HNHacker News
TopNewBestAskShowJobs

auxym

2,296 karma · joined May 2, 2014

I'm mechanical engineer with a long-lived interest in programming and computers.

Get in touch:

frtherien@gmail.com github.com/auxym

submissionscomments
auxym··on Darktable 4.0
Sorry for the misunderstanding, DT does handle the color space of the camera sensor, via the "Input Color Profile" module but this camera calibration data is not published by camera manufacturers. For Canon for example, DT has a single generic "Canon" color matrix profile, which was probably created by a user. You would get better results by calibrating your specific camera. Commercial software makers are able to obtain camera profiles (and other stuff like processing algorithms to match in-camera processing) under NDA from camera makers.

Sorry, not specific link, just stuff I picked up by following the pixls.us forum, where the DT and RT devs hang out.

auxym··on Darktable 4.0
It's not a simple case of reading data from the raw though. Out of those you listed, yes, white balance can be trivially read from the metadata and applied to the processing automatically, and in fact DT does do exactly that.

"Color space", by which I assume you mean input color profile aka the color matrix of the camera's sensor, has to be calibrated or reverse engineered by shooting standardized color targets under standard illuminants. DT does have this data and applies it automatically, but the calibration is provided by users and might not be ideal. You would probably get better results by doing your own calibration, but that is somewhat involved.

Picture control things like saturation etc that you select in camera are post processing steps that the camera applies to the raw in order to get a "developed" jpeg. They are proprietary to the camera maker and not published. Even if you can read settings like "saturation" and "portrait mode" from the metadata, you don't know which algorithm exactly that implies. Lightroom and other commercial software works with camera makers under NDA in order to reproduce the camera's processing. Could it be reverse engineered by open source software? Maybe, but the DT devs decided that they prefer to go in another direction.

auxym··on Darktable 4.0
That's quite OK. It's just that DT then is not the right software for you, since the dev team has an opposing viewpoint to yours.

Rawtherapee I think has a feature that tries to apply curves in order to match the built in jpeg preview.

Otherwise you might be better served by commercial software such as Lightroom or Capture One, as they work with camera manufacturers in order to match the in-camera processing.

auxym··on Darktable 4.0
I understand why this turns you off, but I wanted to note that what you want: some default magic settings that make the photo look like Canon's in camera processing, is an explicit non-goal for the DT devs. Don't get your hopes up.

What you are looking at when entering the darkroom is an "undeveloped" raw. Darktable's philosophy is that developing this raw photo should be 100% the user's responsibility and that whatever the camera maker's JPEG engine does is irrelevant.

I still use DT because it is very powerful, but I do agree that it has a very steep learning curve, and therefore is frustrating for anyone who isnt fully committed to learning about digital color science, etc, myself included.

auxym··on Python for Data Analysis, 3rd Edition
Not sure about colab, but its important to note that Anaconda is not free for commercial use.
auxym··on Python for Data Analysis, 3rd Edition
No experience with polars, but I've had quite a positive experience with xarray.

I still use pandas for tabular data, but anytime I have to deal with ND data, xarray is a lifesaver. No more dealing with raw, unlabeled 5-D numpy arrays, trying to remember which dimension is what.

auxym··on Ask HN: Why not have reverse life insurance that rewards longevity?
Annuities are like insurance against the worst case scenario of outliving your savings (longevity risk).
auxym··on Tell HN: Brother printers now locking out non-OEM paraphernalia
I had an Epson printer in college, it screwed me over one night where I had to print a paper to turn in the next day by refusing to print some black text because the cyan cartridge was running low.

I understand this is standard practice from all printer makers now, but it really turned me off the brand.

auxym··on Ask HN: Which password manager do you use?
Same, and Keepass2Android on my phone, using Nextcloud to sync the db. I use a combination of password and keyfile, the keyfile I only transfer offline between devices.
auxym··on Ask HN: LeetCode Just for Fun?
I enjoy doing advent of code every year.
auxym··on MIT Ends Elsevier Negotiations (2020)
I'm a research engineer in mechanical engineering. I hear about open access from time to time but it's still very fringe. Some journals offer the option of the author paying a fee in exchange for their paper being published as OA, but from my experience not many PIs see the value in this, and prefer to use their grant dollars for other things.

