Human Shader
humanshader.com
humanshader.com
We'll start with inference, but maybe we'll eventually convince some folks to do a full training exercise.
It may take 4-5 days and 1000+ people, but it's definitely created a greater sense of community than any faster rendering system. I like looking over the different pixels and knowing that they represent some nerd-sniped engineer like me.
By the way, there's more going on in the shader algorithm than you might expect. Here's an explanation of the worksheet:
u, v are coordinates relative to the center of the image
h is radius from center, squared
Section B generates the ball:
B3-8 generates the reflected color on the ball.
B9-11 applies the diffuse illumination to the ball.
B12 adds the illumination highlight.
Section C creates the ground:
C5 puts a shadow directly under the ball.
C13 is the cast shadow of the ball.
Section D creates the sky with a simple gradient
Section E converts the image from two-color to three-colorAnother interesting thing is to try to reverse-engineer the worksheet: section B is the sphere, section C is the ground, and section D is the sky. But then there's the lighting model, shadow, etc.
Somehow I got RGB(255, 50, 194), which is different from the value posted on the chart. Actually, I'm not sure how originally they got 0, since G comes from R and B, which are both positive, and the expression is multiplication and addition.
When completed, it would show a QR code, that upon scanning would mark completion of the workbook and show you the answer key. The implicit idea being you didn't have to be perfect, just good enough so the error-correction algorithm of QR codes was enough to pick up the final image. (Probably extra credit if you could figure out how to do as few problems as needed and then use EC to still figure out the secret link)
".. Based on the unbelievably true life stories of three of these women, known as "human computers", we follow these women as they quickly rose the ranks of NASA..."
> mathematicians that served as the brains behind one of the greatest operations in U.S. history
In this case, aren't they more like hands than brains?
The brains should be the ones who created the code. These people are just processing it, it's like manual labor.
They used their brains to perform the calculations, I don't see how you could see it otherwise.
Is the McDonald's cashier the "brains behind the operation" because they count change as part of their duties?
Counting change accurately is very important to the continued successful operation of a retail establishment. But it's menial work.
Even Wikipedia agrees with me:
> Alan Turing described the "human computer" as someone who is "supposed to be following fixed rules; he has no authority to deviate from them in any detail."
I think the term "human computer" is extremely misleading without the cultural context behind the term, i.e. that these people were essentially doing 5th grade math worksheets all day. Reading numbers, plugging them into a calculator (yes, really), and writing down the results.
Usually we call these mechanical turks?
The work could require superhuman endurance, though. “They had to keep working eight hours a day doing the same equation over and over again—it must have been mind-numbing,” notes Paul Ceruzzi, author of Reckoners: The Prehistory of the Digital Computer.
from [this](https://www.smithsonianmag.com/science-nature/history-human-...) article
If I take the words of Alan Turing then your interpretation of menial work, then programming is menial work too. You have fixed rules that you can't deviate from. Is all programming really menial? I would argue that it can be very mentally taxing, the same way math or any intellectual work can be.
At its bases, NASA employed nearly 80 black women as computers, says Margot Lee Shetterly, author of Hidden Figures. One of them, Katherine Johnson, was so revered for her abilities that in 1962, John Glenn asked her to personally verify the flight path of his first launch into space on the Friendship 7 mission. The astronauts didn’t trust the newfangled digital computers, which were prone to crashing. Glenn wanted human eyes on the problem.
“They had a tremendous amount of respect for these women and their mathematical abilities,” says Shetterly. “The male engineers often were not good mathematicians. So the women made their work possible.” Still, some friction existed. Women who asked for promotions got stonewalled or turned down: “For women who wanted to move up, who wanted to be supervisors—particularly if that involves supervising men? Not so much.”
The women wouldn't have been employed without the engineering work, but the engineering work wouldn't have been possible without these women. They were equally the brains behind getting things to space. A sizable number of these women later became programmers, because building the computers was seen as the really difficult task. Coding was dull work. Writing code that gets people safely to the moon and back was obviously trivial.
