Actually it would be quite easy for a robot to decode our messages, since the rules are known. We would just need to look at the pixels, and then compare the two frames. That what changed is the drawn line. Even if the static changed every frame so black turns white and vice versa, it would still be easy propably.
Literally just XORing two consecutive frames would reveal the pattern immediately. Here’s the truth table for xor (exclusive or): True xor True = False False xor True = True True xor False = True False xor False = False Basically, if the pixel before is different from the pixel previous, it’s true and if it just stays the same it’s false. Hope this illustrates how insanely easy this would be for computers to decrypt. Computers can see the hidden stuff in this so much easier than humans can.
Imagine some torrent-like network where you receive only a frame per minute or some, so people would be able to exchange the whole videos but none of the middle nodes would have any data that would make any sense by itself.
if anyone gets aura migraines, this really reminds me of how those look. it has this effect of a line that's not there but still flashes at you on your vision
Thanks for the comment! I don't get migraines, but people I care about do... this seems like a very useful insight. Perhaps something about a migraine causes the same 'something changed' neurons to fire when they shouldn't.
@@ChrisBLong i get migranes with the scintillating scotoma aura, within the field the visual information is lost, like an extension of the natural blind spot, and all you see there is a moving static like this. You can see something similar by rubbing your eyes with a lot of pressure.
Hey, I have an ocular condition called "Blonde fundus" that often (and I mean VERY frequently) causes aura migraines. If you suffer such, you might want to talk to your optometrist about getting an AMBER 3 lens for your prescription. It's not a normal lens, and as such they will almost never be even mentioned, let alone on hand, but it works by shifting a few frequencies of light and can help a ton. Doesn't work for everyone, but for some folks, it's a life changer. Me? I went from migraines several times a week and a continuous and persistent low grade headache to only really suffering a migraine every few months. Cuts down on the afterglow/aura effect quite a bit too, although pinks/neon orange tend to flip, blues and greens tend to flip, and some shades of yellow are absolutely invisible.................. don't expect to see yellow paint on cardboard boxes for example, but it's SO much better than living with debilitating pain if they work for you. Some optical plans even cover them, though if not, they are pricey (around $300 per lens without coverage).
@@kimudo thanks, ill have to check that out... for me i don't get them that often... leaning on maybe once a month to every few months... or at least that was the case. last week i had three... i thought it was just stress related, but who knows. either way i got my rizatriptan prescription refilled to help with that and that's been doing wonders.
There was a short indie game called "Lost in the Static" which used the same effect for the whole gameplay, making it impossible to see anything out of any screenshots.
Thanks, I found it here: silverspaceship.com/static/ That game's dev credit's Sean Howard for the idea: www.squidi.net/three/entry.php?id=56 I've learnt so much from stuff these comments, thanks to everyone who took the time to add their thoughts.
one major step in video compression is taking the difference between consecutive frames, the moving line in a non moving noise, and a moving line with no noise would have the same bit-rate
@@MartinDxtI'm also sure that it being black and white might help quite a bit, I assume that most codecs would have some sort of grayscale optimization
RU-vid seems to do a really good job at monochrome compression. In fullscreen, at 1080p, this video looks as sharp as an uncompressed video, those edges around the pixels are so crisp!
T Rex actually had really good eye sight, comparable to a bird of prey and it isn't based on movement. Jurassic Park lied to you and velociraptors are the size of turkeys too.
Thanks for the comment. Can you link Bad Apple videos? I'd like to see them. I hadn't heard of Bad Apple before - I found some cool/trippy videos but none using this effect.
@@ChrisBLong For the litterature you may want to begin by this article: Detection of Moving Local Density Differences in Dynamic Random Patterns by Human Observers. To my knowledge, most of the research on the topic has been done in the 80s-90s.
another thing that's interesting is if you skip by several seconds at a time, you see the shape that is essentially the sum of the movement during that time
1:28 _"You're not seeing a line, you're seeing a line of pixels changing from one state to another..."_ Yeah, that's still a line. You even said so yourself.
That is a really interesting observation, thank you! It only works (for me, at least) when blinking very fast. It's like your brain forgets the current state of the screen every time you blink, so the change of state doesn't register. This must tie in with the idea of 'saccadic suppression' which is a crazy topic in its own right.
