I've worked in a People role at seven early-stage firms for over 15 years (I enjoy chasing a dream!). Few of the seven achieved profitability, including unicorn status or acquisition.

Did early-stage startups share anything? Was there a difference between winners and losers? YES.

I support founders and entrepreneurs building financially sustainable enterprises with a compelling cause. This isn't something everyone would do. A company's success demands more than guts. Founders drive startup success.

Six Qualities of Successful Startups

Successful startup founders either innately grasped the correlation between strong team engagement and a well-executed business model, or they knew how to ask and listen to others (executive coaches, other company leaders, the team itself) to learn about it.

Successful startups:

1. Co-founders agreed and got along personally.

Multi-founder startups are common. When co-founders agree on strategic decisions and are buddies, there's less friction and politics at work.

As a co-founder, ask your team if you're aligned. They'll explain.

I've seen C-level leaders harbor personal resentments over disagreements. A co-departure founder's caused volatile leadership and work disruptions that the team struggled to manage during and after.

2. Team stayed.

Successful startups have low turnover. Nobody is leaving. There may be a termination for performance, but other team members will have observed the issues and agreed with the decision.

You don't want organizational turnover of 30%+, with leaders citing performance issues but the team not believing them. This breeds suspicion.

Something is wrong if many employees leave voluntarily or involuntarily. You may hear about lack of empowerment, support, or toxic leadership in exit interviews and from the existing team. Intellectual capital loss and resource instability harm success.

3. Team momentum.

A successful startup's team is excited about its progress. Consistently achieving goals and having trackable performance metrics. Some describe this period of productivity as magical, with great talents joining the team and the right people in the right places. Increasing momentum.

I've also seen short-sighted decisions where only some departments, like sales and engineering, had goals. Lack of a unified goals system created silos and miscommunication. Some employees felt apathetic because they didn't know how they contributed to team goals.

4. Employees advanced in their careers.

Even if you haven't created career pathing or professional development programs, early-stage employees will grow and move into next-level roles. If you hire more experienced talent and leaders, expect them to mentor existing team members. Growing companies need good performers.

New talent shouldn't replace and discard existing talent. This creates animosity and makes existing employees feel unappreciated for their early contributions to the company.

5. The company lived its values.

Culture and identity are built on lived values. A company's values affect hiring, performance management, rewards, and other processes. Identify, practice, and believe in company values. Starting with team values instead of management or consultants helps achieve this. When a company's words and actions match, it builds trust.

When company values are beautifully displayed on a wall but few employees understand them, the opposite is true. If an employee can't name the company values, they're useless.

6. Communication was clear.

When necessary information is shared with the team, they feel included, trusted, and like owners. Transparency means employees have the needed information to do their jobs. Disclosure builds trust. The founders answer employees' questions honestly.

Information accessibility decreases office politics. Without transparency, even basic information is guarded and many decisions are made in secret. I've seen founders who don't share financial, board meeting, or compensation and equity information. The founders' lack of trust in the team wasn't surprising, so it was reciprocated.

The Choices

Finally. All six of the above traits (leadership alignment, minimal turnover, momentum, professional advancement, values, and transparency) were high in the profitable startups I've worked at, including unicorn status or acquisition.

I've seen these as the most common and constant signals of startup success or failure.

These characteristics are the product of founders' choices. These decisions lead to increased team engagement and business execution.

Here's something to consider for startup employees and want-to-bes. 90% of startups fail, despite the allure of building something new and gaining ownership. With the emotional and time investment in startup formation, look for startups with these traits to reduce your risk.

Both you and the startup will thrive in these workplaces.

Ali's $1,054,053.74 YouTube Adsense haul.

YouTuber, entrepreneur, and former doctor Ali Abdaal. He began filming productivity and financial videos in 2017. Ali Abdaal has 3 million YouTube subscribers and has crossed $1 million in AdSense revenue. Crazy, no?

Ali will share the revenue of his top 5 youtube videos, things he's learned that you can apply to your side hustle, and how many views it takes to make a livelihood off youtube.

First, "The Long Game."

All good things take time to bear fruit. Compounding improves everything. Long-term work yields better returns. Ali made his first dollar after nine months and 85 videos.

Second, "One piece of content can transform your life, but you never know which one."

Had he abandoned YouTube at 84 videos without making any money, he wouldn't have filmed the 85th video that altered everything.

Third Lesson: Your Industry Choice Can Multiply.

