Hacker News new | past | comments | ask | show | jobs | submit | gaudat's comments login

It's very likely that PCB is made in KiCAD seeing the fonts. I really really hope they make the project files available. That would be a sick example project.

Also, obligatory ꙮ


CERN is a big proponent of KiCad, so it wouldn't surprise me.

I didn't see an Open Source Hardware Logo on it, or any marking suggesting it's CERN Open Hardware License, but it might also be on the back.

The big issue is that it's tough to collaborate with something like Altium Designer when its $460/mo (and no one else can open your design unless you pay the fee).


I dig that PCB design. The panopticon (?) artwork and how the decoupling caps are lined up perfectly along the circumference.


I found details about a talk[0][1] given by the hardware team, their names are on the blurry side of the picture in the top link - Authors: GUATIERI, Francesco (Università degli Studi di Trento); MUENSTER, Markus (Student); BERGHOLD, Michael (NEPOMUC / FRM2)

The name OPHANIM seems to be the combination of "Optical Photons and Antimatter Imager" as written in the silkscreen below the exposed ground.

And a crude take at the further inscription - Ex Saxis Homo Fecit Oculos Per Artem Ingeniumque Nunc Monstrum Usum Est ad Universum Resiscendum

"Man made eyes from stones through art and ingenuity. Now the monster has been used to re-emerge from the universe."

I haven't found a recording of the authors giving that talk or any significantly different materials, but the references to 'Ophanim' seem to be associated with Ezekiel's apocalyptic writings, and angels with many eyes, so that seems to fit.

[0] https://indico.frm2.tum.de/event/484/contributions/5093/ [1] https://indico.mlz-garching.de/event/484/contributions/5093/...


At last a biblically accurate camera ;-)


While do !Fear


My Latin isn't great, but "ad universum" would be "to the universe" not "from the universe"; which would be "ab universum" (in the context of the rest of the sentence, "ad universum" is a bit odd; I would think "in universum" -- "into the universe" -- would be better.)


Assuming resiscendum is a typo for resciscendum, a better translation of the second half is probably something like: "Now this wonder is useful to learn about everything."


This post links to Quanta [https://www.quantamagazine.org/new-book-sorting-algorithm-al...], which links to the actual paper [https://arxiv.org/abs/2405.00807].

Tbh I was expecting a supercharged Dewey Decimals...


Imagine that thing painted neon green and hurling down the highway.


> hurling

Assuming you meant 'hurtling' then no, I can't imagine it, strangely the video disavowed me of any such idea.

If you did actually mean hurling, and weren't talking about the rider hurling from that slow, annoying rhythm, then no, I can't imagine that either.


I for one won't be trading in my old ZX-9R.


For someone who has dabbled in electronics my first impression is "this is useless". Soldor wick is basically multi-stranded copper wires. You can already cut it with whatever snips you have on your bench.

But Hakko is Japanese i think? They like to release wild products. I remember Makita releasing a microwave oven and coffee maker that uses their battery and has their color scheme.


The design is pretty modern. But what is with the choice of the Kodak CCD sensor? CCD cameras got a resurgence in Chinese communities with second-hand camera prices increased like tenfold.

Also see Apertus Axiom where they also used the Zynq but used one hell of a CMOS sensor that can do 4K 300FPS.


He explained it in this Video at 13:02 https://www.youtube.com/watch?v=OkfzjmY9cF8&t=13m02s

> Why use CCD instead of CMOS?

> Well let's say I started this project before the recent CCD camera hype so that's not the reason. Part of me just wanted to be special and fullframe CCD is kind of special.


Being a small manufacture means you go to the bottom of the list for the vendors of the sensors. I'm sure the best sensors are pretty much already spoken for, and the lower quality ones are what's available to any one not the likes of Sony, Canon, Nikon, etc. Any other reason is pretty much an excuse.

We could just go back to full sized cameras with 3 CCDs. j/k


CCD sensors render differently than CMOS sensors and, if their strengths are what you are looking for they could still make sense. Compared to CMOS they require more light, but when you are capturing an image properly they do a really good job of rendering colors and details. CCD sensors also introduce less noise as part of their pipeline (though again, the ceiling on how sensitive they can be is much lower than CMOS).

Basically there has always been a community of photographers who like the "CCD look" and it's not surprising to me that someone who's geeky enough to make his own camera went with one.