This seems like an example of the Phi Phenomenon, or autokinetic effect. It's not something I've ever seen done with visual noise or staric like this before, so props to you for having a unique spin on it. I vote that since I'm not finding anything specifically noting the effect related to visual noise, that you get to be the one to name it
Exactly my thought. This feels similar to what a snake might look like in the brush. You can't see the snake itself particularly well, but the colors are changing in the pattern of a snake
i mean... not sure. Do you see tigers running a lot when they stalk? They tend to be still. Humans see movement, in fact most prey animals only detect movement but when the tiger moves its probably already too late to flee.
@@konakg8174 Not really. There was experiment of 3 extremely blury images, with 3 animals. One of those were a snake and snake was spotted WAAAY before the other 2 animals could be recognised, and the snake was spotted when the image was still insanely blury.
I remember as a kid looking at a tv tuned into static and every so often the static would look as if parts of it were spinning round even though it was just random. No idea why. This vid is fascinating though and definitely something I've never noticed before. 😎👍
Interesting but different effect. This is more to do with video compression and how highly compressed videos like on RU-vid handle drastically changing frames. Still very cool though.
It is a known effect; shape from movement in camouflage is the informal name. Notice if use equilumimant blue and yellow in a dark room, the effect will go away. You can tune on color while the other moves until you find a purple that destroys the effect. There has been some work on unconscious detection of shapes even in this invisible equiluminant case which is very strange too.
There is so much to hate about RU-vid, but reading the comments here has been amazing. One guy made a video about a cool thing he noticed, and lots of comments appear talking about related stuff, and explaining various ideas. I wish more online interactions were like this.
That is what RU-vid used to be like in the early days. There was even an option of making response videos that were directly linked to the original; you could have long discussions in comments, illustrate a point by making a video of your own.
When I was a kid I could sweari saw patterns like this in tv static. It's probably a paradolia-like effect of my brain trying to make sense of adjacent pixels changing, but it was still interesting.
I don't often see static anymore, but even now if I stare at it for long enough I'd swear it looks like a flat image of static being rotated rather than it being actually random. I believe it is random but like you I think my brain is just finding a pattern where there is none.
The visual system detects more than just brightness and color. Motion is a major component of the visual information transmitted from the retina's neural mat. This is less a persistence of vision phenomenon as it is a showcase of how sensitive our eyes are to motion, even in the absence of coherent and persistent visual markers. Still very neat though. Pure motion information.
@@SioxerNikita They very much are. Not as sensitive to it as other animals, and if you're talking only of the retina and not the neurons that measure it then you're correct, but the composite structure of the eye is very much sensitive to motion.
@@Teth47 The structure of the eye itself is not sensitive to motion, that is the visual cortex actually handling the input coming from the cones and rods in your eye. It is a technicality since that is in the brain, not the eyes itself. The eyes is not much more than a camera. Motion Detecting cameras is not really a question of the sensor itself, but rather the software inside.
@@SioxerNikitaFalse. You eyes contain Direction-Selective Ganglion Cells that fire when objects move in their preferred direction in the visual scene.
@@SioxerNikita It's weird to me that people will continue to be confidently incorrect when Google exists and you can just, like check. Seriously, google it, the first result is a whitepaper about the neural circuits on the surface of the retina that sense movement and transmit that data to the brain. The eye detects chromaticity, luminosity, contrast, edges, and motion right at the surface of the retina, the brain works with that input to segment the visual field into objects and determine their distance, size, motion, and direction. The eye is MUCH more than a camera, even without the brain.
The cube or line only exists as a concept of change to the background, rather than its own distinct shape. When you pause the video, you pause the change, and thus the cube is no more.
@@nagualdesign It is too loose to lose. Count the 'o's. Of course, this assumes you know the difference between to and too, which I realize these days isn't a given.
Because there isn’t a shape. It’s just changing the pixels of the imaginary shape, but the still image is just static. You’re following the 1-0 swap, not any image.
Subliminal pattern recognition, an ancient survival instinct. Shaking your screen will diminish the ability to sharply recognize the change. Weird stuff!
Cool discovery! This effect reminds me of stereograms: Your technique creates images that can only be seen in motion, while stereograms creates images that can only be seen when two images are overlapped. Of course the two could be combined to e.g. show a spinning cube with stereoscopic depth effect. Without eye crossing you would see two spinning cubes from slightly offset angles. But I wonder whether we could combine the two techniques in a way that shows the cube (line) _only_ if the image is _both_ animated and overlapped. We would need to create some subtle correlation between the left and right frame that is only visible in motion when overlapping the frames, and not visible when overlapping them without motion.