Sci-hub has been a huge help though. Our library subscriptions cover probably 80-90% of what I'm interested in and scihub covers the vast majority of the remaining fraction. For the last 1 or 2 %, I turn to emailing friends/contacts in other universities to ask them if they have a subscription and can get me a copy. Our library also offers the same service (inter library loans for papers) but charges 2$ and has a 1-2 week delay.

auxym··on MIT Ends Elsevier Negotiations (2020)
> Also for many fields (steam) the papers usually show up in preprinted repositories like arxiv, biorxiv...etc

To my knowledge preprint repositories are popular in physics, math, CS, bio, but not much elsewhere. I work in research in mechanical engineering and no one uses arxiv or anything like it (unfortunately).

That said scihub has not failed me often.

auxym··on Ask HN-Beginner Arduino Kits
Unless he has a specific interest/project in mind already, I'd look into something like this to start with:

https://www.sparkfun.com/products/18577

It's from sparkfun so you generally get better support and docs than a random box of stuff from Aliexpress. The QWIIC ecosystem means that you can daisy chain a whole bunch of sensors easily and without soldering. It includes a small but nice selection of sensors, actuators, buttons and LEDs to play with and some suggestion projects.

Soldering and more advanced stuff can come a bit later if he gets into it. No need to discourage a kid from learning electronics because soldering is hard and you want to save 5$ by buying a bare IC instead of a breakout board (not to mention that the current shortage means that breakouts are often more easily available to hobbyists compared to bare chips).

auxym··on Shimano Forces Hammerhead to Remove All Di2 Related Functionality From Karoo
Yeah, definitely for wider tires.

1.25in (32 mm) and 1.5 in (38 mm) tires were pretty common on the "sport touring" category of 10-speeds from the mid-70s bike boom to the early 90s or something like that.

Then marketing decided that everyone, even casual cyclists, should aspire to ride on "racing" road bikes with 23 mm (or even 21 mm) tires, even if the aero and acceleration benefits really didn't make any sense for them.

Really happy that wider tires are making a comeback and that the "gravel" category is in many ways a comeback of sport touring.

auxym··on Science needs more research software engineers
It's pretty widely used in "traditional" engineering, at least in mechanical/civil/nuclear/aerospace. My impression is that it's probably a holdover from the good old days FORTRAN-IV.

See for example Code_Aster (as in, "a code"), which is one of the largest projects for open source FEA (and it mostly rooted in academia).

auxym··on Science needs more research software engineers
Not just viewed, but literally called an assistant in the official title.

I am a licensed professional engineer (mechanical) and I work in academia (though my work is varied and does involve some code), my official title (in french) is "Professionnel de recherche", which translates to Research Professional. It is my understanding that in most of the english-speaking world, this position (someone that works for a research lab, who is qualified above the technician level, but not a PI nor a postdoc or student) is called "Research Assistant".

I write "Research Engineer" on my résumé/linkedin, because TBH for most people who don't have much experience in academia, "Research Assistant" sounds like an admin assistant (secretary).

auxym··on Build your own FPGA (2012)
Yep, blinky is the Hello, World of hardware.
auxym··on Ask HN: How to invest savings?
https://rationalreminder.ca/podcast/150

The data shows that gold actually isn't that great of an inflation hedge. It does keep a relatively stable real value, but only on a timescale of centuries. For any investment horizon on the scale of a human life, it's too volatile.

