900K impressions, 8K clicks, and $$$ orders…

I recently started dropshipping. Now that I own my business and can charge it as a business expense, it feels less like money wasted if it doesn't work. I also made t-shirts to sell. I intended to open a t-shirt store and had many designs on a hard drive. I read that Tiktok advertising had a high conversion rate and low cost because they were new. According to many, the advertising' cost/efficiency ratio would plummet and become as bad as Google or Facebook Ads. Now felt like the moment to try Tiktok marketing and dropshipping. I work in marketing for a SaaS firm and have seen how poorly ads perform. I wanted to try it alone.

I set up $250 and ran advertising for a week. Before that, I made my own products, store, and marketing. In this post, I'll show you my process and results.

Setting up the store

Dropshipping is a sort of retail business in which the manufacturer ships the product directly to the client through an online platform maintained by a seller. The seller takes orders but has no stock. The manufacturer handles all orders. This no-stock concept increases profitability and flexibility.

In my situation, I used previous t-shirt designs to make my own product. I didn't want to handle order fulfillment logistics, so I looked for a way to print my designs on demand, ship them, and handle order tracking/returns automatically. So I found Printful.

I needed to connect my backend and supplier to a storefront so visitors could buy. 99% of dropshippers use Shopify, but I didn't want to master the difficult application. I wanted a one-day project. I'd previously worked with Big Cartel, so I chose them.

Big Cartel doesn't collect commissions on sales, simply a monthly flat price ($9.99 to $19.99 depending on your plan).

After opening a Big Cartel account, I uploaded 21 designs and product shots, then synced each product with Printful.

Developing the ads

I mocked up my designs on cool people photographs from placeit.net, a great tool for creating product visuals when you don't have a studio, camera gear, or models to wear your t-shirts.

I opened an account on the website and had advertising visuals within 2 hours.

Because my designs are simple (black design on white t-shirt), I chose happy, stylish people on plain-colored backdrops. After that, I had to develop an animated slideshow.

Because I'm a graphic designer, I chose to use Adobe Premiere to create animated Tiktok advertising.

Premiere is a fancy video editing application used for more than advertisements. Premiere is used to edit movies, not social media marketing. I wanted this experiment to be quick, so I got 3 social media ad templates from motionarray.com and threw my visuals in. All the transitions and animations were pre-made in the files, so it only took a few hours to compile. The result:

I downloaded 3 different soundtracks for the videos to determine which would convert best.

After that, I opened a Tiktok business account, uploaded my films, and inserted ad info. They went live within one hour.

The (poor) outcomes

As a European company, I couldn't deliver ads in the US. All of my advertisements' material (title, description, and call to action) was in English, hence they continued getting rejected in Europe for countries that didn't speak English. There are a lot of them:

I lost a lot of quality traffic, but I felt that if the images were engaging, people would check out the store and buy my t-shirts. I was wrong.

• 51,071 impressions on Day 1. 0 orders after 411 clicks

• 114,053 impressions on Day 2. 1.004 clicks and no orders

• Day 3: 987 clicks, 103,685 impressions, and 0 orders

• 101,437 impressions on Day 4. 0 orders after 963 clicks

• 115,053 impressions on Day 5. 1,050 clicks and no purchases

• 125,799 impressions on day 6. 1,184 clicks, no purchases

• 115,547 impressions on Day 7. 1,050 clicks and no purchases

• 121,456 impressions on day 8. 1,083 clicks, no purchases

• 47,586 impressions on Day 9. 419 Clicks. No orders

My overall conversion rate for video advertisements was 0.9%. TikTok's paid ad formats all result in strong engagement rates (ads average 3% to 12% CTR to site), therefore a 1 to 2% CTR should have been doable.

My one-week experiment yielded 8,151 ad clicks but no sales. Even if 0.1% of those clicks converted, I should have made 8 sales. Even companies with horrible web marketing would get one download or trial sign-up for every 8,151 clicks. I knew that because my advertising were in English, I had no impressions in the main EU markets (France, Spain, Italy, Germany), and that this impacted my conversion potential. I still couldn't believe my numbers.