This is perceived by the magnocellular visual system (perception of motion), not what is called the parvocellular (which perceives color and detail). We use this in eyecare to detect early glaucoma using FDT (frequency doubling technology) visual field testing.
This is a well known property of the xor function and is commonly used for encryption. Let’s say you have plaintext, P, and you want to encrypt it. You XOR it with a random encryption key, E, and that will give you the encrypted text, T. To get P from T, you just XOR T and E again. This is what we’re doing: the line at one point in time is XOR’d with the random image at that point in time, giving us the image at the next point in time. Both images are completely random, but because we have T (the image at the second point in time) and E (the image at the first point in time), we can visually XOR them to get P (the line location). When we pause the video, all we’re doing is removing our access to E, thus making T incomprehensible.
Yes, in a random dot stereo pair picture (not a single picture auto-stereogram) each side is completely indistinguishable from a randomly generated field of dots. The information about the 3D subject is only there in the combination of the two pictures.
What's even more intriguing to me is that after having paused the video, I can still see the line or cube for about 2 seconds, after which my brain kind of gets tired of looking at the mess and the shape fades.
I dont know if it's a similar principle but this reminds me of it. In my work we have test objects that we expose under X-Rays, and have to count the contrast details to judge the image quality of the X-Ray unit. I and my colleagues agree that it is much easier to spot the contrast details when it is a live X-Ray, as the noise moving around in the X-Ray image makes it easier to spot static pixels and thus the boundaries between the details and the background.
Interesting, thanks for the comment. I remember reading something similar years ago about some signal detection algorithm that worked better when a bit of random noise was added, but I don't remember the details... EDIT: maybe this: www.ele.uri.edu/faculty/kay/New%20web/downloadable%20files/00809511.pdf
It's also neat when you scrub through the video, large chunks of pixels are clearly visible as they indicate the area changed by the moving line(s) over time compared to the background.
The interesting thing I noticed about pausing on the cube versus the line is that the line disappears into the noise immediately, whereas the cube slowly fades. It seems like, the more complex the pattern, the more points and variation your brain and vision have to hold on to.
I didn't notice until I read this, but this is actually the same effect as Pareidolia at play when looking at the cube. While commonly known for how people see faces where there are none, it is the same effect for why we see clear images in a popcorn ceiling or clouds. It isn't there, but the brain insists it is until it is able to process the entire shape and find nothing. The more complex shape likely takes slightly longer to process. Likely a result of the "edge finding" trait that another commenter talks about. Think about it like this...We can see a moving creature in the forest only because it's moving, and when it stops, our brain knows something is there, so you keep seeing it, but you can point it out to anyone else and they will never see it unless it moves again. And what's really interesting is that if you keep looking, you will continue to see it, even if it actually isn't there at all.
Yeah I noticed that too. I thought it was just my eyes doing a brief retina burn in from stopping the motion or my brain attempting to (what it thinks might be) predictive continuation of the animation even though all movement has stopped. Quite a weird side effect all the same though.
not weird at all, mammal perception is oriented around motion, i think its rods that are primarily responsible, so it works quite well outside the fovea
At "The Exploratorium", an hands-on science museum in San Francisco, CA there was a similar demonstration I recall seeing in the late 80s or maybe even 1990. The CRT was a random pattern of on/off pixels, but a coherent block of random pixels could be moved around on the screen, eg, a circle or a square, or something in the shape of an animal, say. But as soon as the motion stopped, the figure complete disappeared from perception.
Congrats on getting picked up by the algorithm. Here’s a comment so you get some extra engagement points. I would love to see more videos like this! I’ve been wanting to write a book on noise for some time, and I think something like this would fit in super well!
This may become an important topic in AI. In 3D samples (aka videos), their latent space representation should be able to encode changing random patterns. Using your observation, you could create training samples with the moving lines as ground truth labels. Probably deserves a paper :)
This gave me the idea that you could create a 3D game in which objects will only be visible when moving This would probably be a pretty interesting idea
It sure is. What would be even more interesting if this was an overlay that can be turned on and off. I think it should be possible as there is only visual things affected and nothing else. That way we could in principle play any game like this.
I wonder if there's a limit in pixel density where this effect stops happening. In super high resolution, we can't see the individual pixels, so would we see the motion image? If so, would the image appear finer, as a slight color change, or fainter? And how would this apear with greater amounts of colors? There's so much to learn about how resolution effects this phenomenon.