The industry or niche you target as a business owner or side hustler can have a major impact on how much money you make.

Here are the top 5 videos.

1) 9.8m views: $191,258.16 for 9 passive income ideas

Ali made 2 points.

We should consider YouTube videos digital assets. They're investments, which make us money. His investments are yielding passive income.

Investing extra time and effort in your films can pay off.

2) How to Invest for Beginners — 5.2m Views: $87,200.08.

This video did poorly in the first several weeks after it was published; it was his tenth poorest performer. Don't worry about things you can't control. This applies to life, not just YouTube videos.

He stated we constantly have anxieties, fears, and concerns about things outside our control, but if we can find that line, life is easier and more pleasurable.

3) How to Build a Website in 2022— 866.3k views: $42,132.72.

The RPM was $48.86 per thousand views, making it his highest-earning video. Squarespace, Wix, and other website builders are trying to put ads on it and competing against one other, so ad rates go up.

Because it was beyond his niche, Ali almost didn't make the video. He made the video because he wanted to help at least one person.

4) How I take notes on my iPad in medical school — 5.9m views: $24,479.80

85th video. It's the video that affected Ali's YouTube channel and his life the most. The video's success wasn't certain.

5) How I Type Fast 156 Words Per Minute — 8.2M views: $25,143.17

Ali didn't know this video would perform well; he made it because he can type fast and has been practicing for 10 years. So he made a video with his best advice.

How many views to different wealth levels?

It depends on geography, niche, and other monetization sources. To keep things simple, he would solely utilize AdSense.

How many views to generate money?

To generate money on Youtube, you need 1,000 subscribers and 4,000 hours of view time. How much work do you need to make pocket money?

Ali's first 1,000 subscribers took 52 videos and 6 months. The typical channel with 1,000 subscribers contains 152 videos, according to Tubebuddy. It's time-consuming.

After monetizing, you'll need 15,000 views/month to make $5-$10/day.

How many views to go part-time?

Say you make $35,000/year at your day job. If you work 5 days/week, you make $7,000/year each day. If you want to drop down from 5 days to 4 days/week, you need to make an extra $7,000/year from YouTube, or $600/month.

What's the quit-your-job budget?

Silicon Valley Girl is in a highly successful niche targeting tech-focused folks in the west. When her channel had 500k views/month, she made roughly $3,000/month or $47,000/year, enough to quit your work.

Marina has another 1.5m subscriber channel in Russia, which has a lower rpm because fewer corporations advertise there than in the west. 2.3 million views/month is $4,000/month or $50,000/year, enough to quit your employment.

Marina is an intriguing example because she has three YouTube channels with the same skills, but one is 16x more profitable due to the niche she chose.

In Ali's case, he made 100+ videos when his channel was producing enough money to quit his job, roughly $4,000/month.

How many views make you rich?

Depending on how you define rich. Ali felt prosperous with over $100,000/year and 3–5m views/month.

Conclusion

YouTubers and artists don't treat their work like a company, which is a mistake. Businesses have been attempting to figure this out for decades, if not centuries.

We can learn from the business world how to monetize YouTube, Instagram, and Tiktok and make them into sustainable enterprises where we can hire people and delegate tasks.

Bonus

Watch Ali's video explaining all this:

This post is a summary. Read the full article here

Internet's fantastic, right?

We've never had a better way to share our creativity.

I can now draw on my iPad and tweet or Instagram it to thousands. I may get some likes.

With such a great, free tool, you're not alone.

Millions more bright-eyed artists are sharing their work online.

The issue is getting innovative work noticed, not sharing it.

In a world where creators want attention, attention is valuable.

We build for attention.

Attention helps us establish a following, make money, get notoriety, and make a difference.

Most of us require attention to stay sane while creating wonderful things.

I know how hard it is to work hard and receive little views.

How do we receive more attention, more often, in a sea of talent?

Advertising and celebrity endorsements are options. These may work temporarily.

To attract true, organic, and long-term attention, you must create in high quality, high volume, and consistency.

Adapting Steve Martin's Be so amazing, they can't ignore you (with a mention to Dan Norris in his great book Create or Hate for the reminder)

Create a lot.

Eventually, your effort will gain traction.

Traction shows your work's influence.

Traction is when your product sells more. Traction is exponential user growth. Your work is shared more.

No matter how good your work is, it will always have minimal impact on the world.

