so the diagonals go one way and the straight line goes the other or the horizontal goes the other direction and it will fold flat. At first it won't. It'll kind of make this little arch, and then it will fold flat into an arch shape. And so if you wanted to, I don't know, make this

out of large metal and store a bunch of shelters, you can store them as a small pack and then setting them up is pretty easy. Yoshi Morphold. I like them. So I made a little program that would let me try different angles. The one I described would be 45, 45 angles. But you can do them at different angles and get different arches and different shapes, blah, blah, blah. But then I really like to do curved origami. So I added sine waves, which led to some pretty weird and cool things. And to be fair, I saw somebody else had done this

and I wanted to try it. And I had saw someone else else, Katerina Lukasheva, did something that I thought might be a result of Yoshimura folds and sine waves. So I did a bunch of those and they turned out really good. Weird patterns, but still very comfortable patternish sorts of things. And then I added a little bit of randomness with how long each one was and where it started. And that was pretty cool but weird. And then I started adding random start points for the sign. And that's where things went really strange

because it's almost disconcerting to look at. Like, there is a pattern there because there are sine waves. Ah. But the pattern has been randomized in a way that I find... What you should do is you should take a signal for something and you should Fourier transform it and pick the big peaks and use those as the sine wave frequencies. And so you could have an origami that has a secret message

but it's broken down into... Okay, well, I'll put that on my list. So as I'm doing this, I'm making more and more Python code to do it. And I do it on a collab because I don't mind sharing these things. And that's the easiest way for me to do it. And then I output SVGs and they go to my Cricut.

But I've heard recently that SVGs using points are bad for printers like that. And I need to switch to Bezier's. It depends on if the printer supports the interpolation, I think. I can remember not all 3D printers could do that, I thought. But I think the Bezier's are good for everybody. And the points are kind of bad for most things. So I have that to do. I have more randomness. I have more folding.

I've been pretty happy with the origami lately. It's weird that I now probably should be more entertained by the pop-ups, but I'm still back in week two of the paper engineering class of playing with these tessellations, although rounded instead

of straight. And there's a book that I want from Robert Lang. He was on the show. Twists, Tiling, and Tessellations. I mean, I want, I have, I want to read. Ah, yeah. But it's full of mathematical proofs. Oh, no.

And they don't really get to this sort of tiling and tessellation thing until chapter two and chapter one was boring and math heavy. So I just need to get to a mental place where I can do that. And that's astronomy and origami for those of you who, sorry. Okay, let's see. What else? Let's see. else let's see next on my list is

burnout socks yes burnout socks how are you doing with that I don't know I'm too busy to be burned out I think I'm still burned out but I've added so many activities that I'm ignoring it

so like I'm taking another music class the previous music class was great, but it was really intense. This music class is less intense, but also takes some time and effort. And it's a totally different area for you. And, you know, I did the astronomy stuff, and I'm doing some more session work for music. People ask you to do drums or bass. Yeah, so that's coming along.

So I feel like I'm, like I've had some very manic days in the last couple of weeks where I've been like, oh my God, I've done like eight things. What am I, you know, exhausted, which has been great. I enjoy that.

But I'm still not really feeling like I want to work. So I don't know. I mean, we've talked about burnout a lot. And I don't think I'm at the place where I just want to sit around. I have a lot of stuff I'm excited to do. It's just none of it is work. So, I don't know what that is. That may not be burnout.

That may just be I need to retire. Because truthfully, you know, I sit down and spend three hours poking in PixInsight on, you know, and be perfectly happy to sit at my computer doing that. But if somebody paid me to do it, I probably wouldn't want to do it. So I don't know what that is. That's weird. It's usually me who can't read a book as soon as somebody says you should read this I don't want to process anything sold, bought or

buy anything processed so yeah I mean the work I am doing is kind of interesting so I'm having a little more fun with that but that's wrapping up soon so we'll see I am enjoying the music stuff I wish I could figure out a way to make that a thing But that's wrapping up soon. So we'll see. Yeah, I am enjoying the music stuff.

I wish I could figure out a way to make that a thing. But I think it's just going to have to be an uncompensated vocation for a long time. But you are still in burnout zone. I am. But you do all this paper engineering stuff and you sit and focus on your class so you can do things. I do, but I'm really tired afterwards. Yeah. And I can only do it for a little while.

And I do have fun with it. And what I'm working on for work is interesting and almost done and probably pretty good, but... Well, that work was not that was it was i totally misestimated it anytime you're looking at something you say the hardest part about of this is going to be how to test it you should not estimate it without thinking oh my god i'm going to have to write a full simulator. Because if the hardest part about it

is testing it, that means that's the part you're going to have to do. That's the actual work. Well, and the work, I mean, I've watched you do some of it and, you know, tried to help you with a couple of things that I probably didn't have answers for, but

it didn't seem exciting. It seemed like just a little couple steps above kind of scut work. So I don't know if that factors into it, but I think if you were working on an image scanner that was going to fly on the Hubble Space Telescope, you might be a little more into it. Maybe. But I only get a couple of really good brain hours

a day. And then I have to do something else because my brain just says no. Oh, okay. I'm still very tired a lot of time like that, but if I switch to something else, I can continue. No, I I'm still very tired a lot of time like that, but if I switch to something else, I can continue.