I dug into the statistics and found that Tiktok's stats didn't match my store traffic data.

Looking more closely at the numbers

My ads were approved on April 26 but didn't appear until April 27. My store dashboard showed 440 visitors but 1,004 clicks on Tiktok. This happens often while tracking campaign results since different platforms handle comparable user activities (click, view) differently. In online marketing, residual data won't always match across tools.

My data gap was too large. Even if half of the 1,004 persons who clicked closed their browser or left before the store site loaded, I would have gained 502 visitors. The significant difference between Tiktok clicks and Big Cartel store visits made me suspicious. It happened all week:

• Day 1: 440 store visits and 1004 ad clicks

• Day 2: 482 store visits, 987 ad clicks

• 3rd day: 963 hits on ads, 452 store visits

• 443 store visits and 1,050 ad clicks on day 4.

• Day 5: 459 store visits and 1,184 ad clicks

• Day 6: 430 store visits and 1,050 ad clicks

• Day 7: 409 store visits and 1,031 ad clicks

• Day 8: 166 store visits and 418 ad clicks

The disparity wasn't related to residual data or data processing. The disparity between visits and clicks looked regular, but I couldn't explain it.

After the campaign concluded, I discovered all my creative assets (the videos) had a 0% CTR and a $0 expenditure in a separate dashboard. Whether it's a dashboard reporting issue or a budget allocation bug, online marketers shouldn't see this.

Tiktok can present any stats they want on their dashboard, just like any other platform that runs advertisements to promote content to its users. I can't verify that 895,687 individuals saw and clicked on my ad. I invested $200 for what appears to be around 900K impressions, which is an excellent ROI. No one bought a t-shirt, even an unattractive one, out of 900K people?

Would I do it again?

Nope. Whether I didn't make sales because Tiktok inflated the dashboard numbers or because I'm horrible at producing advertising and items that sell, I’ll stick to writing content and making videos. If setting up a business and ads in a few days was all it took to make money online, everyone would do it.

Video advertisements and dropshipping aren't dead. As long as the internet exists, people will click ads and buy stuff. Converting ads and selling stuff takes a lot of work, and I want to focus on other things.

I had always wanted to try dropshipping and I’m happy I did, I just won’t stick to it because that’s not something I’m interested in getting better at.

If I want to sell t-shirts again, I'll avoid Tiktok advertisements and find another route.

This report from Mark Gurman has stuck with me:

In the News and Stocks apps, the display ads are no different than what you might get on an ad-supported website. In the App Store, the ads are for actual apps, which are probably more useful for Apple users than mortgage rates. Some people may resent Apple putting ads in the News and Stocks apps. After all, the iPhone is supposed to be a premium device. Let’s say you shelled out $1,000 or more to buy one, do you want to feel like Apple is squeezing more money out of you just to use its standard features? Now, a portion of ad revenue from the News app’s Today tab goes to publishers, but it’s not clear how much. Apple also lets publishers advertise within their stories and keep the vast majority of that money. Surprisingly, Today ads also appear if you subscribe to News+ for $10 per month (though it’s a smaller number).

I use Apple News often. It's a good general news catch-up tool, like Twitter without the BS. Customized notifications are helpful. Fast and lovely. Except for advertisements. I have Apple One, which includes News+, and while I understand why the magazines still have brand ads, it's ridiculous to me that Apple enables web publishers to introduce awful ads into this experience. Apple's junky commercials are ridiculous.

We know publishers want and probably requested this. Let's keep Apple News ad-free for the much smaller percentage of paid users, and here's your portion. (Same with Stocks, which is more sillier.)

Paid app placement in the App Store is a wonderful approach for developers to find new users (though far too many of those ads are trying to trick users, in my opinion).

Apple is also planning to increase ads in its Maps app. This sounds like Google Maps, and I don't like it. I never find these relevant, and they clutter up the user experience. Apple Maps now has a UI advantage (though not a data/search one, which matters more).