Which art?
We needed a man to repair the machines, to keep them going and everything. And the army was always going to send this fellow they had, but he was always delayed.
Now, we always were in a hurry. Everything we did, we tried to do as quickly as possible. In this particular case, we worked out all the numerical steps that the machines were supposed to do — multiply this, and then do this, and subtract that. Then we worked out the program, but we didn’t have any machine to test it on. So we set up this room with girls in it. Each one had a Marchant: one was the multiplier, another was the adder. This one cubed — all she did was cube a number on an index card and send it to the next girl.
We went through our cycle this way until we got all the bugs out. It turned out that the speed at which we were able to do it was a hell of a lot faster than the other way, where every single person did all the steps. We got speed with this system that was the predicted speed for the IBM machine. The only difference is that the IBM machines didn’t get tired and could work three shifts. But the girls got tired after a while.
https://calteches.library.caltech.edu/34/3/FeynmanLosAlamos....
Human generative reverse-diffusion AI.
1. You have no prompt.
2. You have 1 iteration.
Hence why I suggested prompt and multiple iterations. It's a very subtle tweak, but in aggregate behavior, everything is subtle and has huge effects.
Of course...
It you are into shaders and don't know that guy, well you are one of the "lucky 10000" and you should check him out. ( https://iquilezles.org/ )
If you compare them to nearby "correct" pixels, they usually just have one of the three RGB values that are sharply different.
But there’s no promise that they will. The photo feature may be more of an automated speed bump, a way to reduce silly answers and pixel spam, and let people self-select, than an active human moderation tool. Moderating is boring and expensive in time and/or money, why would anyone actually sift through thousands of hand-written pages of arithmetic?
2840 pages. Doesn't seem crazy. Page looks like it was written by hand? Next. At two pages per second it's less than half an hour of work.
Yes obviously it “can” be done, I’m suggesting it won’t for long, because there’s a big difference between submitting something to the site mods versus submitting something that is anonymously exposed to the public.
I’m saying this as the owner of a site where I made the mistake of allowing crowd-sourced image content to be anonymously served to others. It didn’t take long before not only was there NSFW content, but there was also illegal content.
* edit OH BTW I only just noticed this site was created by IQ, who has already dealt with this exact issue on ShaderToy and had to restrict and remove user submitted images due to abuse! The decision to not show people’s images is almost certainly intentional and by design.
My understanding from looking at the worksheet is the person who created the target image has created three separate formulas (depending on the area of the image), that when you feed in the X and the Y coordinates, it spits out the correct RGB value for that pixel. Is that correct? That’s wild.
A really basic one returning red for every pixel would make a red screen. The most common use would be rendering a textured mesh and determining how lighting should effect it (is it skin? A metal surface?).
Then you can look at crazy things like https://www.shadertoy.com/ which are all purely made in pixel shaders (also that site is run by the creator of the Human Shader)
So a simple shader would be something like (in pseudo code):
function shader( x, y ) { if( x > 0.5 ) return white; else return black; }
would color one half of the screen black, and the other white.
Highly recommended for anyone who wants a chill afternoon challenge. And the best part is seeing the final image come together.
This is 0,000002157210402 FPS so far.
Shouldn't have claimed a pixel.
Heuristically we can still get an estimate: Group names to individuals and groups. Guess average group size. Take difference between claim and submission for time it took to compute.
Why not 1 at a time, and you can claim another immediately after completing the previous?
And if you have 4 pixels awaiting verification, you cannot claim another pixel unless and until one of those pixels is verified?
Well I got another pixel anyway.
It could be split up into easier steps, like adding a pair of 4-digit numbers or multiplying a pair of 1/2 digit numbers.
Let it cook
Saying the mistakes make this worse is like saying "I'd really enjoy spaghetti, I think, but all the noodles are ruining it."
> Thanks a lot for being part of the Human Shader, go find your pixel in the public canvas! Tip: if its color looks wrong to you, feel free to review your calculations and submit again with the same code!