Few sentences can have global significance.

Ethan Zuckerman invented the pop-up commercial in 1997.

He was working for Tripod.com, an online service that let people make little web pages for free. Tripod offered advertising to make money. Advertisers didn't enjoy seeing their advertising next to filthy content, like a user's anal sex website.

Zuckerman's boss wanted a solution. Wasn't there a way to move the ads away from user-generated content?

When you visited a Tripod page, a pop-up ad page appeared. So, the ad isn't officially tied to any user page. It'd float onscreen.

Here’s the thing, though: Zuckerman’s bit of Javascript, that created the popup ad? It was incredibly short — a single line of code:

window.open('http://tripod.com/navbar.html'
"width=200, height=400, toolbar=no, scrollbars=no, resizable=no, target=_top");

Javascript tells the browser to open a 200-by-400-pixel window on top of any other open web pages, without a scrollbar or toolbar.

Simple yet harmful! Soon, commercial websites mimicked Zuckerman's concept, infesting the Internet with pop-up advertising. In the early 2000s, a coder for a download site told me that most of their revenue came from porn pop-up ads.

Pop-up advertising are everywhere. You despise them. Hopefully, your browser blocks them.

Zuckerman wrote a single line of code that made the world worse.

I read Zuckerman's story in How 26 Lines of Code Changed the World. Torie Bosch compiled a humorous anthology of short writings about code that tipped the world.

Most of these samples are quite short. Pop-cultural preconceptions about coding say that important code is vast and expansive. Hollywood depicts programmers as blurs spouting out Niagaras of code. Google's success was formerly attributed to its 2 billion lines of code.

It's usually not true. Google's original breakthrough, the piece of code that propelled Google above its search-engine counterparts, was its PageRank algorithm, which determined a web page's value based on how many other pages connected to it and the quality of those connecting pages. People have written their own Python versions; it's only a few dozen lines.

Google's operations, like any large tech company's, comprise thousands of procedures. So their code base grows. The most impactful code can be brief.

The examples are fascinating and wide-ranging, so read the whole book (or give it to nerds as a present). Charlton McIlwain wrote a chapter on the police beat algorithm developed in the late 1960s to anticipate crime hotspots so law enforcement could dispatch more officers there. It created a racial feedback loop. Since poor Black neighborhoods were already overpoliced compared to white ones, the algorithm directed more policing there, resulting in more arrests, which convinced it to send more police; rinse and repeat.

Kelly Chudler's You Are Not Expected To Understand This depicts the police-beat algorithm.

Even shorter code changed the world: the tracking pixel.

Lily Hay Newman's chapter on monitoring pixels says you probably interact with this code every day. It's a snippet of HTML that embeds a single tiny pixel in an email. Getting an email with a tracking code spies on me. As follows: My browser requests the single-pixel image as soon as I open the mail. My email sender checks to see if Clives browser has requested that pixel. My email sender can tell when I open it.

Adding a tracking pixel to an email is easy:

<img src="URL LINKING TO THE PIXEL ONLINE" width="0" height="0">

An older example: Ellen R. Stofan and Nick Partridge wrote a chapter on Apollo 11's lunar module bailout code. This bailout code operated on the lunar module's tiny on-board computer and was designed to prioritize: If the computer grew overloaded, it would discard all but the most vital work.

When the lunar module approached the moon, the computer became overloaded. The bailout code shut down anything non-essential to landing the module. It shut down certain lunar module display systems, scaring the astronauts. Module landed safely.

22-line code

POODOO    INHINT
    CA  Q
    TS  ALMCADR

    TC  BANKCALL
    CADR  VAC5STOR  # STORE ERASABLES FOR DEBUGGING PURPOSES.

    INDEX  ALMCADR
    CAF  0
ABORT2    TC  BORTENT

OCT77770  OCT  77770    # DONT MOVE
    CA  V37FLBIT  # IS AVERAGE G ON
    MASK  FLAGWRD7
    CCS  A
    TC  WHIMPER -1  # YES.  DONT DO POODOO.  DO BAILOUT.

    TC  DOWNFLAG
    ADRES  STATEFLG

    TC  DOWNFLAG
    ADRES  REINTFLG

    TC  DOWNFLAG
    ADRES  NODOFLAG

    TC  BANKCALL
    CADR  MR.KLEAN
    TC  WHIMPER

This fun book is worth reading.

I'm a contributor to the New York Times Magazine, Wired, and Mother Jones. I've also written Coders: The Making of a New Tribe and the Remaking of the World and Smarter Than You Think: How Technology is Changing Our Minds. Twitter and Instagram: @pomeranian99; Mastodon: @clive@saturation.social.

Deep learning-based generative art is being researched. As usual, self-learning is better. Some models, like OpenAI's DALL-E 2, require registration and can only be used online, but others can be used locally, which is usually more enjoyable for curious users. I'll demonstrate the Stable Diffusion model's operation on a standard PC.

Let’s get started.

What It Does

Stable Diffusion uses numerous components:

• A generative model trained to produce images is called a diffusion model. The model is incrementally improving the starting data, which is only random noise. The model has an image, and while it is being trained, the reversed process is being used to add noise to the image. Being able to reverse this procedure and create images from noise is where the true magic is (more details and samples can be found in the paper).

• An internal compressed representation of a latent diffusion model, which may be altered to produce the desired images, is used (more details can be found in the paper). The capacity to fine-tune the generation process is essential because producing pictures at random is not very attractive (as we can see, for instance, in Generative Adversarial Networks).

• A neural network model called CLIP (Contrastive Language-Image Pre-training) is used to translate natural language prompts into vector representations. This model, which was trained on 400,000,000 image-text pairs, enables the transformation of a text prompt into a latent space for the diffusion model in the scenario of stable diffusion (more details in that paper).

This figure shows all data flow:

The weights file size for Stable Diffusion model v1 is 4 GB and v2 is 5 GB, making the model quite huge. The v1 model was trained on 256x256 and 512x512 LAION-5B pictures on a 4,000 GPU cluster using over 150.000 NVIDIA A100 GPU hours. The open-source pre-trained model is helpful for us. And we will.

Install

Before utilizing the Python sources for Stable Diffusion v1 on GitHub, we must install Miniconda (assuming Git and Python are already installed):

wget https://repo.anaconda.com/miniconda/Miniconda3-py39_4.12.0-Linux-x86_64.sh
chmod +x Miniconda3-py39_4.12.0-Linux-x86_64.sh
./Miniconda3-py39_4.12.0-Linux-x86_64.sh
conda update -n base -c defaults conda

Install the source and prepare the environment:

git clone https://github.com/CompVis/stable-diffusion
cd stable-diffusion
conda env create -f environment.yaml
conda activate ldm
pip3 install transformers --upgrade

Download the pre-trained model weights next. HiggingFace has the newest checkpoint sd-v14.ckpt (a download is free but registration is required). Put the file in the project folder and have fun:

python3 scripts/txt2img.py --prompt "hello world" --plms --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

Almost. The installation is complete for happy users of current GPUs with 12 GB or more VRAM. RuntimeError: CUDA out of memory will occur otherwise. Two solutions exist.

Running the optimized version

Try optimizing first. After cloning the repository and enabling the environment (as previously), we can run the command:

python3 optimizedSD/optimized_txt2img.py --prompt "hello world" --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

Stable Diffusion worked on my visual card with 8 GB RAM (alas, I did not behave well enough to get NVIDIA A100 for Christmas, so 8 GB GPU is the maximum I have;).

Running Stable Diffusion without GPU

If the GPU does not have enough RAM or is not CUDA-compatible, running the code on a CPU will be 20x slower but better than nothing. This unauthorized CPU-only branch from GitHub is easiest to obtain. We may easily edit the source code to use the latest version. It's strange that a pull request for that was made six months ago and still hasn't been approved, as the changes are simple. Readers can finish in 5 minutes:

• Replace if attr.device!= torch.device(cuda) with if attr.device!= torch.device(cuda) and torch.cuda.is available at line 20 of ldm/models/diffusion/ddim.py ().

• Replace if attr.device!= torch.device(cuda) with if attr.device!= torch.device(cuda) and torch.cuda.is available in line 20 of ldm/models/diffusion/plms.py ().

• Replace device=cuda in lines 38, 55, 83, and 142 of ldm/modules/encoders/modules.py with device=cuda if torch.cuda.is available(), otherwise cpu.

• Replace model.cuda() in scripts/txt2img.py line 28 and scripts/img2img.py line 43 with if torch.cuda.is available(): model.cuda ().

Run the script again.

Testing

Test the model. Text-to-image is the first choice. Test the command line example again:

python3 scripts/txt2img.py --prompt "hello world" --plms --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

The slow generation takes 10 seconds on a GPU and 10 minutes on a CPU. Final image:

Hello world is dull and abstract. Try a brush-wielding hamster. Why? Because we can, and it's not as insane as Napoleon's cat. Another image:

Generating an image from a text prompt and another image is interesting. I made this picture in two minutes using the image editor (sorry, drawing wasn't my strong suit):

I can create an image from this drawing:

python3 scripts/img2img.py --prompt "A bird is sitting on a tree branch" --ckpt sd-v1-4.ckpt --init-img bird.png --strength 0.8

It was far better than my initial drawing:

I hope readers understand and experiment.

Stable Diffusion UI

Developers love the command line, but regular users may struggle. Stable Diffusion UI projects simplify image generation and installation. Simple usage:

• Unpack the ZIP after downloading it from https://github.com/cmdr2/stable-diffusion-ui/releases. Linux and Windows are compatible with Stable Diffusion UI (sorry for Mac users, but those machines are not well-suitable for heavy machine learning tasks anyway;).

• Start the script.

Done. The web browser UI makes configuring various Stable Diffusion features (upscaling, filtering, etc.) easy:

V2.1 of Stable Diffusion

I noticed the notification about releasing version 2.1 while writing this essay, and it was intriguing to test it. First, compare version 2 to version 1:

• alternative text encoding. The Contrastive LanguageImage Pre-training (CLIP) deep learning model, which was trained on a significant number of text-image pairs, is used in Stable Diffusion 1. The open-source CLIP implementation used in Stable Diffusion 2 is called OpenCLIP. It is difficult to determine whether there have been any technical advancements or if legal concerns were the main focus. However, because the training datasets for the two text encoders were different, the output results from V1 and V2 will differ for the identical text prompts.

• a new depth model that may be used to the output of image-to-image generation.

• a revolutionary upscaling technique that can quadruple the resolution of an image.

• Generally higher resolution Stable Diffusion 2 has the ability to produce both 512x512 and 768x768 pictures.

The Hugging Face website offers a free online demo of Stable Diffusion 2.1 for code testing. The process is the same as for version 1.4. Download a fresh version and activate the environment:

conda deactivate  
conda env remove -n ldm  # Use this if version 1 was previously installed
git clone https://github.com/Stability-AI/stablediffusion
cd stablediffusion
conda env create -f environment.yaml
conda activate ldm

Hugging Face offers a new weights ckpt file.

The Out of memory error prevented me from running this version on my 8 GB GPU. Version 2.1 fails on CPUs with the slow conv2d cpu not implemented for Half error (according to this GitHub issue, the CPU support for this algorithm and data type will not be added). The model can be modified from half to full precision (float16 instead of float32), however it doesn't make sense since v1 runs up to 10 minutes on the CPU and v2.1 should be much slower. The online demo results are visible. The same hamster painting with a brush prompt yielded this result:

It looks different from v1, but it functions and has a higher resolution.

The superresolution.py script can run the 4x Stable Diffusion upscaler locally (the x4-upscaler-ema.ckpt weights file should be in the same folder):

python3 scripts/gradio/superresolution.py configs/stable-diffusion/x4-upscaling.yaml x4-upscaler-ema.ckpt

This code allows the web browser UI to select the image to upscale:

The copy-paste strategy may explain why the upscaler needs a text prompt (and the Hugging Face code snippet does not have any text input as well). I got a GPU out of memory error again, although CUDA can be disabled like v1. However, processing an image for more than two hours is unlikely:

Stable Diffusion Limitations

When we use the model, it's fun to see what it can and can't do. Generative models produce abstract visuals but not photorealistic ones. This fundamentally limits The generative neural network was trained on text and image pairs, but humans have a lot of background knowledge about the world. The neural network model knows nothing. If someone asks me to draw a Chinese text, I can draw something that looks like Chinese but is actually gibberish because I never learnt it. Generative AI does too! Humans can learn new languages, but the Stable Diffusion AI model includes only language and image decoder brain components. For instance, the Stable Diffusion model will pull NO WAR banner-bearers like this:

V1:

V2.1:

The shot shows text, although the model never learned to read or write. The model's string tokenizer automatically converts letters to lowercase before generating the image, so typing NO WAR banner or no war banner is the same.

I can also ask the model to draw a gorgeous woman:

V1:

V2.1:

The first image is gorgeous but physically incorrect. A second one is better, although it has an Uncanny valley feel. BTW, v2 has a lifehack to add a negative prompt and define what we don't want on the image. Readers might try adding horrible anatomy to the gorgeous woman request.

If we ask for a cartoon attractive woman, the results are nice, but accuracy doesn't matter:

V1:

V2.1:

Another example: I ordered a model to sketch a mouse, which looks beautiful but has too many legs, ears, and fingers:

V1:

V2.1: improved but not perfect.

V1 produces a fun cartoon flying mouse if I want something more abstract:

I tried multiple times with V2.1 but only received this:

The image is OK, but the first version is closer to the request.

Stable Diffusion struggles to draw letters, fingers, etc. However, abstract images yield interesting outcomes. A rural landscape with a modern metropolis in the background turned out well:

V1:

V2.1:

Generative models help make paintings too (at least, abstract ones). I searched Google Image Search for modern art painting to see works by real artists, and this was the first image:

I typed "abstract oil painting of people dancing" and got this:

V1:

V2.1:

It's a different style, but I don't think the AI-generated graphics are worse than the human-drawn ones.

The AI model cannot think like humans. It thinks nothing. A stable diffusion model is a billion-parameter matrix trained on millions of text-image pairs. I input "robot is creating a picture with a pen" to create an image for this post. Humans understand requests immediately. I tried Stable Diffusion multiple times and got this:

This great artwork has a pen, robot, and sketch, however it was not asked. Maybe it was because the tokenizer deleted is and a words from a statement, but I tried other requests such robot painting picture with pen without success. It's harder to prompt a model than a person.

I hope Stable Diffusion's general effects are evident. Despite its limitations, it can produce beautiful photographs in some settings. Readers who want to use Stable Diffusion results should be warned. Source code examination demonstrates that Stable Diffusion images feature a concealed watermark (text StableDiffusionV1 and SDV2) encoded using the invisible-watermark Python package. It's not a secret, because the official Stable Diffusion repository's test watermark.py file contains a decoding snippet. The put watermark line in the txt2img.py source code can be removed if desired. I didn't discover this watermark on photographs made by the online Hugging Face demo. Maybe I did something incorrectly (but maybe they are just not using the txt2img script on their backend at all).

Conclusion

The Stable Diffusion model was fascinating. As I mentioned before, trying something yourself is always better than taking someone else's word, so I encourage readers to do the same (including this article as well;).

Is Generative AI a game-changer? My humble experience tells me:

• I think that place has a lot of potential. For designers and artists, generative AI can be a truly useful and innovative tool. Unfortunately, it can also pose a threat to some of them since if users can enter a text field to obtain a picture or a website logo in a matter of clicks, why would they pay more to a different party? Is it possible right now? unquestionably not yet. Images still have a very poor quality and are erroneous in minute details. And after viewing the image of the stunning woman above, models and fashion photographers may also unwind because it is highly unlikely that AI will replace them in the upcoming years.

• Today, generative AI is still in its infancy. Even 768x768 images are considered to be of a high resolution when using neural networks, which are computationally highly expensive. There isn't an AI model that can generate high-resolution photographs natively without upscaling or other methods, at least not as of the time this article was written, but it will happen eventually.

• It is still a challenge to accurately represent knowledge in neural networks (information like how many legs a cat has or the year Napoleon was born). Consequently, AI models struggle to create photorealistic photos, at least where little details are important (on the other side, when I searched Google for modern art paintings, the results are often even worse;).

• When compared to the carefully chosen images from official web pages or YouTube reviews, the average output quality of a Stable Diffusion generation process is actually less attractive because to its high degree of randomness. When using the same technique on their own, consumers will theoretically only view those images as 1% of the results.

Anyway, it's exciting to witness this area's advancement, especially because the project is open source. Google's Imagen and DALL-E 2 can also produce remarkable findings. It will be interesting to see how they progress.

We may never tap again in ten years.

I discussed styling websites and web apps on smartwatches in my earlier article on W3C standardization.

The Parallax Chronicles

Section containing examples and flying objects

Another intriguing Working Draft I found applies to all devices, including smartphones.

These pages may have something intriguing. Take your time. Return after scrolling:

• Exhibit 1

• Exhibit 2

• Exhibit 3

What connects these three pages?

JustinWick at English Wikipedia • CC-BY-SA-3.0

Scroll-linked animation, commonly called parallax, is the effect.

WordPress theme developers' quick setup and low-code tools made the effect popular around 2014.

Parallax: Why Designers Love It

The chapter that your designer shouldn't read

Online video playback required searching, scrolling, and clicking ten years ago. Scroll and click four years ago.

Some video sites let you swipe to autoplay the next video from an endless list.

UI designers create scrollable pages and apps to accommodate the behavioral change.

Web interactivity used to be mouse-based. Clicking a button opened a help drawer, and hovering animated it.

However, a large page with more material requires fewer buttons and less interactiveness.

Designers choose scroll-based effects. Design and frontend developers must fight the trend but prepare for the worst.

How to Create Parallax

The component that you might want to show the designer

JavaScript-based effects track page scrolling and apply animations.

Javascript libraries like lax.js simplify it.

Using it needs a lot of human mathematical and physical computations.

Your asset library must also be prepared to display your website on a laptop, television, smartphone, tablet, foldable smartphone, and possibly even a microwave.

Overall, scroll-based animations can be solved better.

CSS Scroll-linked Animations

CSS makes sense since it's presentational. A Working Draft has been laying the groundwork for the next generation of interactiveness.

The new CSS property scroll-timeline powers the feature, which MDN describes well.

Before testing it, you should realize it is poorly supported:

Firefox 103 currently supports it.

There is also a polyfill, with some demo examples to explore.

Summary

Web design was a protracted process. Started with pages with static backdrop images and scrollable text. Artists and designers may use the scroll-based animation CSS API to completely revamp our web experience.

It's a promising frontier. This post may attract a future scrollable web designer.

Ps. I have created flashcards for HTML, Javascript etc. Check them out!

Elon's rides start rough, but then...

Elon Musk has never been so hated.

They don’t get Elon.

• He began using PayPal in this manner.

• He began with SpaceX in a similar manner.

• He began with Tesla in this manner.

Disruptive.

Elon had rocky starts. His creativity requires it. Just like writing a first draft.

His fastest way to find the way is to avoid it.

PayPal's pricey launch

PayPal was a 1999 business flop.

They were considered insane.

Elon and his co-founders had big plans for PayPal. They adopted the popular philosophy of the time, exchanging short-term profit for growth, and pulled off a miracle just before the bubble burst.

PayPal was created as a dollar alternative. Original PayPal software allowed PalmPilot money transfers. Unfortunately, there weren't enough PalmPilot users.

Since everyone had email, the company emailed payments. Costs rose faster than sales.

The startup wanted to get a million subscribers by paying $10 to sign up and $10 for each referral. Elon thought the price was fair because PayPal made money by charging transaction fees. They needed to make money quickly.

A Wall Street Journal article valuing PayPal at $500 million attracted investors. The dot-com bubble burst soon after they rushed to get financing.

Musk and his partners sold PayPal to eBay for $1.5 billion in 2002. Musk's most successful company was PayPal.

SpaceX's start-up error

Elon and his friends bought a reconditioned ICBM in Russia in 2002.

He planned to invest much of his wealth in a stunt to promote NASA and space travel.

Many called Elon crazy.

The goal was to buy a cheap Russian rocket to launch mice or plants to Mars and return them. He thought SpaceX would revive global space interest. After a bad meeting in Moscow, Elon decided to build his own rockets to undercut launch contracts.

Then SpaceX was founded.

Elon’s plan was harder than expected.

Explosions followed explosions.

• Millions lost on cargo.

• Millions lost on the rockets.

Investors thought Elon was crazy, but he wasn't.

NASA's biggest competitor became SpaceX. NASA hired SpaceX to handle many of its missions.

Tesla's shaky beginning

Tesla began shakily.

• Clients detested their roadster.

• They continued to miss deadlines.

Lotus would handle the car while Tesla focused on the EV component, easing Tesla's entry. The business experienced elegance creep. Modifying specific parts kept the car from getting worse.

Cost overruns, delays, and other factors changed the Elise-like car's appearance. Only 7% of the Tesla Roadster's parts matched its Lotus twin.

Tesla was about to die.

Elon saved the mess as CEO.

He fired 25% of the workforce to reduce costs.

Elon Musk transformed Tesla into the world's most valuable automaker by running it like a startup.

Tesla hasn't spent a dime on advertising. They let the media do the talking by investing in innovation.

Elon sheds. Elon tries. Elon learns. Elon refines.

Twitter doesn't worry me.

The media is shocked. I’m not.

This is just Elon being Elon.

• Elon makes lean.

• Elon tries new things.

• Elon listens to feedback.

• Elon refines.

Besides Twitter will always be Twitter.

Writing for readers:

14 years of daily writing.

I post practically everything on social media. I authored hundreds of articles, thousands of tweets, and numerous volumes to almost no one.

Tens of thousands of readers regularly praise me.

I despised writing. I'm stuck now.

I've learned what readers like and what doesn't.

Here are some essential guidelines for writing with impact:

Readers won't understand your work if you can't.

Though obvious, this slipped me up. Share your truths.

Stories engage human brains.

Showing the journey of a person from worm to butterfly inspires the human spirit.

Overthinking hinders powerful writing.

The best ideas come from inner understanding in between thoughts.

Avoid writing to find it. Write.

Writing a masterpiece isn't motivating.

Write for five minutes to simplify. Step-by-step, entertaining, easy steps.

Good writing requires a willingness to make mistakes.

So write loads of garbage that you can edit into a good piece.

Courageous writing.

A courageous story will move readers. Personal experience is best.

Go where few dare.

Templates, outlines, and boundaries help.

Limitations enhance writing.

Excellent writing is straightforward and readable, removing all the unnecessary fat.

Use five words instead of nine.

Use ordinary words instead of uncommon ones.

Readers desire relatability.

Too much perfection will turn it off.

Write to solve an issue if you can't think of anything to write.

Instead, read to inspire. Best authors read.

Every tweet, thread, and novel must have a central idea.

What's its point?

This can make writing confusing.

️ Don't direct your reader.

Readers quit reading. Demonstrate, describe, and relate.

Even if no one responds, have fun. If you hate writing it, the reader will too.