No, I'm still in the stage of, at some point, on almost every day, even on the weekend, it just is like, no, no, there'll be no more output. We were talking about this in the paper engineering class about needing some forms of input before you can have output. And one person said, it's like juice. And I'm a juice machine. I can output juice, but at some point, if I don't have oranges, you don't have juice. And

my brain is a juice machine seemed very yes, like that. So you're winding down your project. I'm winding down my project. Are we asking people to give us more projects? No. No. Not right now. Not quite yet. I would very much like to have a few months off, which would be the first actual few months off I can remember since...

Grad school? Yeah, probably. So I fully intend to take some of the summer. It's February. Some of the, you know, the summer in its penumbra. Huh. Well, I too am planning on taking a little bit of time off. But probably not that much. We'll see.

I don't know. Maybe I'll just take spring off or whatever. I don't know. We'll figure. Whatever season. It doesn't matter. Once we start needing to mine salt again, we will start mining salt. Yeah. Anyway. Um, but yeah, yeah, it, yeah. It's not like I want to do something for myself.

Like I don't have any ideas for a product and I don't want to do that stuff. I don't want to do that. You know, it's kind of like marketing music. I don't want to do that stuff. I don't want to do that. You know, it's kind of like marketing music. I don't want to do that. I certainly don't want to figure out how to sell something and manufacture it, even if it's software. So I don't know. I need a break. But not a break where I don't do anything, just a break where I don't do what people tell me to do. But one of the things you have been doing, Fork, is some actual CAD and 3D printing. Oh, God. So many people are going to be like, oh, yeah, I do that all the time. Oh, we hug. But then there are other people like me who are like,

yeah, I don't know how to do that, and I heard it's hard. I complained on previous episodes, recent previous episodes about all this. Yeah. I'm doing some drone stuff. And I have a big drone over here. You can look at it. It looks like an Imperial probe droid. Oh my god. That's enormous.

You haven't seen it? Did you not see it? I don't think I saw it outside the garage and it looked smaller in the garage. Well, that's with all of its blades folded down too. Those arms come out. So those two foot long arms. So when it's fully extended, it's quite a terrifying thing. The blades folded down, too. Those arms come out, so those two-foot-long arms. So it's fully extended.

It's quite a terrifying thing. If you told me that was named Star Exterminator, I totally would believe you. It does look very imperial. Yes. And it has an 11-pound payload capacity, so it could carry an infant if you wanted to have an infant delivery service. Stork. Keep it away from the blades. So what I'm doing with that is for a client,

and without going into great details, we're doing some stuff with neural networks, and I needed to mount an NVIDIA Xavier to it that runs our code, and I delayed doing this for months, trying to figure out how to do this, because I knew it was going to come down to me CAD designing something, and I'm terrible at it. I tried to CAD some stuff for other parts of this project and failed miserably, getting to the 95% point, but the 5% point where the screws align or where the board fits perfectly. It wasn't working, and I gave up. But this one, I had to figure it out.

So, yeah, I came up with a mounting mechanism that will securely mount this. It wasn't that bad, because it was a simple structure. It was basically a plate with some curved places for rods to sit and then screw holes. So that was within my capabilities, it turns out. And it looks good. It looks solid. curved places for rods to sit and then screw holes. So that was within my capabilities, it turns out. And it looks good.

It looks solid. But that's a pure extrusion. It was easy because I made the profile and then just extruded it and drilled some holes. It's pretty easy in CAD programs. So since that was the extent of it, it was within my capabilities. But boxes and things with lids and stuff like that and places to mount boards, that gets tricky. So that's still beyond me. Yeah. So I was pretty happy with that. But this drone is, oh my God, it's a really expensive professional drone and it is complicated. Really? What a shock.