I'm playing with that - I expect there's a limit where your eye can't see it, even with a perfect lossless video display, but there's also a limit for these online videos where compression artefacts start to interfere with the low-res pixels...
there absolutely ought to be such a limit but I suspect it's higher than our ability to differentiate between *static* pixels. I.e. I'm guessing if you have perfect binary white noise so fine that it looks uniform grey to you, you could *still* detect moving shapes for a bit, before it's truly gone. Very curious if that hypothesis turns out to be true. It'd be a bit like how you may not see much of anything in the dark at night, *unless* there is motion. Compression really makes it tricky though. In both directions. A compression algorithm may just make it *actually* uniformly grey, losing the pixel-level detail, or it may spot and emphasize the motion, making the shape visible even when pausing, depending on how it happens to be lossy... It'd probably be easier to have like a demo website that calculates the thing on the fly in browser, needing no compression but also not a whole lot of space
There's an eerie dream like quality to watching the cube. I get the same feeling looking at autostereograms, particularly the animated ones. It's like the image is never actually there on the screen but the pixels encode straight into your mind. Very cool.
that was my first thought too - it seems very dream like, especially how we can experience in real time the sensation/aura of the line fade from memory the same way the memory of dreams often fade away soon after waking.
At one point I studied computer vision back in college where I learned about the concept of "edge detection". Edge detection is a type of pre-processing that our vision system performs automatically, we don't even perceive it. What it allows us to do is to denote the boundaries of objects very clearly. Without it, you would not be able to simply look at a table and quickly identify all of the objects laying on it such as pens and papers. You would have to study it intently to figure out where the boundaries of different objects are rather than it just being obvious. Effectively it's performing a 2nd derivative over the luminous values in your visual field and noticing when the sign flips. Edge detection (to my knowledge) works on static images so I'm not sure if this explains what's happening here. But it's possible that the sudden flipping of the pixel values in a row causes a strong activation of that edge detection circuit.
In this case it would need to first involve comparing the two images and then determine if the changes constitutes edges or not. I think that is where more advanced computer vision is expanding right now or at least need to, considering the problems self driving cars have with confusing some static images with real objects. With something that can identify where borders are your more likely to identify real 3D objects from flat surfaces, which is why some tricks work best through a screen where you cannot really get the full depth. But it ads quite a lot of extra complexity and more ways to identify things.
@@lukasegeling5205 Thanks for bringing up this concept. It's interesting though that no pixel is actually moving at all, just the line of pixels is simply flipping its state. But perhaps it's tricking us into "seeing" movement which isn't there?
It's still there, it's just hard to decipher where. It's the redraw line. It's like horizontal screen tear while gaming. But of course if this were not an indistinguishable white vs black dot array, it would be easy to see.
after reading the wikipedia page for 'Motion perception', i think the heading titled 'Second-order motion perception' is relevant. the article defines as: Second-order motion is when the moving contour is defined by contrast, texture, flicker or some other quality that does not result in an increase in luminance or motion energy in the Fourier spectrum of the stimulus. which seems to fit what the demo shows and it appears theres a huge amount of interesting reading to do, like the different hypotheses on how brains evolved to perceive such things. this is my favorite kind of youtube videos, simple, unique and great jumping off point for researching
in Jurassic Park only. In reality, T. Rex had some of the best binocular vision for a land based animal ever. A T. Rex could spot you from 3 miles away and stalk you
wow i didn't even notice this had under 1k views until u pointed out. I could've believed it had millions. It's obviously not high production or anything but short and simple informative videos like this do very well all the time
This is standard persistence of vision that is well studied and know. Persistence of vision allows you to see a series changing pictures as a smooth video.
You're not drawing lines, you're inverting pixels along a line. In the special case of a blank background that happens to result in a line, but for noisy backgrounds it just results in different noise. You're not drawing lines, you're drawing *change* and when it's paused there's no change happening and nothing to see.
That's quite novel pausing the video while the cube is moving and then watching it fade away from Clear to 'Was It Ever Really There' over a single heartbeat like a record spindown.
When I started watching, I formulated the idea that this works because our brains recognize patterns in motion, and that there is something about the consistent motion that help trick our brain. But when you started going frame by frame, I got to see each line appear crystal clear, then fade without actually fading. It felt like real magic. Amazing!
We call it relative motion. The relative motion of all the pixels "moving" give a directionality -- causing a "vector" to appear as a LATENT image in your visual cortex.