Your work can eventually dent or puncture. Daily, people work to dent.

To achieve this tipping point, you must consistently produce exceptional work.

Expect traction after hundreds of outputs.

Dilbert creator Scott Adams says repetition persuades. If you don't stop, you can persuade practically anyone with anything.

Volume lends believability. So make more.

I worked as an illustrator for at least a year and a half without any recognition. After 150 illustrations on iStockphoto, my work started selling.

With 350 illustrations on iStock, I started getting decent client commissions.

Producing often will improve your craft and draw attention.

It's the only way to succeed. More creation means better results and greater attention.

Austin Kleon says you can improve your skill in relative anonymity before you become famous. Before obtaining traction, generate a lot and become excellent.

Most artists, even excellent ones, don't create consistently enough to get traction.

It may hurt. For makers who don't love and flow with their work, it's extremely difficult.

Your work must bring you to life.

To generate so much that others can't ignore you, decide what you'll accomplish every day (or most days).

Commit and be patient.

Prepare for zero-traction.

Anticipating this will help you persevere and create.

My online guru Grant Cardone says: Anything worth doing is worth doing every day.

Do.

On January 1, "Public Domain Day," thousands of creative works immediately join the public domain. The original creator or copyright holder loses exclusive rights to reproduce, adapt, or publish the work, and anybody can use it. It happens with movies, poems, music, artworks, books (where creative rights endure 70 years beyond the author's death), and sometimes source code.

Public domain creative works open the door to new uses. 400,000 sound recordings from before 1923, including Winnie-the-Pooh, were released this year.  With most of A.A. Milne's 1926 Winnie-the-Pooh characters now available, we're seeing innovative interpretations Milne likely never planned. The ancient hyphenated version of the honey-loving bear is being adapted for a horror movie: "Winnie-the-Pooh: Blood and Honey"... with Pooh and Piglet as the baddies.

Counterintuitively, experimenting and recombination can occasionally increase IP value. Open source movements allow the public to build on (or fork and duplicate) existing technologies. Permissionless innovation helps Android, Linux, and other open source software projects compete. Crypto's success at attracting public development is also due to its support of open source and "remix culture," notably in NFT forums.

Production memes

NFT projects use several IP strategies to establish brands, communities, and content. Some preserve regular IP protections; others offer NFT owners the opportunity to innovate on connected IP; yet others have removed copyright and other IP safeguards.

By using the "Creative Commons Zero" (cc0) license, artists can intentionally select for "no rights reserved." This option permits anyone to benefit from derivative works without legal repercussions. There's still a lot of confusion between copyrights and NFTs, so nothing here should be considered legal, financial, tax, or investment advice. Check out this post for an overview of copyright vulnerabilities with NFTs and how authors can protect owners' rights. This article focuses on cc0.

Nouns, a 2021 project, popularized cc0 for NFTs. Others followed, including: A Common Place, Anonymice, Blitmap, Chain Runners, Cryptoadz, CryptoTeddies, Goblintown, Gradis, Loot, mfers, Mirakai, Shields, and Terrarium Club are cc0 projects.

Popular crypto artist XCOPY licensed their 1-of-1 NFT artwork "Right-click and Save As Guy" under cc0 in January, exactly one month after selling it. cc0 has spawned many derivatives.

"Right-click Save As Guy" by XCOPY (1)/derivative works (2)

XCOPY said Monday he would apply cc0 to "all his existing art." "We haven't seen a cc0 summer yet, but I think it's approaching," said the artist. - predicting a "DeFi summer" in 2020, when decentralized finance gained popularity.

Why do so many NFT authors choose "no rights"?

Promoting expansions of the original project to create a more lively and active community is one rationale. This makes sense in crypto, where many value open sharing and establishing community.

Creativity depends on cultural significance. NFTs may allow verifiable ownership of any digital asset, regardless of license, but cc0 jumpstarts "meme-ability" by actively, not passively, inviting derivative works. As new derivatives are made and shared, attention might flow back to the original, boosting its reputation. This may inspire new interpretations, leading in a flywheel effect where each derivative adds to the original's worth - similar to platform network effects, where platforms become more valuable as more users join them.

cc0 licence allows creators "seize production memes."