It's got like this flight controller computer. It's got an Android tablet that it comes with. That's that thing there that runs this weird software. And it's got this flight manager and mission planner stuff and all this calibration and setup. And we got a gimbal for it that we're going to mount the camera to. And that took me like three hours to assemble this morning because it was fiddly with little Allen keys, and the entirety of the documentation was mostly that sheet of paper

right there on the floor. But it also comes with 50 cables, and I don't know which cables to use because some of it has to plug into the flight controller, some of it has to plug into power, some of it plugs into itself. And some of the cables, it turns out, it just gives you all the cables you might need, depending on what kind of drone you're hooking it to. And then there was one that was like for Bluetooth,

but I don't see any Bluetooth. Anyway, complicated. And then once you get that, you have to configure it in the Android thing and get it all to work. And this is not stuff I'm enjoying that much. It's a lot of fiddly, some assembly required stuff. And then this thing's got to fly

and work and not crash. Oh, and then I've got to get the power to work. I didn't even talk about the power. I've got to power the Xavier, right? The Xavier is this little NVIDIA computer. Didn't you decide you were powering it separate from it? Maybe, but it's got a barrel jack that takes 19 volts at like 5 amps or something,

maybe 4 amps nominal. And so it got it powered from the drone because the drone's in the sky. I'm not going to have a wall wart flying in the sky with an extension cord. So I found some DC-to-DC converters at Amazon that were in the right voltage range. And I tested that and hooked them up, and there's an aux port from the batteries on those drones. So that all worked, but I don't really know if it's going to work in flight because I don't know if the voltage is going to wiggle around a lot

while the throttle's being jammed up. So I don't know. So I've got to figure out how to make that stable or test it on the ground. Because I can't really test this thing. You've seen it. I cannot take this in my backyard and fly it without chopping down half a forest. And I can't. There's nowhere around here I can take it.

And I think you have to be FAA certified anyway to fly something like this. So this is not a 250-gram drone. So I thought about, you know, strapping it down and throttling it up, just to see what the power thing does. But yeah, that's my main worry, is conditioning the power so the DC converter doesn't

trip out. Why don't you just pop some more batteries in there? Because I'm not sure I have batteries that are at the right voltage for the DC to DC converters I have. Because I'm not sure I have batteries that are at the right voltage for the DC-DTC converters I have. Because I do not have the 24-volt batteries that this monstrosity takes. I have normal drone batteries, which are like

9.6. So, instead of a buck, I'd need a bunch of boost. I got a bunch of buck converters. That's my story. I will be glad to have this out of my room where it stares at me. It's a little beady GPS antenna.

Oh, and it's so loud. You turn the thing on and it beeps at like 130 decibels until you push the big safety button, which you have to hold down, which is between the blades. That's so bad.

Okay, look, look, look. See those two yellow connectors right there? Yeah, those. The safety button's to the left of that. And so those two blade arms, you've got to unfold before you do any of this. And so you're sticking your hand between vinyl record-sized helicopter blades to push off the safety button. It's just a little terrifying. Well, I think, why don't you take the blades off?

You wouldn't want to push the safety with the blades on, would you? It's extremely hard to take the blades off. You have to Allen key them off, and you wouldn't want to do that while it was put them back on with the safety off. You can't get the blades off. You have to allen key them off, and you wouldn't want to do that while it was put them back on with the safety off. You can't get the blades off. So yeah, it's a very vexing machine. I'm vexed by it.

And once you turn the safety off, it beeps in different ways. Oh, and then the Android computer starts yelling at you in this autopilot, you're about to crash kind of voice, and says things like, GPS not connected. No, satellite's found.

Let's see. Let's talk about something more positive. That was positive. Was it? You made her sound very mean. Well, she was mean. You see, CircuitPython

now has a Jupyter Notebooks host. It has for a while. I didn't know this. It came across in the Slack. MicroPython does. But it turns out CircuitPython also does. So if you want to use Jupyter Notebooks and have the Python

code in the notebook run on your CircuitPython target, that works. There's a link we'll put in the show notes, but you can set up a Jupyter notebook and connect it to your CircuitPython compatible board. And, you know, it runs each cell just like it was running on the, it is running on the device. So you can do all your embedded stuff in development. I thought it was really cool. Because that's, you know, normally the pattern with CircuitPython is you have an editor on your computer and you edit the code and then you save it to the device over USB and then run it.

So the integration time is pretty fast. But this is like, if you want to develop stuff, it's like running in the REPL, basically. But you've still got the code that you can copy and paste. So I thought it was really cool. I really like Jupyter Notebooks. I don't think it's that hard to run. So one of the things that I didn't understand about Jupyter Notebooks to start with

was that it always needs a server. You can't just run Jupyter Notebooks on your computer. to start with was that it always needs a server. Yeah, yeah. You can't just run Jupyter notebooks on your computer unless you have something for it to talk to, like a Python-speaking backend that is a Jupyter notebook backend. This is hidden from you when you do Colab because it looks just like Jupyter Notebook, but you're already attached to your fake machine that somebody else owns.

But Jupyter Notebooks is really nice because you can have the code and documentation in pretty documentation form. Yeah. And so it's really easy to say, and now I'm doing this without getting too bogged down in how pretty your comments are because they look like text comments. They look like word comments. Yeah. So, yeah. Yeah, no, I shouldn't have been surprised that this works, but I never considered it.

