3,055 karma · joined September 21, 2017
Interesting to see someone evoke DjVu.
With the exception of IW44 wavelet compression, basically everything the DjVu file format supports has a PDF equivalent. I built a tool to convert DjVu to PDF that preserves the image layers and file structure with nearly equivalent compression.
My tool did expose some edge cases in the PDF standard which was frustrating. For instance, PDF supports applying a bitonal mask to an image, but it does not specify how to apply it if the two images have different resolution (DPI). It took many years to get Apple to bring their implementation into consistency.
Both of these intermediate stages between verilog and the final working hardware are where all of the secret sauce in semiconductor design actually lives. Both of these steps will be affected by the process target. Are you building for an FPGA or CPLD? Which one? If for an ASIC what is the process node? What logic elements are available in your node? Just like when you order a PCB, the process design rules generally come with prescribed information about the manufacturing constraints -- what works and what doesn't. And ultimately you cant rely on hand layout to throw down millions/billions of circuit elements anyway so you have to build solvers and all kinds of other mind bogglingly sophisticated automation to the party.
The diminishing returns of manual optimization here should be evident; if you want to do that work, become a process design engineer.
We do need more open information in this space! Tiny Tapeout is such a great project!
This holds true for any size square keypad; the common multiple will be 1+x^2 and for square-number bases, the value of the common factor will always be “11” in conventional symbology. A 2x2 base 4 and even a 1x1 binary keypad respects the rule although it’s sort of meaningless in the latter context.
I really enjoyed working this way and kind of wish the same experience could be replicated between multiple machines and with more than 2 people. I would like it if anyone could drag an application onto a shared screen where multiple people could control/interact with it while still having the option to take a window from the shared display back to a private display, ie passsing an application from one system to another.
Also, delta trikes (at least the ones on the market currently) cant lean, so the rider has no control over center of gravity, which is generally pretty disastrous in any kind of emergency control scenario on a motorcycle.
What is particularly egregious is that they are specifically marketed and sold to older people who no longer have the strength or dexterity to control a regular motorcycle. They also usually have a lifetime of muscle memory from 2 wheel bikes that does not apply and often actively hurts their ability to control a trike.
The only way to buy insurance on them is through a specific carrier that is essentially underwritten by Harley Davidson. Death benefits are capped at like $30k.
Tadpole trikes like Can Am makes are fantastically safer, especially since the forward wheels camber and the trike can lean; however the comment about old people who have no business continuing to ride a motorcycle still stands as this is still the market segment they target.
It is considerably easier to achieve sync and high framerates with LED i would expect. There are also some dedicated chips coming to the market to drive stuff like those POV video signs that are now available on aliexpress for a song.
For the record I support both, with the exception of 3 wheel delta configuration motorcycles such as Harley's 3 wheeler. Riding one is a death wish; selling one should be a crime.
Before LEDs this type of volumetric display originally rotated a helical or tilted projection surface and used a projector which allowed the entire volume to be scanned once or twice during each rotation. This had the advantage of looking more continuous, being higher resolution, and being less expensive.
I personally thought laser based picoprojectors (of the "focus free" type) were going to explode the market for small and cheap volumetric displays, but for some reason the tech never made it out of a few weird niche products.
Ultimately I concluded that volumetric displays simply don't have a great practical application *except for* the novelty. Is anyone using such a display professionally for an actual practical purpose?
However as a de facto standard you can dump postscript or pcl to port 9100 and most good printers will happily oblige you. The problem AirPrint endeavored to solve was already solved when it came along, IMO. Its current problems are solely related to it unnecessarily being a walled garden.
However in the endless quest to make consumer goods cheaper most inexpensive printers that people purchase now interpret everything on the host and the printer interface is low level machine control, as in “move motor” type commands.
Lantronix and others have produced small “print server” gateway devices in the past that run cups and support AirPrint and talk directly to a downstream “dumb” printer.
To me all of these Mini PCs are essentially uselessly uninteresting without fast network connectivity. When is an actual vendor that is not Quotom or whatever ship one with 10GBaseT or SFP+/SFP28 ports already?
It finally worked the other day! A robot repeated "chrysanthemum" back to me when asked. Sadly it seemed to be limited to this one single trick; as I continued my attempts to exploit it, it abruptly apologized and ended the call.