Apple is nickel-and-diming its customers. We spend thousands for their products and premium services like Apple One. We all know why: income must rise, and new firms are needed to scale. This will eventually backfire.

No wonder brands use Twitter to reach their audience. 53% of Twitter users buy new products first. 

Twitter growth does more than make your brand look popular. It helps clients trust your business. It boosts your industry standing. It shows clients, prospects, and even competitors you mean business.

How can you naturally gain Twitter followers?

• Share useful information

• Post visual content

• Tweet consistently

• Socialize

• Spread your @name everywhere.

• Use existing customers

• Promote followers

Share useful information

Twitter users join conversations and consume material. To build your followers, make sure your material appeals to them and gives value, whether it's sales, product lessons, or current events.

Use Twitter Analytics to learn what your audience likes.

Explore popular topics by utilizing relevant keywords and hashtags. Check out this post on how to use Twitter trends.

Post visual content

97% of Twitter users focus on images, so incorporating media can help your Tweets stand out. Visuals and videos make content more engaging and memorable.

Tweet often

Your audience should expect regular content updates. Plan your ideas and tweet during crucial seasons and events with a content calendar.

Socialize

Twitter connects people. Do more than tweet. Follow industry leaders. Retweet influencers, engage with thought leaders, and reply to mentions and customers to boost engagement.

Micro-influencers can promote your brand or items. They can help you gain new audiences' trust.

Spread your @name everywhere.

Maximize brand exposure. Add a follow button on your website, link to it in your email signature and newsletters, and promote it on business cards or menus.

Use existing customers

Emails can be used to find existing Twitter clients. Upload your email contacts and follow your customers on Twitter to start a dialogue.

Promote followers

Run a followers campaign to boost your organic growth. Followers campaigns promote your account to a particular demographic, and you only pay when someone follows you.

Consider short campaigns to enhance momentum or an always-on campaign to gain new followers.

Increasing your brand's Twitter followers takes effort and experimentation, but the payback is huge.

👋 Follow me on twitter

This post is for you if you can't read or study for 5 minutes.

If you clicked this post, you may be experiencing problems focusing on tasks. A few minutes of reading may tire you. Easily distracted? Using social media and video games for hours without being sidetracked may impair your dopamine system.

When we achieve a goal, the brain secretes dopamine. It might be as simple as drinking water or as crucial as college admission. Situations vary. Various events require different amounts.

Dopamine is released when we start learning but declines over time. Social media algorithms provide new material continually, making us happy. Social media use slows down the system. We can't continue without an award. We return to social media and dopamine rewards.

Mice were given a button that released dopamine into their brains to study the hormone. The mice lost their hunger, thirst, and libido and kept pressing the button. Think this is like someone who spends all day gaming or on Instagram?

When we cause our brain to release so much dopamine, the brain tries to balance it in 2 ways:

1- Decreases dopamine production

2- Dopamine cannot reach its target.

Too many quick joys aren't enough. We'll want more joys. Drugs and alcohol are similar. Initially, a beer will get you drunk. After a while, 3-4 beers will get you drunk.

Social media is continually changing. Updates to these platforms keep us interested. When social media conditions us, we can't read a book.

Same here. I used to complete a book in a day and work longer without distraction. Now I'm addicted to Instagram. Daily, I spend 2 hours on social media. This must change. My life needs improvement. So I started the 50-day challenge.

I've compiled three dopamine-related methods.

Recommendations:

1. Day-long dopamine detox

First, take a day off from all your favorite things. Social media, gaming, music, junk food, fast food, smoking, alcohol, friends. Take a break.

Hanging out with friends or listening to music may seem pointless. Our minds are polluted. One day away from our pleasures can refresh us.

2. One-week dopamine detox by selecting

Choose one or more things to avoid. Social media, gaming, music, junk food, fast food, smoking, alcohol, friends. Try a week without Instagram or Twitter. I use this occasionally.

1. One week all together

One solid detox week. It's the hardest program. First or second options are best for dopamine detox. Time will help you.