Yeah, it's a nice way to do things. It would be a completely different way of doing embedded development, which would be fun to explore. We got a couple of good links recently from our Patreon Slack group. And in fact, that's what the newsletter was last week

since we didn't have a show to talk about. So if you want to sign up for the newsletter, you can get some of the good Slack links from Patreon folks, or you could join the Slack group if you give us a buck on Patreon. I think we should raise that for inflation. It should be like $1.20 now, probably. I'm kidding.

I knew that you were kidding. I just wasn't sure what I was going to say to it. So, cool. All right. All right. Yeah, we don't have to talk about the other thing. Yeah. So, that's the state of the world.

That is the state of the world, in our universe at least. If anybody has a source for really flexible USB 3 cables, drop me a line. Because for the drone, the camera we're using is USB 3, but it has to be on the gimbal. And all the USB 3 cables I have and I've seen are fairly stiff. And so that's not going to work very well because it's going to be fighting the gimbal or the gimbal is going to be fighting the cable. So what I want is a really thin or pliable USB-C cable or USB 3 cable. Not USB-C, it's normal. Why didn't the gimbal come with the right cable?

Gimbal doesn't know what camera you have. You said it came with every cable. Yeah, but most... Sorry. Oh, your life is so hard. And we can't order stuff. Like, everything still has huge lead times with this drone stuff. It's crazy.

It's funny. The chip market is starting to move or has been starting to move. Things aren't nearly as bad as they were. The gimbal took months to get. The camera has taken many weeks to get. The drone itself took like six months. That was incredible.

I don't know. Does the drone have a name? It is the Aurelia X6 Standard. No, no, no. I mean, have you named it? Do you need googly eyes for it? I don't need a connection with it. You're not going to bond with the drone?

No, I do not want to bond with the drone. I can give it really evil... And luckily, I am not going to fly the drone when we do testing. So if it crashes, it's not my fault. It may not be your fault, but you'll probably still be present. You think I should wear a helmet when we do testing. So if it crashes, it's not my fault. It may not be your fault, but you'll probably still be present. You think I should wear a helmet when we go? A hard hat.

A hard hat? Yeah. And that's the show. Thank you to Christopher for producing and co-hosting. Thank you for listening. You can always contact us at embedded.fm or hit the contact link on embedded.fm.

And if you take all of our shows and average them together, there's enough exposure time that you get a decent show out of the 400. Let's see. Where were we in Winnie the Pooh, let's see, Heffalump had eaten most of the honey. Actually, Winnie the Pooh had eaten most of the honey, but had forgotten. Okay, and then he went to go look. Very little, okay, okay. Father, said Pooh, as he got his notice inside the jar. A heffalump has been eating it. And then he thought a little and said, Oh, no, I did. I forgot. Indeed, he had eaten most of it, but there was a little left at the very bottom of the jar, and he pushed his head right in

and began to lick. By and by, Piglet woke up. As soon as he woke, he said to himself, oh. Then he said bravely, yes. And then he said even more bravely, quite so. But he didn't feel very brave for the word which was really jiggering about in his brain was heffalumps. What was a heffalump like? Was it fierce? Did it come when you whistled? And how did it come? Was it fond of pigs at all? If it was fond of pigs, did it make any difference what sort of pig? Supposing it was fierce with pigs, would it make any difference if the pig had a grandfather called Trespasser's William? He didn't know the answer to any of these questions, and he was going to see his first heffalump in about an hour from now. Of course, Pooh would

be with him, and it was much more friendly with two. But suppose heffalumps were very fierce with pigs and bears. Wouldn't it be better that he had a headache and couldn't go to six pine trees this morning? But then suppose it was a very fine day and there was no half a lump in the trap. Here he would be in bed all morning simply wasting his time for nothing. What should he do? And then he had a clever idea. He would go up very quietly to the six pine trees now, peep very cautiously into the trap, and see if there was a heffalump there. And if there was, he would go back to bed, and if there wasn't, he wouldn't. So off he went. At first he thought that there wouldn't be a heffalump in the trap, and then he thought that there would. And as he got nearer, he was sure that they were,

because he could hear it heffalumping around like anything. Oh dear, oh dear, oh dear, said Piglet to himself, and he wanted to run away, but somehow, having gotten so near, he felt he must just see what a heffalump was like, so he crept to the side of the trap and looked in, and all the time Winnie the Pooh had been trying to get the honey jar off his head, the more he shook it, the more tightly it stuck. Bother, he said inside the jar, and oh help, and mostly ow. And he tried bumping it against things, but as he couldn't see what he was bumping it against, it didn't help him. And he tried to climb out of the trap, but as he could see nothing but the jar,

and not much of that, he couldn't find his way. So at last he lifted up his head, jar and all, and made a loud, roaring noise of sadness and despair. And it was at that moment that Piglet looked down.