It’s february. It’s cold out. I should start thinking about working on inside projects.
C3pr has gotten a lot of traction. I’m drawn to working on it just like I was for spp, and that turned out OK. I’m pretty sure that this month I’m going to make a lot of progress on the new camera component. And I think I ought to ‘do’ cabinets and the fan.
Feb 1 sun
Super bowl XXX1X. I can’t believe I ate the whole thing. I can’t believe SHE did too!
Feb 2 mon
I was going to go to Burlington today to pick up the beam, but it was snowing 8-12″ and it dawned on me that it wasn’t a good idea. I tried to upload pics to my blog site, but they wouldn’t go, and the web says it’s probably goDaddy throttling the site, so I may change it. (I finally got it re-hosted in June) I ordered a carburetor kit for the motorcycle and did a lot of opencv. I’ve learned a lot about opencv lately, and I’m sure glad I looked.
feb 3 tue
it’s cold, but it’s not snowing, so I prepared to go to Burlington to get the steel beam for the green house. I wanted to dig out the truck using the tractor, but the cold was messing with the diesel, and the tractor wouldn’t idle, and threatened to stall. Besides, the garage door was frozen shut. So I dug out the truck by hand. Salt on the roads wasn’t melting the snow, so the roads were dicy, but I just drove slow and had no trouble. I went to the vet, got gas, best buy for a USB cable, pet food, vinegar, Healthy Living, and QCS, where they said they didn’t have the steel for the beam. Shit. I hit Starbucks and came home. I read about stereo imaging in the opencv book until Mary got home. We had hamburgers and finally managed to watch Arbitrage on Netflix. Kind of a wasted day.
feb 7 sat
It’s still cold out, only now it’s snowing. Again.
I got the beam. I got it off the truck even though the tractor wouldn’t start. (Which pisses me off. Not only that, but we had a power failure and the generator wouldn’t start. Nothing works around here.) I got the ‘ABC’ for dinner: asparagus, pork belly, and chick peas. It sounded good on paper, but I burnt the belly and filled the house with smoke.
So I’ve been cooped up inside and I’ve done wonderful things with opencv. I drew a test pattern to calibrate my camera and I seem to have gotten the table-to-camera transform working using photos of a piece of plywood on the floor. I tellya, the numbers made no sense whatsoever for most of the day, and then they did. Really impressive. Good job!
Mary is off with her sisters for 3 days.
Pattern for calibrating lens distortion.
feb 10 tue
Snow on the ground. Snow on the way.
I got started on the cabinet fronts for the living room cabinets and you could argue that I’m getting carried away. The problem is that everything is perfectly square in this house. Too square. So I’m building crooked door fronts. They start off straight and, over the course of 8 doors, slant by 2 inches, like a row of books falling over. And over by the porch, I’m making square doors, but rotating the center panels’ grain direction. It might turn out good, or it might look like hell, and it looks like we’re going to find out.
I struggled mightily with how to do a gui for the vision module, and decided to use a python main so all the widgets would be available, and I’d run the opencv threads from c. This meant trying to get the .tcl to work for my widgets, and one problem led to another until … it worked. But I have to be honest: it didn’t work a lot more than it worked. It took me 4 days, but this was a really hard one. Nice work.
feb 11 wed
I glued up as many cabinet panels as I had ready and then scraped the glue and cut 4 to shape. 15 doors is a lot of pieces and I went and made almost all of them different. Go figure. Nonetheless, they’re looking good, and my only concern is that it’s going to look a little too droopy.
Yesterday, I was thinking to myself that it looks like clear sailing from here on out for c3pr. “The hard parts are behind me,” I said. Well I may have spoken too hastily, because today was a struggle to make ‘make’ work. I got it in the end, though, and that’s what counts. Tomorrow will be another problem.
feb 17 tue
It’s been really cold out. Like below zero every morning and windy. The drifts buried the driveway and I dug it out with the tractor. I lost the shift knob doing it. And to think I used to clear the driveway with a shovel in westford! I came in frozen and felt like I’d been defeated. The way I figured it, I worked all of nov, dec, jan outside, so it’s OK to work in the nice warm basement for awhile.
The cabinets are coming out really nicely, but the design is still evolving. What happens is, out of 14 doors, I get 12 of them right and then I have to fix whatever I fucked up on the other two. Like: I splintered 2 edges badly, so I bevelled all of them in order to fix the splinters, and the bevel looks great.
c3pr is always interesting. I’ve got a gui for calibrate/solvepnp, and I “know how to go about” the things I’ve gotta do.
Turning 3/8″ steel bolts into custom cabinet hinges.
Step 1: cut bolts to length.
Step 2: braze them in a jig
A double-door hinge, open all the way.
Good looking units after all!
feb 21 sat
No accelerated colorspace conversion found. That’s what it’s come down to. This web camera thingy has been really hard, and I’m getting this close to getting it to work.
I was going to record videos with my coolpix and then upload them to oneDrive, download to laptop, put it on thumb, move it to linux, convert it with ffmpg, and then read it in with cv. For every video. And still no live feed. So I figured: wouldn’t it be easier to just hook up a webcam? Didn’t you used to have a webcam? It’s in a cardboard box somewhere, and I went looking for it. It wasn’t in my closet. Or in the filing cabinets. Or in the basement. Or in the other part of the basement. Or in Mary’s office or the bedroom. Ah, I said: it must be in Chuck. So I put on my boots and waded thru knee deep crusty snow and looked thru all the boxes in chuck, and it wasn’t there either. So I looked in the barn, and there it was, in a cardboard box. I brought it in and thawed it out and went looking for the paperwork. I found the book, but not the disk, so I dowloaded the software and eventually got it running on the laptop. Then I got it running on Linux, where it is ungodly slow. So I converted my ballinframe() to ballinstream() and it didn’t work. And the fix seemed to use functions unavailable in c, so I had to write a test program in c++ and the includes didn’t work, and the functions wouldn’t link, and I had to use >> operator to read the file, and now I’m getting “no accelerated colorspace conversion found.”
You might not think it, but that’s a lot of fucking progress for a couple of days. Good job.
Plus, I’ve got the cabinet doors glued and half the hinge parts roughed. Good job there too!
feb 29 sun
That’s March 1 to you. It’s been really cold out and I’ve stayed inside. I’ve got most of the cabinet doors glued up and most of the hinges made. I wish I could say that I have ALL of either one done, but I doubt that I’ll get off that easy. It doesn’t sound like a lot, but it’s pretty good.
c3pr is coming along, and I have a good start on the new vision module. I have a camera hooked up and working using opencv, and a bunch of calibration routines working. I’ve even got parts of it working in c++. So all that’s really good. But this ball recognition stuff has me thrown for a loop. Even with a ball 12″ in front of the camera, I haven’t actually detected its position. And it is 100 pixels across and stationary! It is harder than it looks. So I got online and downloaded code from a phD project that tracked a tennis ball. Maybe I can learn something.
Welding goggles as a fashion accessory.