miahi

There is another reason why (especially cheap/"normal") thermal cameras have low frame rates. Basically higher (than 9!) fps cameras can be restricted for export and travel under ITAR (International Traffic in Arms Regulations).

https://www.flircameras.com/export_conditions

Probably yes, as it means less actual sensor surface, so it's a tradeoff between better focus and noise. But it's hard to quantify without comparing electrical signals.

You don't have to think about the physical image sensors as just an array of sensors with 1:1 correspondence in the final picture. There was always a lot of computation in between - bayer filters with more green pixels, different grid patterns, multiple sensors per pixel, more pixels on the sensor than in the actual photo and so on.

Recent sensors have AF phase-detect sensors built into the image sensor, by splitting one "pixel" into two sensors (and having lots of them; sometimes in Canon sensors more than half of the pixels have this functionality; on some cameras you can extract the raw image including the different data on those pixels for bigger resolution). The difference between phase detect and contrast being that a phase-detect sensor will give exact values on how to adjust the focus, whilst contrast detection needs focus hunting in small adjustments, so it's slower.

Skydio doesn't have anything in the EU hot market (under 250 grams; over that weight you have limitations). If you do not trust DJI you can just use a different application to fly the drone instead of DJI Fly; Litchi is the best I think.

Discussion from today: "Do you know if feature X works? A customer is asking about it" "No, it was disabled because of a bug a while back" "Hmm, found the ticket, it's from 4 years ago!" "Yes, basically it's a bug from a new version of a 3rd party software that made it unusable. At that point it was basically decided that we will not try to find a workaround, we just tell the customer that it's a 3rd party issue."

The funny thing is, all the people who made that decision left the company/project and now we have the option to change it. Can't wait to see what happens next, if the new leadership changes that or not (my money is on "not").

The other funny thing is, I can think of at least two workarounds that only need a day of work and a couple for testing it.

I had issues with 10G copper SFP+ modules even when there was only one installed in the switch (Mikrotik CRS305-1G-4S+IN), the other modules installed being DAC/fiber. I got random disconnects I could not attribute until I checked the switch logs - the module was shutting down because it was reaching >90C when the ambient temp was 26C. I had to add a fan.

They are made to be good enough without costing a lot (some Noctua coolers are more expensive than a whole CPU) and usually address only the lower-level enduser market, where a difference of $5 can be an important percentage of the price. But they did evolve. At least for low-power CPUs the Intel stock coolers are not noisy and work well. The AMD Wraith coolers are on par with some of the good 3rd party coolers.

I did almost all the same upgrades to my T430 around 2014, with small variations (BIOS reflash with hardware programmer instead of software workaround for wifi upgrade, as at that point no workarounds existed). I used it for more than 2 years, but I had to upgrade because I needed more RAM. It still lives with a friend. It was surprising how well the CPU behaved, especially since the thermal solution was not really created to handle 45W CPUs (there were no T430 laptops with quad core CPUs).

I remember encountering something similar in the Lamborghini Sian kit - I think around one of the diffs; I left the axle a bit loose just in case, and that was good, because of course I mounted the diff the other way around and had to take it apart.

That is very old information and not up to date (newest laptop in that list is 14 years old). Resetting a BIOS password for newer models requires a bit of voodoo (booting up with what looks like a debug BIOS to retrieve info from somewhere else in the system).

Nice project! I did almost the same (Garmin v3 Lidar + steppers), but based on an ESP32 instead of raspi + arduino. The problem I have is the scanning speed vs resolution. The sensor is quite slow, so it takes a really long time to scan anything (hundred of hertz seems fast, but when you do the math you realize it's not fast at all).

They do have a lot of advantages: a flat country, good infrastructure for cycling, cool climate. That means you don't sweat on the way to work so you don't need showers and to change clothes.

That was a political decision. Having your own grid means that you also have your own power frequency regulation, and importing power from a different network is not as simple as connecting wires - you need to connect them using DC instead of AC. These transfer points (interchanges) are sized just for temporary power import or export, you cannot use them to power the whole Texas grid. You can see more details about the interconnects here: https://www.eia.gov/realtime_grid/#/status

There was no need for KGB to steal everything, as they already had good optics; the Zeiss factories were moved to Russian-controlled teritories during the Second World War[1], including workers.

[1] http://www.novacon.com.br/lenses08.htm

Glossy screens look better in ideal conditions. The matte screen diffuses the light (that's how matte works) a bit and reduces the contrast and it looks a bit blurry.

On phones, capacitive screens use a glass layer. Apple extended the glass layer to the laptops because it looked better (probably because matte screens looked worse than phones). Before them the screens were matte. Of course, everybody copied Apple and introduced glossy screens, because when looking at a glossy and a matte screen next to one another (in a shop) the glossy will win. And of course, everybody introduced glossy screens without the anti-reflection coating that Apple uses (cargo cult), so it made everything worse.

There was a time period when you had to really search for a non-glossy screen. Now it's a bit better, you can find them, but the touchscreen ones will always be glossy, and I did not yet see an OLED without glossy screen (could be because they are more fragile, although folding OLEDs do exist, so it's not clear).

The tiles have the format https ://www.micro-pano.com/pearl/panos/STITCH_-_FULL_35x.tiles/l8/[line]/l8_[line]_[column].jpg

where lines start from 001 to 219 and columns from 001 to 192

The strip in the foreground seems to be taxiway B[1], and not the runway (taxiways have yellow short lines, runways white long lines)

[1] https://skyvector.com/files/tpp/2014/pdf/00384AD.PDF

I remember stumbling on this page in 2003-2004. When I saw the title here I thought 'I hope it's the same' - and it is! It looks exactly like it was back then[1]. Love it!