Cute, but one day I am going to get lucky and a scammer is going to get absolutely wrecked.
There is one simple and easy solution to this mess, but good luck getting traction on it.
The revenue to the state is $0.15/gal towards the highway fund and $0.05/gal towards education. Since the EV surcharge does not fund education, I don't feel it is appropriate to include this portion in the comparison.
So with my actual driving data in mind, here's what I would have paid if my 3 vehicles were 30mpg ICE cars: 70,442 mi / 7 years / 30mpg = 335 gal * $0.15 = $50/year to the Texas highway fund.
Vs my actual paid: $200 (2023) * 3 cars + $200 (2024) * 3 cars = $1200
I therefore concede that my initial estimate was a bit high. I paid 24x more tax than I would have if I didn't own EV's, but this was only 20x more than the average driver due to my driving habits.
Had I not purchased and registered my vehicles before the surcharge took effect, I would have been worse off due to the $500 surcharge on initial registration:
$500 (first year registration) * 3 + $200 * 3 = $2100
But if we are going to take it into this territory, we do need to break down those taxes a little bit to make a proper comparison. On the state tax side, 75% of the fee goes to the highway fund, so @400gal the actual road tax portion is $60. The remainder goes to education per the Texas constitution, but the EV surcharge does not fund education, so I do not feel it is fair to include this portion in the comparison.
The federal fuel tax applies to fuel used for electricity generation as well, and is passed to the consumer as a surcharge. I'd argue that EVs are already paying the same federal fuel tax when they are getting their electricity from taxable fuel sources. Even so, it's not the state's job to collect it, and Texas also doesn't pay any of the EV surcharges to the federal government. Because of this, I'd argue that the federal fuel tax is completely out of scope. If the federal government decides this is unbalanced and needs to be corrected, they can change their own tax code.
Finally, don't forget that the first year surcharge is $500. I amortized the total registration fees over 5 years as 500+200x4 = $260/yr.
So at 16k miles @30mpg = 400gal * 0.15/gal = $60 in state road taxes for an ICE car vs $260 for an EV, or stated another way the EV pays the equivalent road tax of driving an ICE car 56,000 miles per year.
Here in Texas, I would ordinarily pay about $30/year in road taxes on gasoline driving a 30mpg vehicle 12,000/mi anually.
But I have an EV instead so instead I pay:
$500 in surcharge for the first year of registration and $200 surcharge for every year thereafter.
Oh whoops I misspoke; I actually have 3 EVs so despite being one person, I pay approximately 25x more road tax than the average driver here.
I'm not necessarily complaining about the /amount/ of tax but the simple fact that it is both disproportionately applied and far too low overall. The state should charge based on actual mileage, but since they just eliminated state inspections, good luck with that. Second best alternative is to make it a flat surcharge for all.
Ideally the Pi5 wants 5V@5A, which is out of the capabilities of almost every USB PD supply and certainly not something that any iPad or any other computer can supply directly.
By default the firmware is configured for a 15W power budget so normally will work fine with a 5V@3A.
Only Thunderbolt iPads can supply this 15W; older USB-C ipads are limited to 7.5W and should not be used in this application without a power splitter. Pi zero 2W or similar would be a better fit if anyone wants to do this on older iPads or any other tablet that has limites current source capability.
Also this entire process probably also applies on lightning iOS devices using the camera connection kit and a power splitter as lightning can only source a couple of watts.
The irony is they will probably still have to burn it on a CD and mail it to you. Good old DICOM!
Replying to myself here; seems that the combination of being ex vivo and 7T means that the resolution increase over a typical in vivo patient scan is really really tremendous since your dead brain is happy to remain completely still through multi-hour sequences. The resolution you can get is primarily dependent on field strength but like many things in signal processing, you can trade integration time for higher resolution to a point. So this dataset is at about 100micron while 7T in vivo can only get down to maybe 300-1000micron (best case depending on patient movement)
I only really know about the physics of MRI, so I admit I don't fully understand what the advantage of this project is against other work; I assume it would have a lot to do with the fact that the data is so much higher resolution and can identify finer structure within the brain.
Incorrect. An accelerometer glued to the mountain will show -9.8m/s^2 because it's being accelerated by said mountain.