You can walk, read, or pray during a dopamine detox. Many options exist. If you want to succeed, you must avoid instant gratification. Success after hard work is priceless.

Elon Musk, Matt Taibbi, the 'Twitter Files,' and Hunter Biden's laptop: what gives?

Explaining Twitter Files

Matt Taibbi released "The Twitter Files," a batch of emails sent by Twitter executives discussing the company's decision to stop an October 2020 New York Post story online.

What's on Twitter? New York Post and Fox News call them "bombshell" documents. Or, as a Post columnist admitted, are they "not the smoking gun"? Onward!

What started this?

The New York Post published an exclusive, potentially explosive story in October 2020: Biden's Secret Emails: Ukrainian executive thanks Hunter Biden for'meeting' veep dad. The story purported to report the contents of a laptop brought to the tabloid by a Delaware computer repair shop owner who said it belonged to President Biden's second son, Hunter Biden. Emails and files on the laptop allegedly showed how Hunter peddled influence with Ukranian businessmen and included a "raunchy 12-minute video" of Hunter smoking crack and having sex.

Twitter banned links to the Post story after it was published, calling it "hacked material." The Post's Twitter account was suspended for multiple days.

Why? Yoel Roth, Twitter's former head of trust and safety, said the company couldn't verify the story, implying they didn't trust the Post.

Twitter's stated purpose rarely includes verifying news stories. This seemed like intentional political interference. This story was hard to verify because the people who claimed to have found the laptop wouldn't give it to other newspapers. (Much of the story, including Hunter's business dealings in Ukraine and China, was later confirmed.)

Roth: "It looked like a hack and leak."

So what are the “Twitter Files?”

Twitter's decision to bury the story became a political scandal, and new CEO Elon Musk promised an explanation. The Twitter Files, named after Facebook leaks.

Musk promised exclusive details of "what really happened" with Hunter Biden late Friday afternoon. The tweet was punctuated with a popcorn emoji.

Explaining Twitter Files

Three hours later, journalist Matt Taibbi tweeted more than three dozen tweets based on internal Twitter documents that revealed "a Frankensteinian tale of a human-built mechanism grown out of its designer's control."

Musk sees this release as a way to shape Twitter's public perception and internal culture in his image. We don't know if the CEO gave Taibbi the documents. Musk hyped the document dump before and during publication, but Taibbi cited "internal sources."

Taibbi shares email screenshots showing Twitter execs discussing the Post story and blocking its distribution. Taibbi says the emails show Twitter's "extraordinary steps" to bury the story.

Twitter communications chief Brandon Borrman has the most damning quote in the Files. Can we say this is policy? The story seemed unbelievable. It seemed like a hack... or not? Could Twitter, which ex-CEO Dick Costolo called "the free speech wing of the free speech party," censor a news story?

Many on the right say the Twitter Files prove the company acted at the behest of Democrats. Both parties had these tools, writes Taibbi. In 2020, both the Trump White House and Biden campaign made requests. He says the system for reporting tweets for deletion is unbalanced because Twitter employees' political donations favor Democrats. Perhaps. These donations may have helped Democrats connect with Twitter staff, but it's also possible they didn't. No emails in Taibbi's cache show these alleged illicit relations or any actions Twitter employees took as a result.

Even Musk's supporters were surprised by the drop. Miranda Devine of the New York Post told Tucker Carlson the documents weren't "the smoking gun we'd hoped for." Sebastian Gorka said on Truth Social, "So far, I'm deeply underwhelmed." DC Democrats collude with Palo Alto Democrats. Whoop!” The Washington Free Beacon's Joe Simonson said the Twitter files are "underwhelming." Twitter was staffed by Democrats who did their bidding. (Why?)

If "The Twitter Files" matter, why?

These emails led Twitter to suppress the Hunter Biden laptop story has real news value. It's rare for a large and valuable company like Twitter to address wrongdoing so thoroughly. Emails resemble FOIA documents. They describe internal drama at a company with government-level power. Katie Notopoulos tweeted, "Any news outlet would've loved this scoop!" It's not a'scandal' as teased."