Highly recommend this podcast in general, though it does go pretty deep into technical stuff sometimes. They recently had an hour long interview with Nobel prize winner Eugene Fama (of efficient market hypothesis fame) for their 200th episode. On the topic of investing for retirement, I do recommend their interview with Wade Pfau among others. They also had an arc of a few episodes on the topic of the psychology of money and the relation with happiness, which I thought was super interesting.

auxym··on Knots 3D – Learn how to tie over 150 useful knots
Old method for splicing telegram wires: https://en.m.wikipedia.org/wiki/Western_Union_splice
auxym··on QCAD – Open Source CAD System
The other big difference is DWG compatibility for interop with AutoCAD. QCAD has good DWG compat, but only in the commercial version.

LibreCAD has not-great DWG support and is free. Last I remember reading about it, they were looking to get better DWG support via the LibreDWG library, but they needed to strip out and fully rewrite a bunch of old QCAD code to achieve that, due to licensing issues.

auxym··on QCAD – Open Source CAD System
LibreCAD is a fork of QCAD that aims to be 100% free.

https://www.librecad.org/

auxym··on My Unholy Battle with a Rock64
Pine's whole schtick is "cheap hardware with nice specs but zero software support".

You can like it or not but they're pretty open about the business model.

auxym··on Ask HN: Asynchronous FPGAs and flowchart programming
Labview provides a dataflow programming environment that can be used to program FPGAs (eg in the compactRIO line of hardware).

Not saying it's great (it's not), but it works in its niche: programming of one-off test fixtures by people who really have no idea about programming, digital circuits or FPGAs but might need the performance they afford (eg real-time high frequency control of a 100K+ RPM prototype gas turbine). You also have to be willing to shell out for their hardware (and labview itself), but in the testing world, I've found their hardware to actually be on the "pretty affordable for what you get" side, compared to the likes of HBM eDAQ or Siemens LMS setups.

auxym··on Best Raspberry Pi Projects
Pine64 maybe?

https://wiki.gentoo.org/wiki/PINE64_ROCKPro64

auxym··on Electronics Explained: Fundamentals for engineers, technicians and makers (2018)
The problem with 3-way switches is that there's (at least) 8 ways they can be wired:

https://www.easy-do-it-yourself-home-improvements.com/3-way-...

And more generally, the difficult part about building electricity is code. Codebooks are absolutely huge, difficult to read, and can vary by locality (not just by country or state but actually at the municipal level). I just don't have it in me to try and memorize a huge code book just so I can do a bit of wiring around the house (already need to memorize a bunch of standards in my day job as a mechanical engineer).

auxym··on Electronics Explained: Fundamentals for engineers, technicians and makers (2018)
Heard some good from this book: https://www.mheducation.ca/practical-electronics-for-invento...

Also a free one: https://www.analog.com/en/education/education-library/linear...

auxym··on Tcl the Misunderstood (2006)
One fundamental principle in tcl is that everything is a string (EIAS) [1]. Therefore, an empty string "" is the same thing as a "null value" or 0-length list, but generally doesn't require any special treatment (your function that takes a list as input generally should know how to handle empty lists).

[1] In practice things are typed in the interpreter internals for performance reasons, but the language behaves as if EIAS.

auxym··on Tcl the Misunderstood (2006)
Yeah, that may be why it became popular in CAE/EDA tools, one of the the few places it remains used these days. IMO Tcl works well for that use case, but it often catches shit from people who've had to use it. I think it's either because they misunderstand the language (as in the article, but also not that people who use CAE or EDA tools may have very little programming background) or that the vendor APIs provided to interact with the tool are very poorly designed (which I have definitely experienced).
auxym··on Netflix loses subscribers for the first time in 10+ years
The day Netflix has ads is the day I unsubscribe. The price, here in Canada at least, has already doubled since I first subscribed and is getting close to a cable subscription.

Ads and high price is why everyone cut the cable in the first place. (And poor content)

auxym··on Ask HN: Which movie(s) best represent hacker culture for you?
I agree, and was to note that I loved the soundtrack, too.
← PreviousPage 7 of 28Next →