"SEASON 1 MEME CARD 2"

Physical items are also using cc0 NFT assets, thus it's not just a digital phenomenon. The Nouns Vision initiative turned the square-framed spectacles shown on each new NounsDAO NFT ("one per day, forever") into luxury sunglasses. Blitmap's pixel-art has been used on shoes, apparel, and caps. In traditional IP regimes, a single owner controls creation, licensing, and production.

The physical "blitcap" (3rd level) is a descendant of the trait in the cc0 Chain Runners collection (2nd), which uses the "logo" from cc0 Blitmap (1st)! The Logo is Blitmap token #84 and has been used as a trait in various collections. The "Dom Rose" is another popular token. These homages reference Blitmap's influence as a cc0 leader, as one of the earliest NFT projects to proclaim public domain intents. A new collection, Citizens of Tajigen, emerged last week with a Blitcap characteristic.

These derivatives can be a win-win for everyone, not just the original inventors, especially when using NFT assets to establish unique brands. As people learn about the derivative, they may become interested in the original. If you see someone wearing Nouns glasses on the street (or in a Super Bowl ad), you may desire a pair, but you may also be interested in buying an original NounsDAO NFT or related derivative.

Blitmap Logo Hat (1), Chain Runners #780 ft. Hat (2), and Blitmap Original "Logo #87" (3)

Co-creating open source

NFTs' power comes from smart contract technology's intrinsic composability. Many smart contracts can be integrated or stacked to generate richer applications.

"Money Legos" describes how decentralized finance ("DeFi") smart contracts interconnect to generate new financial use cases. Yearn communicates with MakerDAO's stablecoin $DAI and exchange liquidity provider Curve by calling public smart contract methods. NFTs and their underlying smart contracts can operate as the base-layer framework for recombining and interconnecting culture and creativity.

cc0 gives an NFT's enthusiast community authority to develop new value layers whenever, wherever, and however they wish.

Multiple cc0 projects are playable characters in HyperLoot, a Loot Project knockoff.

Open source and Linux's rise are parallels. When the internet was young, Microsoft dominated the OS market with Windows. Linux (and its developer Linus Torvalds) championed a community-first mentality, freely available the source code without restrictions. This led to developers worldwide producing new software for Linux, from web servers to databases. As people (and organizations) created world-class open source software, Linux's value proposition grew, leading to explosive development and industry innovation. According to Truelist, Linux powers 96.3% of the top 1 million web servers and 85% of smartphones.

With cc0 licensing empowering NFT community builders, one might hope for long-term innovation. Combining cc0 with NFTs "turns an antagonistic game into a co-operative one," says NounsDAO cofounder punk4156. It's important on several levels. First, decentralized systems from open source to crypto are about trust and coordination, therefore facilitating cooperation is crucial. Second, the dynamics of this cooperation work well in the context of NFTs because giving people ownership over their digital assets allows them to internalize the results of co-creation through the value that accrues to their assets and contributions, which incentivizes them to participate in co-creation in the first place.

Licensed to create

If cc0 projects are open source "applications" or "platforms," then NFT artwork, metadata, and smart contracts provide the "user interface" and the underlying blockchain (e.g., Ethereum) is the "operating system." For these apps to attain Linux-like potential, more infrastructure services must be established and made available so people may take advantage of cc0's remixing capabilities.

These services are developing. Zora protocol and OpenSea's open source Seaport protocol enable open, permissionless NFT marketplaces. A pixel-art-rendering engine was just published on-chain to the Ethereum blockchain and integrated into OKPC and ICE64. Each application improves blockchain's "out-of-the-box" capabilities, leading to new apps created from the improved building blocks.

Web3 developer growth is at an all-time high, yet it's still a small fraction of active software developers globally. As additional developers enter the field, prospective NFT projects may find more creative and infrastructure Legos for cc0 and beyond.

Electric Capital Developer Report (2021), p. 122

Growth requires composability. Users can easily integrate digital assets developed on public standards and compatible infrastructure into other platforms. The Loot Project is one of the first to illustrate decentralized co-creation, worldbuilding, and more in NFTs. This example was low-fi or "incomplete" aesthetically, providing room for imagination and community co-creation.

Loot began with a series of Loot bag NFTs, each listing eight "adventure things" in white writing on a black backdrop (such as Loot Bag #5726's "Katana, Divine Robe, Great Helm, Wool Sash, Divine Slippers, Chain Gloves, Amulet, Gold Ring"). Dom Hofmann's free Loot bags served as a foundation for the community.