[1] https://web.archive.org/web/20030202031646/http://www.ritsum...

The talk about striping (the fact that all the disks contain both data and the two levels of parity) is glossed over, but what it means is that there are not 7 types of failure/recovery, there are just two - when one drive fails, and when two drives fail. And the recovery actually contains all cases all the time (case 1-3 for 1 disk, case 4-7 for two disks).

It's not wrong when the interest for saving money is actually negative. This happens in Denmark

> In Denmark, negative deposit rates are no longer an exotic phenomenon reserved for the largest corporations. Panel B of Figure 3 shows that in February 2020, three quarters of non-financial corporates' deposits were subject to negative interest. The percentage is even higher for non-bank financial firms

https://voxeu.org/article/negative-interest-rates-danish-exp...

Lenovo P53 has 3 NVMe slots, 4k OLED with touchscreen (and optional pen) and up to 128GB ECC RAM if you choose the Xeon processor. It's big and heavy, but it exists.

I hope AMD will create a better market for the ECC laptop memory (right now it's hard to find + expensive).

It depends on the sensor size. 16mm in this case is not the 35mm-equivalent.

On many mirrorless cameras the shutter is open by default (even when the camera is closed), so not even one shutter "count" on this type of usage (the shutter is only used when taking a photo, and not even then if in silent mode is used). Shutter lifespans are in the tens of thousands to hundred of thousands normally. It's good to know the value when buying used (because you don't know what you are buying, some cameras have such a high burst rate that you can reach hundred of thousands in hours if you keep the button pressed).

For many cameras you can find a power line adapter that replaces the actual battery and connects to power (Canon cameras cannot charge their batteries in-camera so this is a must for them).

It might have a Compact Flash adapter on the back (CF cards are IDE compatible and with better response times than HDDs; they are still in use in some embedded systems, you just have to use a compatible filesystem, as they don't usually have internal wear leveling). Unfortunately I cannot find any pictures of the back of the motherboard, but if you look carefully at the pictures it seems that the board has two PCBs.

The easiest way to test printing at an angle is to rotate the object 45 degrees before printing; you'll need a lot more supports, but it can be done.

The results depend on the forces your object will see. The tensile strength is lower perpendicular to the printing direction, as the adhesion between layers is lower than the continuous filament adhesion. So printing at an angle is better if you need more tensile strength on the Z axis (but then you can rotate it to have that axis flat), but it also reduces the strength on that angled direction.

The beer can is round and somewhat easy to reseal. The round part is important, as the photo paper will be wrapped around the interior[1] for a wide field of view without a lot of distortion.

[1] https://www.instructables.com/Solargraphy/

They work exactly the same as a screen. Orange is not pure red, it is red + yellow, and yellow is red + green. So orange is red + _some_ green. The in-camera processing will render the image based on the sensor input and the known properties of the filter (think of it as a color profile, a mapping between what the sensors read and the color). And the processing includes color interpolation for each pixel, as each pixel (photosite) only has one color filter, but the resulting image pixel has all three colors; these are calculated based on the neighbor photosites.

Different sensors/cameras have different filters, and combined with the manufacturer specific post-processing this gives different cameras/manufacturers a different color rendition and feel.

Deleting the queues is an operational decision that I made to be able to put the system back online after the network device was replaced (the important part was the uptime/SLA). From a quick analysis of the logs the percentage of bad messages was ~90% (there was a ~50% chance that the original "touched" bit was 0 so no change was done, but the messages had multiple "touched" bits at fixed intervals).

There was an escape hatch, but the conditions to hit it were a bit complex. Implementing new message filtering of this kind at 2AM while the system was down was not feasible.

This kind of incident happened to me in a system that was supposed to have high availability. We had failovers for hardware, but it seems that a network device that was supposed to have HA (and was set up to pass the functionality to another device in case of failure) did not have ECC memory. One memory bit got stuck at 0 and the event was not detected at network level, as the data was repacked with a "clean" CRC. For some reason the packet headers were not affected by this, maybe because they were kept in a separate memory zone or because of memory alignment. So the device did not report any kind of suspicious activity, no errors in its statistics.

On the application side the effects were quite bad, as the data was mainly XML and, depending on where the bit was flipped, it could impact the data or the XML structure. The data had its own CRC/hash, so the packets were cleanly rejected by the application. Unfortunately the XML library from the message queue engine and the ESB we were using did not like at all when the bit flipping occurred in the XML tags (it seems fuzzing tests were not done at that point) so the message processing got stuck and we kept getting bad messages in the queues. Even worse, the queues could not be cleaned with the normal procedures because the application wanted to first display info about the messages inside - and that failed.

The network debug was non-trivial because of that header consistency - the network devices did not report any kind of packet issues, so we had to sniff the different network segments to identify the culprit. From the application point of view, we had to delete the whole message queue storage to get rid of the bad messages, and let the application handle the rest (luckily it was designed with eventual consistency and self-healing).

GPIO 3 means "The third pin with General Purpose I/O functionality" - a generic pin that can be used for custom protocols. In the case of ESP8266, the pin is also used for I2S SD (data pin of the I2S bus), and this is what is actually used, because the I2S implementation in ESP8266 can work at 80MHz.

I don't seem to find the specifications of the I2S in Raspberry Pis, all I can find is in the range of kHz. So probably it cannot be done.