Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[methylatedspirit]

It's not .bmp. It's a RAW image, encoded in interleaved (each pixel gets encoded in R-G-B) rather than planar (3 separate bitplanes of R, G, and B), 24-bit (8 bits per color) color. R, G and B only get 1 byte each. Irfanview has this little option to skip n bytes when opening a RAW image, which how I know interleaved images can create hue-shifted versions of themselves by manipulating that byte-skip value. I'll explain my entire process:

Audacity lets you import raw data, and there's a lot of encodings that you can use. I chose "unsigned 8-bit", since that's an exact 1:1 mapping of the underlying image; each color in a pixel gets 1 byte represented as 1 sample. Easy. If you shift the "audio" by 1 sample to the right, then it should be equivalent to using a byte-skip value of 1.

"Should" is a strong word though. Fortunately, I have tested this out; shifted the "audio" by 1 sample, added silence to the 1-sample gap, then exported it. (R, G, B) is now mapped to (G, B, R). 2-sample gap, and it's now (B, R, G). 3-sample, back to (R, G, B). My theoretical model for how this works may be a bit wrong, but the important thing is that it's all in the set {R, G, B}. Nowhere does it become CMY.

Now I'm left with the following: how did any of the programs in the process manage to hue-shift the resulting image so that RGB becomes CMY? Not Irfanview, not Audacity... then is it RTX Voice?

How is that thing at all aware of the audio it's being fed? It's live noise-reduction software, and the neural net it relies on is trained on human voice, to my knowledge. All it knows is that it gets audio piped into it, and it outputs its noise-reduced output to an audio device of my choice. Then Audacity records the audio device, and that's how I get my output image.

Where in any of that can you end up with a CMY version of an image?

[Mathel]

Sorry, I missed the part about RTX Voice in the previous post.

Given how RTX Voice is apparently a sound only software, I am somewhat surprised it gave even a correctable result.
I am guessing that it is not expecting an image, and even at 0 noise reductin, it does something.

I also checked the colors you posted with IrfanView myself. Apparently there is more red in magenta than in yellow, more green in yellow than in cyan, and more blue in cyan than in red.
And not just that. The three instances of the "same" CMY color are not completely equal. For example, first magenta is #B7009F, second is #B400A1, third is #B500A0.

Edit: I just got an idea of what it does. It would not explain the color shift, but would explain why they are not exact. It may compress and uncompress the "sound" with a lossy compression, which results in the image not being exact.

[methylatedspirit]

And not just that. The three instances of the "same" CMY color are not completely equal. For example, first magenta is #B7009F, second is #B400A1, third is #B500A0.

That's what I like to call "within margin of error". I wasn't sampling from the exact same region each time, and it did induce some degradation in the process, even at "0" noise reduction. It's a lossy process. I'll show you what the calibration bars looked like, since the program did degrade them away from "pure" colors. (It's a 32x3 snippet of the calibration bars, scaled up to a more visible size. Not the full thing; that's incredibly long.)

And here's what it's supposed to look like:


It's not a pure color, so some error when sampling colors is expected. I think the noise in the output calibration bar almost looks sinusoidal, luminance-wise.

Ninja'd by an edit:

Edit: I just got an idea of what it does. It would not explain the color shift, but would explain why they are not exact. It may compress and uncompress the "sound" with a lossy compression, which results in the image not being exact.

Exactly.

[MrRoboto75]

Today I had to look up what a "bone folder" was, for work in order to caption a video about tools used to conserve/repair paper artefacts in a museum, in this case for some sort of Chinese kite.  Use it to smooth out folds or mends apparently.

My exact google was something like "folder paper conservation" or something.

[Vector]

Oh, I know those as an "origami knife"

[methylatedspirit]

So, a Calculus video that a group of a few of my classmates and I made got labeled as "made for kids" on Youtube. I think the upbeat music tipped the algo off, since there's no conceivable way that this serious calculus video for serious people could be shown to kids without them sleeping from the moment "calculus" or "differentiation" or "math" gets mentioned. I wrote and solved a sample question for that thing; young children would not at all be interested in this unless they're really into math, bothering on Euler or Erdős levels of math-enthusiasm.

Personally, I would've gone the 3Blue1Brown route; use his open-souce manim library to make nice animations, but I think that would've been too much time commitment. Plus, it's a duplication of scope; dude's explained Calculus far better than I ever could in his Essence of Calculus series.

[wierd]

Youtube is just dumb. 

Realistically, there probably is no way to properly make it NOT DUMB, without angering the special snowflakes in the content industry, (which is where the primary motivational focus for all these automated detection, identification, and classification mechanisms comes from) but it is still a case of youtube just being dumb.


The hilarious scenario of "We introduced these policies to catch people making horrible videos that are not meant for kids, being shown to kids" (such as say, Happy Tree Friends), producing circumstances where content clearly not intended for children, being flagged for children, in 180 degree contravention of stated objective purpose, notwithstanding.