This is probably the best example I've seen of something that comes close to depicting how aphantasia affects me. Though for me it's not just black & white, there's random colours mixed in as well. For anyone wondering, aphantasia is a condition that affects one's ability to picture things in "the mind's eye", or "visualisation", or whatever. Supposedly, most people can visualise things fairly clearly, some can't see anything at all, & some of us are somewhere in between.
One thing I don't see anyone talking about is whether the bit flips are single-directional or self-reseting. Right now the cube changes a bit while it passes through, but doesn't change it back after it "leaves".
They're single flips each time something is drawn over each pixel. Replaying exactly the same set of lines in exactly the same place would completely erase the image, but I don't do that.
Predators (including humans) have more ability to spot an object that is moving than one that is not moving. This ability is even heightened in our peripheries (the area of our vision that is not the center). This is another good reason why we see what you are illustrating, we see that something moved or changed compared to the rest, regardless of the fact that it still blends in to the environment perfectly.
yessssss this illusion also works if you get 32x32 pixels of the color but one is so slightly not the same color you don't realize but if that pixel teleports you can see it but once it stops it blends in
Ive noticed this while playing with noise in Photoshop, selecting areas of noise and regenerating them creates these blinking outlines around the shape... Love it!
You see the Predator when it moves, especially against contrasting backgrounds. But when it holds still it becomes invisible. You can't ever get a good look at it, you can only shoot at the moving blur.
If you run your cursor over the watch bar on the video, you can also see just how the line and then the cube change the image as a whole. it's fascinating.
I really should've just said how cool this video is in my previous comment. It reminds me of a similar video about the 'strange loop' that arises when you train a camera on its own output and a whole world of fractals emerges. Both very cool.
There is a really cool example of this on here called "bad apple but if you pause you cant see" which recreates a music video with this effect. Glad to have seen this video now to explain how it works
I believe this effect has been researched although I don't remember the name now.I believe what is happening is that, during the split second in which the pixel changes, our brain processes the change, and then we try to analize the context in which the change happened, so we basically "fill" the line and understand it as a whole. The best comparasion is: Imagine three equal triangles fscing upwards spaced away from in a manner that they don't touch each other, two of them are aligned by the y axis at a lower level of the screen, the remaining one has its center aligned with the middle point between the lower two by the x axis. Although the three triangles are not touching each other, and are just three triangles scattered in space, the alignment leads our mind into interpreting that alignment with the idea of the three triangles being parts of a bigger one. Another great example is this: imagine you have an equilateral triangle facing upward, now cut two small segments of the same length and at the same height on the two lateral sides of it, and a bigger one on the remaining side, this should make it look like there's another triangle superimposed to the existing trisngle, but facing downward, even though there is none. The principle works similarly here: our minds notices the change in pixel colors(I believe it is due to visual permanence as you said), we interpret these changes as being part of a bigger context, which allows us to see the line and the cube. It is an optical ilusion, but the main diference is that the way our brain is processing it is actually anchored in the reality of the change(that being the fact the pixels are actually a line), instead of just some misinterpretation our brain makes. Really great video.
Isnt this just persistence of vision. Like when you see tires on a road and it appears to roll back or stop based on its speed and design? Like Helicopter blades and when you hold a pen you jiggle it really fast and it seems to be bending? I remember having to learn that as a software developer for Arduinos and LED lights to save wires. Basically its how all Monitors work, scanlines of daisychained chips that turns on and off really fast because your eyes cant conceive it. Its the effect you see when you film a stoplight with more frames then the stoplight uses.
What I think is even more neat is for me anyway, I can see all the leftover randomness vs the new randomness… it looks like someone taking a squeegee to a window with water on it. It’s leaving a streaked effect over the whole screen…anyone else see that? Also, I believe the effect is called Persistence of Vision. Persistence of vision is the optical illusion that occurs when the visual perception of an object does not cease for some time after the rays of light proceeding from it have ceased to enter the eye. The illusion has also been described as "retinal persistence", "persistence of impressions", simply "persistence" and other variations. A very commonly given example of the phenomenon is the apparent fiery trail of a glowing coal or burning stick while it is whirled around in the dark. -Wiki
It's really interesting. I can also sometimes continue to see where the line was, like I can actually identify the pixels that make it up. It is much, MUCH easier in the Bad Apple video as there you can see pretty unique shapes.
This reminds me of an encryption trick to add byte and bit entropy, where if you follow a randomly weighted bit flip you will statistically always tend to a more towards a more entropic state and unless you do all the correct bit flip reverses, (and with some minor modification, enforcing it all to be in order), you will just keep adding more chaos.
