13
$\begingroup$

My boss told me he had something important to tell me, that I needed to read by the end of the day.
I told him I was on it. See, I didn't want to admit I was "Working From Home" from the local bar and didn't have my laptop anywhere near me, so had no idea what he had sent.

But when I finally got home an hour or so ago and looked at what he sent me, I now realize his attachment must have gotten messed up or something, because this is what I got:

An image that just looks like randomly colored static

I can't exactly tell him now that I'm just looking at this, or he's going to know I was not in fact "on it"—and he's going to want to know why not. That's a situation I really don't want to be in. Now I'm panicking, because I have about two hours left to get myself out of this.

So, like, can you folks help me figure out how to unwreck the image so I can see what I was supposed to see, while there's still time left before it's tomorrow??

 

—HINTS—
helpfulness level 0:

Getting something recognizable is not particularly difficult, but you'll need to figure out every bit of the puzzle for it to be a complete and correct solution.

helpfulness level 1:

Spoilery attribution added. I've eliminated a good deal of background information from the source and there are other minor tweaks, but if your result looks substantially different from the source, your method is incomplete. (See also: hint level 0.)

helpfulness level 2:

My boss is above average. Well, in one way anyway—for him, the Wadsworth Constant is more than 30%. I ignore like the first 37.5% of everything he says or writes....

helpfulness level 3:

A handful of Uno cards  

 


SPOILER WARNING: Original Image by Freepik

$\endgroup$
8
  • $\begingroup$ rot13(Gurer'f n tencuvp naq n jevggra zrffntr gung fnl "UNCCL CV QNL") $\endgroup$
    – DanDan面
    Mar 15, 2023 at 8:24
  • 1
    $\begingroup$ @DanDan0101 how did you find that? $\endgroup$
    – Stevo
    Mar 15, 2023 at 11:08
  • $\begingroup$ You have a sense of humo(u)r! $\endgroup$
    – humn
    Mar 15, 2023 at 12:02
  • 1
    $\begingroup$ Staring way too long at the picture thinking it's one of those magic eye puzzles... $\endgroup$
    – user170231
    Mar 15, 2023 at 17:11
  • $\begingroup$ Well I found the ending image, just I don't know how we get there... $\endgroup$
    – Stevo
    Mar 17, 2023 at 8:28

2 Answers 2

4
$\begingroup$

I think I'm 75% there. See update at the end.

Use Aperisolve to show the image bit by bit and find out that only the top 3 bits of each channel are noisy. Mask out those bits and shift the remaining bits 3 to the left. Most of the hidden image is visible, but the colors are all wrong. Look closely at the "black" background to find a pattern of vertical stripes.
Zoomed in on background

Print out the (unshifted) pixel values from the first row of the image to see the first 9 digits of pi, a repeating additive key.

[[3 1 4]
 [1 5 9]
 [2 6 5]
 [3 1 4]
 [1 5 9]
 [2 6 5]]

After subtracting the repeating key, the image is clearer, but still discolored. Print out the set of all values for each of the three color channels.

R [0, 1, 2, 3, 4, 5, 6, 7]
G [0, 1, 2, 3, 4, 5, 6, 7]
B [0, 1, 2, 3, 10]

The 0-7 range in R and G means there are three bits of data in each channel. The R channel has wide, ascending stripes behind the text...: Red channel stripes
... while the G channel has narrow, repeating stripes, suggesting the bits in R and G are part of the same original component, where the R bits are the more significant of the two. Green channel stripes
Combine them into the highest 6 bits of an 8-bit channel. The result looks like it could be an original V or L component of an HSV or HSL mode image. But something is off. The original image's V component is mostly around 0.5, while its L component has large areas of 1.0 intensity outside of the red parts, and this image doesn't match either of them. Combined R and G channels

The B channel is a mystery. The range 0-3 suggests only two bits of data per pixel, but also contains the value 10. Shift left four bits to visualize it and see the value 10 is in all of the places where the source image is red (fuchsia?): B channel with bright tens

Assume B represents an original H component. A linear mapping from 0-10 to 0-255 is not satisfactory. An arbitrary mapping gets me pretty close to the original, but there is obviously something missing: V component with arbitrary H lookup, S constant

PNG dissection reveals a bKGD (background color) chunk with the color 0x6c000a, but I have no idea what to do with it.

Update:

After the setter left a comment below this answer that gave away the bitfields, I was able to reconstruct the image. It turns out it was not HSV mode after all. First, combine the three bits from R, the three bits from G, and the two bits from B (wherever it is not equal to 10) in that order to create a "base component". Then, fill out a new RGB image (initially all zeroes) according to these rules:
1. Wherever the base component is all zeroes, fill the image with white (#ffffff).
2. Wherever the B channel was equal to 10, set the R channel to the base component. (The original Fuschia color can be restored with a bit of extra math here.)
3. Everywhere else, set the top two bits of the R channel to 1, and its next three bits to bits 1-3 of the base component. Set the top four bits of the G channel to the bottom four bits of the base component.
Result:
Recovered image
Source code

$\endgroup$
3
  • 1
    $\begingroup$ This is pretty close, yes :) I'm not sure what arbitrary mapping you're using; it's giving decent results and all the detail seems to be there, but the coloring is, er, off a bit. There are specific color spaces in my image, which can be accurately recovered with some fiddling: Background #ffffff; Yellowish 11xxx000/xxxx0.../...; reddish 0xxxxxx0/.../... (As for the #6c000a i don't recognize that; probably an unintended artifact of image editing) $\endgroup$
    – Rubio
    Mar 20, 2023 at 16:51
  • $\begingroup$ Those bitfields just give away the rest of it. ;) I looked at it some more after posting the answer and noticed the top bit of my combined R+G was always zero wherever B!=10, but I'm not sure I would have made the jump to splitting them back up conditionally like that, even if you had said "It's not HSV"... Anyway, answer updated. $\endgroup$ Mar 20, 2023 at 18:17
  • $\begingroup$ the bits (heh) of the solution you were missing weren't the interesting part of the puzzle anyway, just the last hurdle to a satisfactory looking result. You'd obviously independently gotten through everything else, so no sense making you trial-and-error your way through this last unhinted task. :) Well done! $\endgroup$
    – Rubio
    Mar 20, 2023 at 23:04
11
$\begingroup$

I guess, the times up?

Why?

Using this tool you will see your Boss was trying to wish you - enter image description here But, alas, its already 15th of March 2023.

$\endgroup$
3
  • $\begingroup$ Someone who deserves a place in cat-lover heaven wishes me happy π-day every year.. $\endgroup$
    – humn
    Mar 15, 2023 at 12:07
  • $\begingroup$ That's great but how'd you know? It could be any other "encryption". $\endgroup$ Mar 15, 2023 at 15:33
  • 2
    $\begingroup$ See hint level 0. This has the essentials, but it is possible to reconstruct a much nicer image. $\endgroup$
    – Rubio
    Mar 15, 2023 at 16:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.