[Rolan7]

Wasn't there a big change last year where Youtube required content creators to mark their content as "for kids" or "not for kids?"  With the default presumably being the latter?  I guess I technically have a channel but it only has a few vidya-game shitposts, and I haven't heard anything.

The big controversy, unless I'm misremembering, is that a lot of Youtubers were dismayed to suddenly lose a lot of viewership because their content wasn't for kids, yet was popular with kids.

[Mathel]

You have to either specify the default yourself, of set it manually for every video. But that is for any newly posted video. I have no idea how it works retroactively.

What you, Rolan, presume would be sensible but I am not sure about YT being sensible. They may instead have applied some kind of algorithm to it.

[Mech#4]

You have to either specify the default yourself, of set it manually for every video. But that is for any newly posted video. I have no idea how it works retroactively.

What you, Rolan, presume would be sensible but I am not sure about YT being sensible. They may instead have applied some kind of algorithm to it.

Looking at the oldest videos I have, from early 2016 so I think before they did that, it looks like it's been defaulted to Not for Kids. It looks like it applies the overall setting for your channel to all old videos as well.

[methylatedspirit]

What you, Rolan, presume would be sensible but I am not sure about YT being sensible. They may instead have applied some kind of algorithm to it.

I think the "for kids" thing is retroactive, and it sure wasn't very sensible at the start. I've seen that some older AVGN videos managed to get tagged as "for kids" in the early days of that feature being implemented, so all comments from before that time were wiped clean. Of course, that was retracted some time after because AVGN is not at all suitable for kids (profanity, for one).

Sometimes I think it's a bit of a continuation of the backwards "presumed guilty before innocent" copyright system that YT is forced to implement on their platform.

I suppose I'd be sympathetic if I were any sort of artist, but I refuse to identify with that label. Any work I produce that resembles it is nothing more than some sequence of transformations on some set of media. Why do you think I databend frequently? It's the quintessential form of that philosophy. Minimal human element, just a set of reproducible, repeatable instructions fed to a computer that produces some consistent output. The only thing "human" about it is the choice of media.

[wierd]

I just investigated my co-workers Ford Explorer.

Complaint: Engine oil indicator flashes intermittently-- why it do that?

I check.  Yes. Engine oil pressure cycles up and down at idle. Check oil indicator flashes.  CHECK ENGINE light on.

I ask, "Do you know your check engine light is on?"

She says "it's been on a long time, so I dont worry about it."

I groan.  I go to my hoopty mobile and get out my 100$ midrange code reader, and hook it up to the ODB port.

I get 5 DTCs.  4 of them revolve around the engine running too lean, and one is about the evaporative emission system leaking (small leak).

I also check the engine oil.  It does not even read on the dipstick.  There is clear indication that whoever checked the oil last did not put the dipstick all the way down again after checking, as there is condensed oil vapor residue all over that side of the engine compartment, on the firewall, etc...

(Also, positive battery terminal is green and fuzzy)

I tell her that she is basically out of oil, and needs to put some in "Right away."  (and also about the fuzzy battery terminal. Advise her to clean it with an old toothbrush and some baking soda water)

She says "I have some (oil) in the trunk!"

So, I go, and remove the oil cap.  There is peanut butter under the cap.  I groan again.

I mention that this COULD be benign, and could be a sign of condensation in the crankcase, since this is winter time, and it is cold. I also mention that it could be a sign that the radiator fluid is leaking into the engine block, either from a cracked block, or from a bad head gasket. She indicates she has no idea what that means.  I groan even louder, then investigate the radiator.  This is a newer system with a sealed vapor process, so the fill hole for the radiator is also the overflow tank.  I look inside the overflow tank for signs of peanut butter. Thankfully there is none.  HOWEVER, she tells me that sometimes her vehicle smells "Syrupy". 

I inform her that she needs to take it to a shop for compression testing, emissions compliance testing-- and that if a head gasket or engine block problem is detected to not bother trying to fix it, and to just trade it in.

...

Why?  Why does this shit happen to me? I am JUST her damn co-worker, why does she bug me about these kinds of things?

[King Zultan]

I've noticed that any time you mention that you know those kinds of things to people they automatically assume your the person to go to for that sort of thing, even if you have other stuff you need to be doing.

[wierd]

I am NOT a mechanic!!  My brother? He is the motor-head, I am the tech-head.  I have just assisted and been a spare pair of hands for a lot of bullshit my parents did to their vehicles, since one of their boys was a motor-head, and they could abuse that fact.  As such, I have seen a lot of common engine problems, and what their symptoms are.

I am not qualified to actually fix these kinds of things, but I can direct a reasonable course of action with a proper mechanic. 


I just had that conversation with her, that there are limits to human cognition and ability to acquire and sustain skills.  She felt I was smart enough to gain being a mechanic as well. 


Jeebus H Christ on a pogostick.  Why me?

[LordBaal]

No good deed goes unpunished?