« first day (1524 days earlier)      last day (1398 days later) » 

09:29
That sounds like a good solution. It would allow for switching it off without actually having windows thinks the device is disconnected.
I like how your view of plug and play on Linux is that it requires further tweaking. I guess you could just connect your video feed to an AI that upscales it to 30fps.
 
4 hours later…
13:20
well, plug and play in the sense that it can be used immediately and without the need to install any drivers. It can be handled by the generic "USB Video Class" driver, but I think I recall reading how support for UVC 1.5 is ignored and (for now) you just get the functionality of UVC 1.2 or something
 
5 hours later…
17:53
you definitely did a deep dive into the specs before buying. Is it still common for peripherals not to be supported under Linux?
Not at all. Linux these days is fairly easy to just jump into. The basic stuff like gpu drivers, ethernet and wifi, keyboards and mice just work.
If you have more specialized peripherals (video capture cards, digital audio workstations, digital art tablets like Wacom), there can be some friction.
Welcome @KrisGoncalves always good to see more people here
Thanks. Had stumbled upon this a bit ago and never had the fortune of being on when others were on.
but, your comment on you recent question brought me back
Yeah, at this point it's mostly been me and Kasper trying to keep chat alive.
I might just go and start on that story about that image rotation problem before fox gets here.
after I make another push on modifying a utility class, that is
@Kasper have I told you that story before?
would've involed words like rotation, CT scan, and Radon Transform
18:15
The fox has arrived.
heya fox! Is that nickname fine, or do you prefer something else?
No I haven't heard that story yet!
Like I mentioned, any visual aids will have to wait, but there's a chance this story can be conveyed with words alone.

So, I was taking a Digital Image Processing course towards the end of my college days. I was working on a team, and we had several choices for what we could work on for the end-of-semester project. We settled on writing a program (C++) to take an image, perform a "CT scan" on it, and then reconstruct the original image from the "scan" data.
What's a CT scan in this context? I'm assuming you didn't CAT scan it at your local hospital
I don't think I've ever seen four people in here :D
In a nutshell, a CT scan works by firing X-rays (I think) through you and having a sensor record the intensity of the beam. By rotating around the subject, you can build up a sinogram, which you can then run through the inverse Radon Transform (en.wikipedia.org/wiki/Radon_transform) to get a reasonable reconstruction of the image
18:25
but how do you perform a ct scan on an image?
including partially and totally opaque things inside of your body
For doing this "scan" on an image, it's basically just taking the maximum grayscale value of an image that lies along the line that your virtual "photon" travels through
The simplest way to do that is to assume that the photon path aligns with the columns of pixels, and take max values from each column of the image.
The part that I was in charge of was rotating the image.
I think I understand, did just skim the wikipedia article on it
we weren't using any libraries, really, so I had to build the rotation function from scratch (given a 1-D array representation of the image as input)
Now, there are some angles of rotation that work out pretty nice. 90 degrees, 180 degrees, etc... but for the vast majority of rotation angles, you're going to have a lot of input pixels end up off of the pixel grid.
so interpolation is required
makes sense
My problem was that I knew where the output pixel grid was going to be (forcing it to be on the grid), but I couldn't work out how to determine which pixels from the input I should be using to interpolate.
Assuming most of the input pixels are off-grid, which ones should you pay attention to at any given output pixel location? Each pixel is surrounded by 8 others. Do you take the ones that are closest? That'd be a lot of repeated work to determine distances, and would it even give you the right result?
So I went to the course's instructor and asked for a bit of help
18:40
i've played around a bit with image scaling once so this actually sounds familiar
The thing that got me past that roadblock was "instead of trying to rotate the input image and interpolating, consider trying to un-rotate the output grid"
i gotta run to the store real quick before they close, be back in 20 min
@DerekF had to reboot. I'm pretty sure most people know me as Brian by now, so it's up to you :p
Nice.
by reversing the process (taking the output pixel location, rotating it by (-1) * whatever angle you want to rotate it at), that gives you an output pixel that is likely off-grid, but you have input pixels that are on-grid and have well-deffined intensity values
That makes it so much easier to do the interpolation
So since then, when I get stuck on a problem, I'll eventually think to myself "is there a way that I can do whatever it is I'm trying to do in reverse? approach the problem by working from the end instead of the beginning?"
@sfdcfox Gotcha. And your suggestion about the order of package dependencies made me think "ah, reverse it!", which in turn reminded me of that college project
Yeah, makes sense. I love it when an apparently complicated algorithm turns out to be trivial if you reorder it.
19:06
Agreed, makes sense. That's a nice way to explain it
I'll plan to accept that answer tomorrow (barring getting some equally illuminating insight from another answer). 24 hours seems like a generally good threshold to me for accepting answers (and I've already upvoted).
@KrisGoncalves For now, I treat this chat room as more of a comfy, off-topic "things that are almost interesting" discussion area.
Like a mid 00s forum
Makes me wonder how many forums running software like PHPBB are still kicking around.
I think PHPBB is still under active development. I believe the Ars Technica forum is built on it
19:22
yep, phpBB is still being developed
though the advent of services like Facebook, Instagram, Twitter, and Discord have shifted at least a portion of the audience that would have otherwise used forums away from forums
yeah there's few forums left anymore. I'm guessing only privacy enthusiasts still use them. That and maybe some stragglers
I still think there's a place for forums in the digital landscape, acting as a centralized hub for an organized community.
I'd like there to be a place but I'm guessing users generally prefer a facebook page or something similar
In the end discussions go where the users are
Forums being more community-centric than Twitter, Instgram, TikTok, etc... which are centered around individual creators/influencers/brands.

Kinda hard to discuss Ford on, say, Toyota's twitter (on a regular and consistent basis at least.)
good point, i hadn't thought about that. Must say I don't visit forums that much anymore, unless you count reddit
19:39
Me either. I have started lurking on the Level1Techs forum a bit (mostly for updates on LookingGlass), but I'm not actively participating in any.
and maybe Reddit is a hybrid? or just so large that it's not just one community (each community having their own sub or two or three)
yeah reddit is a bit of an outlier
definitely more forum-like than social media-like, I'd say
yeah i like scrolling through their popular section once every while
but i don't participate much either
Well I gotta go, good luck splitting your org up into packages. Let me know how that goes
the answer is going to be "slowly" :p

« first day (1524 days earlier)      last day (1398 days later) »