That sounds like the stuff you read on audiophile websites where they are trying to sell you a special "audio-tuned" ethernet cable for thousands of dollar. The sensor simply converts photons into electrons binned into colors based on the bayer pattern, how can they "render" colors differently? That's a function of the digital processing pipline.

Also I don't understand how a CCD can introduce less noise than a CMOS, but is less sensitive at the same time? Light sensitivity is largely governed by the noise floor of the sensor I would say.


I don't really know either. My understanding is that the process of increasing ISO (i.e. boosting signal) works a lot better in CMOS, but also that because of how CMOS works read noise is higher overall. So a CCD will give you a cleaner record at base sensitivity (generally iso 100) or slightly elevated iso, but as soon as you get away from that CMOS sensors will do better. Whatever read noise they introduce is less impactful than the fact that they are much better at amplifying the light you did detect in a clean(er) way.

You are also welcome to dismiss me as a goofy photophile. I am not trying to sell you anything, just reporting what I understand people think. Maybe this article seems less snake oil-y[1]?

[1] https://petapixel.com/what-is-ccd-cmos-sensor/


Sorry if it came across like it, I didn't mean to accuse you of snake oil pandering. I just read the article and I think you misremembered, the article states CMOS has in fact a significantly lower noise floor (I wondered about this, because I remember when the next gen CMOS sensors came out in the pentax K5 and the Nikon D400(?) it was possible to push the shadows so strongly that you could essentially take a completely black photo and push it up to the same as if you would have taken with higher ISO). CCDs have a "more pleasing/film-like" nonlinearity, but I believe if you shoot raw a linear response is always better, because it gives you more flexibility for post processing.


IIRC the author scored a few trays on local equivalent of eBay. Obsoleted Kodak/OnSemi KAF line of sensors are rare kind of large film-sized sensors with public full datasheets; most manufacturers don't even confirm or deny existences of sensors in public.


A part of me really wants to push this and see how far it would go. Why stop at AM? Add in shortwave radio and let us listen to stations on the other side of the earth.


Shortwave doesn’t work well as a mobile, moving medium. It is weak-signal reception (by design) and is likely to be compromised every time your car turns in a different direction.


Why not we just have a system like the beacons of Gondor?


Do you need latency measured in nanoseconds or multiplexing hundreds of sensors? Otherwise it makes a lot more sense to use a Cortex-M or Cortex-A.


Aren't FPGAs also helpful for high bandwidth/throughput applications like continuous input from a camera? (I'm not an FPGA expert, just curious)


Right. Fpgas shine if you have high data rates (multi Gigabit on single/dual pins), low latency, lots of similar computation (multiply accumulate), a lot of complex pipelined logic and if you need to be really compact at the same time.

The effort to get this running is also crazy huge. Large designs "compile" for hours, simulation and in situ testing can take months. If your fpga has 1000 pins, even PCB design, manufacturing and debugging is hard, if you even have a sufficient scope at hand. I would guess 10x compared to solving a problem with an embedded cpu, which is 10x compared to solving a problem on a standard PC. Expect another 10x if you go from Fpga to custom ASICs.

I'm curious what could be expected / possible with 5-10k. Tweaking an existing design, on proven and available hardware? Testbench for a module only? Maybe "just" VHDL/verilog coding (instead of dealing with the Fpga)?

(sorry, no marketplace)


Many of the common usecases, like camera input, will have dedicated peripheral a available in a System on Chip or microcontroller. Which would then generally be preferred over FPGA because it will require less custom development, quality assurance, etc.


That's true, I was thinking of the more "extreme" use cases like those cameras which sample at hundreds (thousands?) of images a second.


In this case, my maximum data rate is about ~256 Mbit. So not crazy, but not microcontroller either.


For your second example it's impossible to find a YouTube video of Chika dance that is available in every country. Try this: https://streamable.com/0hrdh

It's remarkable how smooth the animation is. And how natural the character's movements are. This scene jumps over the uncanny valley straight into being enjoyable.

And I think there's an AI paper from a Chinese researcher using this video as a demo.


>And the beautiful thing about adaptive attacks is that they look exactly like Hollywood, it's beautiful! Because you see them flipping and going through values, getting it right and moving to the next one, which you all told it was fake, it was not!


Several of the CPU side channel attack demos look like that too. The speed could probably be improved in weaponization but currently you can watch the ASLR defeat routine, with a percentage and everything just like you said.


  C P E 1 7 0 4 T K S


Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: