https://tomconnor.me/wp-content/uploads/2021/08/10x-Engineer-Mental-Models.pdf

I keep returning to the same mental models and tricks after writing and reading about a wide range of topics.

Top 12 mental models

12.

Survival bias - We perceive the surviving population as remarkable, yet they may have gotten there through sheer grit.

Survivorship bias affects us in many situations. Our retirement fund; the unicorn business; the winning team. We often study and imitate the last one standing. This can lead to genuine insights and performance improvements, but it can also lead us astray because the leader may just be lucky.

11.

The Helsinki Bus Theory - How to persevere Buss up!

Always display new work, and always be compared to others. Why? Easy. Keep riding. Stay on the fucking bus.

10.

Until it sticks… Turning up every day… — Artists teach engineers plenty. Quality work over a career comes from showing up every day and starting.

9.

WRAP decision making process (Heath Brothers)

Decision-making WRAP Model:

W — Widen your Options

R — Reality test your assumptions

A — Attain Distance

P — Prepare to be wrong or Right

8.

Systems for knowledge worker excellence - Todd Henry and Cal Newport write about techniques knowledge workers can employ to build a creative rhythm and do better work.

Todd Henry's FRESH framework:

1. Focus: Keep the start in mind as you wrap up.

2. Relationships: close a loop that's open.

3. Pruning is an energy.

4. Set aside time to be inspired by stimuli.

5. Hours: Spend time thinking.

7.

Black Box Thinking…..

BBT is learning from mistakes. Science has transformed the world because it constantly updates its theories in light of failures. Complexity guarantees failure. Do we learn or self-justify?

6.

The OODA Loop - Competitive advantage

O: Observe: collect the data. Figure out exactly where you are, what’s happening.

O: Orient: analyze/synthesize the data to form an accurate picture.

D: Decide: select an action from possible options

A: Action: execute the action, and return to step (1)

Boyd's approach indicates that speed and agility are about information processing, not physical reactions. They form feedback loops. More OODA loops improve speed.

5.

Know your Domain 

Leaders who try to impose order in a complex situation fail; those who set the stage, step back, and allow patterns to develop win.

https://vimeo.com/640941172?embedded=true&source=vimeo_logo&owner=11999906

4.

The Three Critical Gaps

• Information Gap - The discrepancy between what we know and what we would like to know

• Gap in Alignment - What individuals actually do as opposed to what we wish them to do

• Effects Gap - the discrepancy between our expectations and the results of our actions

3.

Theory of Constraints — The Goal  - To maximize system production, maximize bottleneck throughput.

• Goldratt creates a five-step procedure:

1. Determine the restriction

2. Improve the restriction.

3. Everything else should be based on the limitation.

4. Increase the restriction

5. Go back to step 1 Avoid letting inertia become a limitation.

Any non-constraint improvement is an illusion.

2.

Serendipity and the Adjacent Possible - Why do several amazing ideas emerge at once? How can you foster serendipity in your work?

You need specialized abilities to reach to the edge of possibilities, where you can pursue exciting tasks that will change the world. Few people do it since it takes a lot of hard work. You'll stand out if you do.

Most people simply lack the comfort with discomfort required to tackle really hard things. At some point, in other words, there’s no way getting around the necessity to clear your calendar, shut down your phone, and spend several hard days trying to make sense of the damn proof.

1.

Boundaries of failure - Rasmussen's accident model.

Rasmussen modeled this. It has economic, workload, and performance boundaries.

The economic boundary is a company's profit zone. If the lights are on, you're within the economic boundaries, but there's pressure to cut costs and do more.

Performance limit reflects system capacity. Taking shortcuts is a human desire to minimize work. This is often necessary to survive because there's always more labor.

Both push operating points toward acceptable performance. Personal or process safety, or equipment performance.

If you exceed acceptable performance, you'll push back, typically forcefully.

Here’s why that’s baloney

Mark Cuban popularized entrepreneurship.

Shark Tank (which made Mark famous) made starting a business glamorous to attract more entrepreneurs. First off

This isn't an anti-billionaire rant.

Mark Cuban has done excellent. He's a smart, principled businessman. I enjoy his Web3 work. But Mark's work and productivity theories are absurd.

You don't need to outwork everyone in this recession to live well.

You won't be able to outwork me.

Yuck! Mark's words made me gag.

Why do boys think working is a football game where the winner wins a Super Bowl trophy? To outwork you.

Hard work doesn't equal intelligence.

Highly clever professionals spend 4 hours a day in a flow state, then go home to relax with family.

If you don't put forth the effort, someone else will.

- Mark.

He'll burn out. He's delusional and doesn't understand productivity. Boredom or disconnection spark our best thoughts.

TikTok outlaws boredom.

In a spare minute, we check our phones because we can't stand stillness.

All this work p*rn makes things worse. When is it okay to feel again? Because I can’t feel anything when I’m drowning in work and haven’t had a holiday in 2 years.

Your rivals are actively attempting to undermine you.

Ohhh please Mark…seriously.

This isn't a Tom Hanks war film. Relax. Not everyone is a rival. Only yourself is your competitor. To survive the recession, be better than a year ago.

If you get rich, great. If not, there's more to life than Lambos and angel investments.

Some want to relax and enjoy life. No competition. We witness people with lives trying to endure the recession and record-high prices.

This fictitious rival worsens life and work.

If you are truly talented, you will motivate others to work more diligently and effectively.

No Mark. Soz.

If you're a good leader, you won't brag about working hard and treating others like cogs. Treat them like humans. You'll have EQ.

Silly statements like this are caused by an out-of-control ego. No longer watch Shark Tank.

Ego over humanity.

Good leaders will urge people to keep together during the recession. Good leaders support those who are laid off and need a reference.

Not harder, quicker, better. That created my mental health problems 10 years ago.

Truth: we want to work less.

The promotion of entrepreneurship is ludicrous.

Marvel superheroes. Seriously, relax Max.

I used to write about entrepreneurship, then I quit. Many WeWork Adam Neumanns. Carelessness.

I now utilize the side hustle title when writing about online company or entrepreneurship. Humanizes.

Stop glorifying. Thinking we'll all be Elon Musks who send rockets to Mars is delusional. Most of us won't create companies employing hundreds.

OK.

The true epidemic is glorification. fewer selfies Little birdy needs less bank account screenshots. Less Uber talk.

We're exhausted.

Fun, ego-free business can transform the world. Take a relax pill.

Work as if someone were attempting to take everything from you.

I've seen people lose everything.

Myself included. My 20s startup failed. I was almost bankrupt. I thought I'd never recover. Nope.

Best thing ever.

Losing everything reveals your true self. Unintelligent entrepreneur egos perish instantly. Regaining humility revitalizes relationships.

Money's significance shifts. Stop chasing it like a puppy with a bone.

Fearing loss is unfounded.

Here is a more effective approach than outworking nobody.

(You'll thrive in the recession and become wealthy.)

Smarter work

Overworking is donkey work.

You don't want to be a career-long overworker. Instead than wasting time, write down what you do. List tasks and processes.

Keep doing/outsource the list. Step-by-step each task. Continuously systematize.

Then recruit a digital employee like Zapier or a virtual assistant in the same country.

Intelligent, not difficult.

If your big break could burn in hell, diversify like it will.

People err by focusing on one chance.

Chances can vanish. All-in risky. Instead of working like a Mark Cuban groupie, diversify your income.

If you're employed, your customer is your employer.

Sell the same abilities twice and add 2-3 contract clients. Reduce your hours at your main job and take on more clients.

Leave brand loyalty behind

Mark desires his employees' worship.

That's stupid. When times are bad, layoffs multiply. The problem is the false belief that companies care. No. A business maximizes profit and pays you the least.

To care or overpay is anti-capitalist (that run the world). Be honest.

I was a banker. Then the bat virus hit and jobs disappeared faster than I urinate after a night of drinking.

Start being disloyal now since your company will cheerfully replace you with a better applicant. Meet recruiters and hiring managers on LinkedIn. Whenever something goes wrong at work, act.

Loyalty to self and family. Nobody.

Outwork this instead

Mark doesn't suggest outworking inflation instead of people.

Inflation erodes your time on earth. If you ignore inflation, you'll work harder for less pay every minute.

Financial literacy beats inflation.

Get a side job and earn money online

So you can stop outworking everyone.

Internet leverages time. Same effort today yields exponential results later. There are still whole places not online.

Instead of working forever, generate money online.

Final Words

Overworking is stupid. Don't listen to wealthy football jocks.

Work isn't everything. Prioritize diversification, internet income streams, boredom, and financial knowledge throughout the recession.

That’s how to get wealthy rather than burnout-rich.

Restructuring and layoffs cause career reevaluation. Your career can benefit.

Once you become institutionalized, the corporate ladder is all you know.

You're bubbled. Extremists term it the corporate Matrix. I'm not so severe because the business world brainwashed me, too.

This boosted my corporate career.

Until I hit bottom.

15 months later, I view my corporate life differently. You may wish to advance professionally. Read this before you do.

Your happiness in the workplace may be deceptive.

I've been fortunate to spend time with corporate aces.

Working for 2.5 years in banking social media gave me some of these experiences. Earlier in my career, I recorded interviews with business leaders.

These people have titles like Chief General Manager and Head Of. New titles brought life-changing salaries.

They seemed happy.

I’d pass them in the hallway and they’d smile or shake my hand. I dreamt of having their life.

The ominous pattern

Unfiltered talks with some of them revealed a different world.

They acted well. They were skilled at smiling and saying the correct things. All had the same dark pattern, though.

Something felt off.

I found my conversations with them were generally for their benefit. They hoped my online antics as a writer/coach would shed light on their dilemma.

They'd tell me they wanted more. When you're one position away from CEO, it's hard not to wonder if this next move will matter.

What really displeased corporate ladder chasers

Before ascending further, consider these.

Zero autonomy

As you rise in a company, your days get busier.

Many people and initiatives need supervision. Everyone expects you to know business details. Weak when you don't. A poor leader is fired during the next restructuring and left to pursue their corporate ambition.

Full calendars leave no time for reflection. You can't have a coffee with a friend or waste a day.

You’re always on call. It’s a roll call kinda life.

Unable to express oneself freely

My 8 years of LinkedIn writing helped me meet these leaders.

I didn't think they'd care. Mistake.

Corporate leaders envied me because they wanted to talk freely again without corporate comms or a PR firm directing them what to say.

They couldn't share their flaws or inspiring experiences.

They wanted to.

Every day they were muzzled eroded by their business dream.

Limited family time

Top leaders had families.

They've climbed the corporate ladder. Nothing excellent happens overnight.

Corporate dreamers rarely saw their families.

Late meetings, customer functions, expos, training, leadership days, team days, town halls, and product demos regularly occurred after work.

Or they had to travel interstate or internationally for work events. They used bags and motel showers.

Initially, they said business class flights and hotels were nice. They'd get bored. 5-star hotels become monotonous.

No hotel beats home.

One leader said he hadn't seen his daughter much. They used to Facetime, but now that he's been gone so long, she rarely wants to talk to him.

So they iPad-parented.

You're miserable without your family.

Held captive by other job titles

Going up the business ladder seems like a battle.

Leaders compete for business gains and corporate advancement.

I saw shocking filthy tricks. Leaders would lie to seem nice.

Captives included top officials.

A different section every week. If they ran technology, the Head of Sales would argue their CRM cost millions. Or an Operations chief would battle a product team over support requests.

After one conflict, another began.

Corporate echelons are antagonistic. Huge pay and bonuses guarantee bad behavior.

Overly centered on revenue

As you rise, revenue becomes more prevalent. Most days, you'd believe revenue was everything. Here’s the problem…

Numbers drain us.

Unless you're a closet math nerd, contemplating and talking about numbers drains your creativity.

Revenue will never substitute impact.

Incapable of taking risks

Corporate success requires taking fewer risks.

Risks can cause dismissal. Risks can interrupt business. Keep things moving so you may keep getting paid your enormous salary and bonus.

Restructuring or layoffs are inevitable. All corporate climbers experience it.

On this fateful day, a small few realize the game they’ve been trapped in and escape. Most return to play for a new company, but it takes time.

Addiction keeps them trapped. You know nothing else. The rest is strange.

You start to think “I’m getting old” or “it’s nearly retirement.” So you settle yet again for the trappings of the corporate ladder game to nowhere.

Should you climb the corporate ladder?

Let me end on a surprising note.

Young people should ascend the corporate ladder. It teaches you business skills and helps support your side gig and (potential) online business.

Don't get trapped, shackled, or muzzled.

Your ideas and creativity become stifled after too much gaming play.

Corporate success won't bring happiness.

Find fulfilling employment that matters. That's it.

DeCap is DeFi's killer app.

“Software is eating the world.” Marc Andreesen, venture capitalist

DeFi. Imagine a blockchain-based alternative financial system that offers the same products and services as traditional finance, but with more variety, faster, more secure, lower cost, and simpler access.

Decentralised finance (DeFi) is a marketplace without gatekeepers or central authority managing the flow of money, where customers engage directly with smart contracts running on a blockchain.

DeFi grew exponentially in 2020/21, with Total Value Locked (an inadequate estimate for market size) topping at $100 billion. After that, it crashed.

The accumulation of funds by individuals with high discretionary income during the epidemic, the novelty of crypto trading, and the high yields given (5% APY for stablecoins on established platforms to 100%+ for risky assets) are among the primary elements explaining this exponential increase.

No longer your older brothers DeFi

Since transactions are anonymous, borrowers had to overcollateralize DeFi 1.0. To borrow $100 in stablecoins, you must deposit $150 in ETH. DeFi 1.0's business strategy raises two problems.

• Why does DeFi offer interest rates that are higher than those of the conventional financial system?;

• Why would somebody put down more cash than they intended to borrow?

Maxed out on their own resources, investors took loans to acquire more crypto; the demand for those loans raised DeFi yields, which kept crypto prices increasing; as crypto prices rose, investors made a return on their positions, allowing them to deposit more money and borrow more crypto.

This is a bull market game. DeFi 1.0's overcollateralization speculation is dead. Cryptocrash sank it.

The “speculation by overcollateralisation” world of DeFi 1.0 is dead

At a JP Morgan digital assets conference, institutional investors were more interested in DeFi than crypto or fintech. To me, that shows DeFi 2.0's institutional future.

DeFi 2.0 protocols must handle KYC/AML, tax compliance, market abuse, and cybersecurity problems to be institutional-ready.

Stablecoins gaining market share under benign regulation and more CBDCs coming online in the next couple of years could help DeFi 2.0 separate from crypto volatility.

DeFi 2.0 will have a better footing to finally decouple from crypto volatility

Then we can transition from speculation through overcollateralization to DeFi's genuine comparative advantages: cheaper transaction costs, near-instant settlement, more efficient price discovery, faster time-to-market for financial innovation, and a superior audit trail.

Akin to Amazon for financial goods

Amazon decimated brick-and-mortar shops by offering millions of things online, warehouses by keeping just-in-time inventory, and back-offices by automating invoicing and payments. Software devoured retail. DeFi will eat banking with software.

DeFi is the Amazon for financial items that will replace fintech. Even the most advanced internet brokers offer only 100 currency pairings and limited bonds, equities, and ETFs.

Old banks settlement systems and inefficient, hard-to-upgrade outdated software harm them. For advanced gamers, it's like driving an F1 vehicle on dirt.

It is like driving a F1 car on a dirt road, for the most sophisticated players

Central bankers throughout the world know how expensive and difficult it is to handle cross-border payments using the US dollar as the reserve currency, which is vulnerable to the economic cycle and geopolitical tensions.

Decentralization is the only method to deliver 24h global financial markets. DeFi 2.0 lets you buy and sell startup shares like Google or Tesla. VC funds will trade like mutual funds. Or create a bundle coverage for your car, house, and NFTs. Defi 2.0 consumes banking and creates Global Wall Street.

Defi 2.0 is how software eats banking and delivers the global Wall Street

Decentralized Capitalism is Emerging

90% of markets are digital. 10% is hardest to digitalize. That's money creation, ID, and asset tokenization.

90% of financial markets are already digital. The only problem is that the 10% left is the hardest to digitalize

Debt helped Athens construct a powerful navy that secured trade routes. Bonds financed the Renaissance's wars and supply chains. Equity fueled industrial growth. FX drove globalization's payments system. DeFi's plans:

If the 20th century was a conflict between governments and markets over economic drivers, the 21st century will be between centralized and decentralized corporate structures.

Offices vs. telecommuting. China vs. onshoring/friendshoring. Oil & gas vs. diverse energy matrix. National vs. multilateral policymaking. DAOs vs. corporations Fiat vs. crypto. TradFi vs.

An age where the network effects of the sharing economy will overtake the gains of scale of the monopolistic competition economy

This is the dawn of Decentralized Capitalism (or DeCap), an age where the network effects of the sharing economy will reach a tipping point and surpass the scale gains of the monopolistic competition economy, further eliminating inefficiencies and creating a more robust economy through better data and automation. DeFi 2.0 enables this.

DeFi needs to pay the piper now.

DeCap won't be Web3.0's Shangri-La, though. That's too much for an ailing Atlas. When push comes to shove, DeFi folks want to survive and fight another day for the revolution. If feasible, make a tidy profit.

Decentralization wasn't meant to circumvent regulation. It circumvents censorship. On-ramp, off-ramp measures (control DeFi's entry and exit points, not what happens in between) sound like a good compromise for DeFi 2.0.

The sooner authorities realize that DeFi regulation is made ex-ante by writing code and constructing smart contracts with rules, the faster DeFi 2.0 will become the more efficient and safe financial marketplace.

More crucially, we must boost system liquidity. DeFi's financial stability risks are downplayed. DeFi must improve its liquidity management if it's to become mainstream, just as banks rely on capital constraints.

This reveals the complex and, frankly, inadequate governance arrangements for DeFi protocols. They redistribute control from tokenholders to developers, which is bad governance regardless of the economic model.

But crypto can only ride the existing banking system for so long before forming its own economy. DeFi will upgrade web2.0's financial rails till then.

From a simple landing where we tested our MVP to a platform that distributes 20,000 codes per month, we learned a lot.

Starting point

Kwotet was my first startup. Everyone might post book quotes online.

I wanted a change.

Kwotet lacked attention, thus I felt stuck. After experiencing the trials of starting Kwotet, I thought of developing a waitlist service, but I required a strong co-founder.

I knew Dries from school, but we weren't close. He was an entrepreneurial programmer who worked a lot outside school. I needed this.

We brainstormed throughout school hours. We developed features to put us first. We worked until 3 am to launch this product.

Putting in the hours is KEY when building a startup

The instant that we lost our spark

In Belgium, college seniors do their internship in their last semester.

As we both made the decision to pick a quite challenging company, little time was left for Lancero.

Eventually, we lost interest. We lost the spark…

The only logical choice was to find someone with the same spark we started with to acquire Lancero.

And we did @ MicroAcquire.

Sell before your product dies. Make sure to profit from all the gains.

What did we do following the sale?

Not far from selling Lancero I lost my dad. I was about to start a new company. It was focused on positivity. I got none left at the time.

We still didn’t let go of the dream of becoming full-time entrepreneurs. As Dries launched the amazing company Plunk, and I’m still in the discovering stages of my next journey!

Dream!

You’re an entrepreneur if:

• You're imaginative.

• You enjoy disassembling and reassembling things.

• You're adept at making new friends.

• YOU HAVE DREAMS.

You don’t need to believe me if I tell you “everything is possible”… I wouldn't believe it myself if anyone told me this 2 years ago.

Until I started doing, living my dreams.

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.