Several projects have begun metaphorical (lore) and practical (game development) world-building in a short time, with artists contributing many variations to the collective "Lootverse." They've produced games (Realms & The Crypt), characters (Genesis Project, Hyperloot, Loot Explorers), storytelling initiatives (Banners, OpenQuill), and even infrastructure (The Rift).

Why cc0 and composability? Because consumers own and control Loot bags, they may use them wherever they choose by connecting their crypto wallets. This allows users to participate in multiple derivative projects, such as  Genesis Adventurers, whose characters appear in many others — creating a decentralized franchise not owned by any one corporation.

Genesis Project's Genesis Adventurer (1) with HyperLoot (2) and Loot Explorer (3) versions

When to go cc0

There are several IP development strategies NFT projects can use. When it comes to cc0, it’s important to be realistic. The public domain won't make a project a runaway success just by implementing the license. cc0 works well for NFT initiatives that can develop a rich, enlarged ecosystem.

Many of the most successful cc0 projects have introduced flexible intellectual property. The Nouns brand is as obvious for a beer ad as for real glasses; Loot bags are simple primitives that make sense in all adventure settings; and the Goblintown visual style looks good on dwarfs, zombies, and cranky owls as it does on Val Kilmer.

The ideal cc0 NFT project gives builders the opportunity to add value:

• vertically, by stacking new content and features directly on top of the original cc0 assets (for instance, as with games built on the Loot ecosystem, among others), and

• horizontally, by introducing distinct but related intellectual property that helps propagate the original cc0 project’s brand (as with various Goblintown derivatives, among others).

These actions can assist cc0 NFT business models. Because cc0 NFT projects receive royalties from secondary sales, third-party extensions and derivatives can boost demand for the original assets.

Using cc0 license lowers friction that could hinder brand-reinforcing extensions or lead to them bypassing the original. Robbie Broome recently argued (in the context of his cc0 project A Common Place) that giving away his IP to cc0 avoids bad rehashes down the line. If UrbanOutfitters wanted to put my design on a tee, they could use the actual work instead of hiring a designer. CC0 can turn competition into cooperation.

Community agreement about core assets' value and contribution can help cc0 projects. Cohesion and engagement are key. Using the above examples: Developers can design adventure games around whatever themes and item concepts they desire, but many choose Loot bags because of the Lootverse's community togetherness. Flipmap shared half of its money with the original Blitmap artists in acknowledgment of that project's core role in the community. This can build a healthy culture within a cc0 project ecosystem. Commentator NiftyPins said it was smart to acknowledge the people that constructed their universe. Many OG Blitmap artists have popped into the Flipmap discord to share information.

cc0 isn't a one-size-fits-all answer; NFTs formed around well-established brands may prefer more restrictive licenses to preserve their intellectual property and reinforce exclusivity. cc0 has some superficial similarities to permitting NFT owners to market the IP connected with their NFTs (à la Bored Ape Yacht Club), but there is a significant difference: cc0 holders can't exclude others from utilizing the same IP. This can make it tougher for holders to develop commercial brands on cc0 assets or offer specific rights to partners. Holders can still introduce enlarged intellectual property (such as backstories or derivatives) that they control.

Blockchain technologies and the crypto ethos are decentralized and open-source. This makes it logical for crypto initiatives to build around cc0 content models, which build on the work of the Creative Commons foundation and numerous open source pioneers.

NFT creators that choose cc0 must select how involved they want to be in building the ecosystem. Some cc0 project leaders, like Chain Runners' developers, have kept building on top of the initial cc0 assets, creating an environment derivative projects can plug into. Dom Hofmann stood back from Loot, letting the community lead. (Dom is also working on additional cc0 NFT projects for the company he formed to build Blitmap.) Other authors have chosen out totally, like sartoshi, who announced his exit from the cc0 project he founded, mfers, and from the NFT area by publishing a final edition suitably named "end of sartoshi" and then deactivating his Twitter account. A multi-signature wallet of seven mfers controls the project's smart contract. 

cc0 licensing allows a robust community to co-create in ways that benefit all members, regardless of original creators' continuous commitment. We foresee more organized infrastructure and design patterns as NFT matures. Like open source software, value capture frameworks may see innovation. (We could imagine a variant of the "Sleepycat license," which requires commercial software to pay licensing fees when embedding open source components.) As creators progress the space, we expect them to build unique rights and licensing strategies. cc0 allows NFT producers to bootstrap ideas that may take off.

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.