Twitter's new owner calls it "the de facto public town square," implying public accountability. Like a government agency. Though it's exciting to receive once-hidden documents in response to a FOIA, they may be boring and tell you nothing new. Like Twitter files. We learned how Twitter blocked the Post's story, but not why. Before these documents were released, we knew Twitter had suppressed the story and who was involved.

These people were disciplined and left Twitter. Musk fired Vijaya Gadde, the former CLO who reportedly played a "key role" in the decision. Roth quit over Musk's "dictatorship." Musk arrived after Borrman left. Jack Dorsey, then-CEO, has left. Did those who digitally quarantined the Post's story favor Joe Biden and the Democrats? Republican Party opposition and Trump hatred? New York Post distaste? According to our documents, no. Was there political and press interference? True. We knew.

Taibbi interviewed anonymous ex-Twitter employees about the decision; all expressed shock and outrage. One source said, "Everyone knew this was fucked." Since Taibbi doesn't quote that expletive, we can assume the leaked emails contained few or no sensational quotes. These executives said little to support nefarious claims.

Outlets more invested in the Hunter Biden story than Gizmodo seem vexed by the release and muted headlines. The New York Post, which has never shied away from a blaring headline in its 221-year history, owns the story of Hunter Biden's laptop. Two Friday-night Post alerts about Musk's actions were restrained. Elon Musk will drop Twitter files on NY Post-Hunter Biden laptop censorship today. Elon Musk's Twitter dropped Post censorship details from Biden's laptop. Fox News' Apple News push alert read, "Elon Musk drops Twitter censorship documents."

Bombshell, bombshell, bombshell… what, exactly, is the bombshell? Maybe we've heard this story too much and are missing the big picture. Maybe these documents detail a well-documented decision.

The Post explains why on its website. "Hunter Biden laptop bombshell: Twitter invented reason to censor Post's reporting," its headline says.

Twitter's ad hoc decision to moderate a tabloid's content is not surprising. The social network had done this for years as it battled toxic users—violent white nationalists, virulent transphobes, harassers and bullies of all political stripes, etc. No matter how much Musk crows, the company never had content moderation under control. Buzzfeed's 2016 investigation showed how Twitter has struggled with abusive posters since 2006. Jack Dorsey and his executives improvised, like Musk.

Did the US government interfere with the ex-social VP's media company? That's shocking, a bombshell. Musk said Friday, "Twitter suppressing free speech by itself is not a 1st amendment violation, but acting under government orders with no judicial review is." Indeed! Taibbi believed this. August 2022: "The laptop is secondary." Zeynep Tufecki, a Columbia professor and New York Times columnist, says the FBI is cutting true story distribution. Taibbi retracted the claim Friday night: "I've seen no evidence of government involvement in the laptop story."

What’s the bottom line?

I'm still not sure what's at stake in the Hunter Biden scandal after dozens of New York Post articles, hundreds of hours of Fox News airtime, and thousands of tweets. Briefly: Joe Biden's son left his laptop with a questionable repairman. FBI confiscated it? The repairman made a copy and gave it to Rudy Giuliani's lawyer. The Post got it from Steve Bannon. On that laptop were videos of Hunter Biden smoking crack, cavorting with prostitutes, and emails about introducing his father to a Ukrainian businessman for $50,000 a month. Joe Biden urged Ukraine to fire a prosecutor investigating the company. What? The story seems to be about Biden family business dealings, right?

The discussion has moved past that point anyway. Now, the story is the censorship of it. Adrienne Rich wrote in "Diving Into the Wreck" that she came for "the wreck and not the story of the wreck" No matter how far we go, Hunter Biden's laptop is done. Now, the crash's story matters.

I'm dizzy. Katherine Miller of BuzzFeed wrote, "I know who I believe, and you probably do, too. To believe one is to disbelieve the other, which implicates us in the decision; we're stuck." I'm stuck. Hunter Biden's laptop is a political fabrication. You choose. I've decided.

This could change. Twitter Files drama continues. Taibbi said, "Much more to come." I'm dizzy.

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.