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.

Great News about the new Google Business Intelligence Solution

Google has renamed Data Studio to Looker Studio and Looker Studio Pro.

Now, Google releases Looker Studio Pro. Similar to the move from Data Studio to Looker Studio, Looker Studio Pro is basically what Looker was previously, but both solutions will merge. Google says the Pro edition will acquire new enterprise management features, team collaboration capabilities, and SLAs.

In addition to Google's announcements and sales methods, additional features include:

Looker Studio assets can now have organizational ownership. Customers can link Looker Studio to a Google Cloud project and migrate existing assets once. This provides:

• Your users' created Looker Studio assets are all kept in a Google Cloud project.

• When the users who own assets leave your organization, the assets won't be removed.

• Using IAM, you may provide each Looker Studio asset in your company project-level permissions.

• Other Cloud services can access Looker Studio assets that are owned by a Google Cloud project.

Looker Studio Pro clients may now manage report and data source access at scale using team workspaces.

Google announcing these features for the pro version is fascinating. Both products will likely converge, but Google may only release many features in the premium version in the future. Microsoft with Power BI and its free and premium variants already achieves this.

Sources and Further Readings

Google, Release Notes (2022)

Google, Looker (2022)

Combining AIs creates realistic AI News Videos.

Powerful AI tools like Chat GPT-3 are trending. Have you combined AIs?

The 1-minute fake news video below is startlingly realistic. Artificial Intelligence developed NASA's Mars exploration breakthrough video (AI). However, integrating the aforementioned AIs generated it.

• AI-generated text for the Chat GPT-3 based on a succinct tagline

• DALL-E-2 AI generates an image from a brief slogan.

• Artificial intelligence-generated avatar and speech

This article shows how to use and mix the three AIs to make a realistic news video. First, watch the video (1 minute).

Talk GPT-3

Chat GPT-3 is an OpenAI NLP model. It can auto-complete text and produce conversational responses.

Try it at the playground. The AI will write a comprehensive text from a brief tagline. Let's see what the AI generates with "Breakthrough in Mars Project" as the headline.

Amazing. Our tagline matches our complete and realistic text. Fake news can start here.

DALL-E-2

OpenAI's huge transformer-based language model DALL-E-2. Its GPT-3 basis is geared for image generation. It can generate high-quality photos from a brief phrase and create artwork and images of non-existent objects.

DALL-E-2 can create a news video background. We'll use "Breakthrough in Mars project" again. Our AI creates four striking visuals. Last.

Synthesia

Synthesia lets you quickly produce videos with AI avatars and synthetic vocals.

Avatars are first. Rosie it is.

Upload and select DALL-backdrop. E-2's

Copy the Chat GPT-3 content and choose a synthetic voice.

Voice: English (US) Professional.

Finally, we generate and watch or download our video.

Synthesia AI completes the AI video.

Overview & Resources

We used three AIs to make surprisingly realistic NASA Mars breakthrough fake news in this post. Synthesia generates an avatar and a synthetic voice, therefore it may be four AIs.

These AIs created our fake news.

• AI-generated text for the Chat GPT-3 based on a succinct tagline

• DALL-E-2 AI generates an image from a brief slogan.

• Artificial intelligence-generated avatar and speech

This resulted to obimy's new audience.

Hi! obimy's official account. Here, we'll teach app developers and marketers. In 2022, our downloads increased dramatically, so we'll share what we learned.

obimy is what we call a ‘senseger’. It's a new method to communicate digitally. Instead of text, obimy users connect through senses and moods. Feeling playful? Flirt with your partner, pat a pal, or dump water on a classmate. Each feeling is an interactive animation with vibration. It's a wordless app. App Store and Google Play have obimy.

We had 20,000 users in 2022. Two to five thousand of them opened the app monthly. Our DAU metric was 500.

We have 6 million users after 6 months. 500,000 individuals use obimy daily. obimy was the top lifestyle app this week in the U.S.

And TikTok helped.

TikTok fuels obimys' growth. It's why our app exploded. How and what did we learn? Our Head of Marketing, Anastasia Avramenko, knows.

our actions prior to TikTok

We wanted to achieve product-market fit through organic expansion. Quora, Reddit, Facebook Groups, Facebook Ads, Google Ads, Apple Search Ads, and social media activity were tested. Nothing worked. Our CPI was sometimes $4, so unit economics didn't work.

We studied our markets and made audience hypotheses. We promoted our goods and studied our audience through social media quizzes. Our target demographic was Americans in long-distance relationships. I designed quizzes like Test the Strength of Your Relationship to better understand the user base. After each quiz, we encouraged users to download the app to enhance their connection and bridge the distance.

We got 1,000 responses for $50. This helped us comprehend the audience's grief and coping strategies (aka our rivals). I based action items on answers given. If you can't embrace a loved one, use obimy.

We also tried Facebook and Google ads. From the start, we knew it wouldn't work.

We were desperate to discover a free way to get more users.

Our journey to TikTok

TikTok is a great venue for emerging creators. It also helped reach people. Before obimy, my TikTok videos garnered 12 million views without sponsored promotion.

We had to act. TikTok was required.

I wasn't a TikTok user before obimy. Initially, I uploaded promotional content. Call-to-actions appear strange next to dancing challenges and my money don't jiggle jiggle. I learned TikTok. Watch TikTok for an hour was on my to-do list. What a dream job!

Our most popular movies presented the app alongside text outlining what it does. We started promoting them in Europe and the U.S. and got a 16% CTR and $1 CPI, an improvement over our previous efforts.

Somehow, we were expanding. So we came up with new hypotheses, calls to action, and content.

Four months passed, yet we saw no organic growth.

Russia attacked Ukraine.

Our app aimed to be helpful. For now, we're focusing on our Ukrainian audience. I posted sloppy TikToks illustrating how obimy can help during shelling or air raids.

In two hours, Kostia sent me our visitor count. Our servers crashed.

Initially, we had several thousand daily users. Over 200,000 users joined obimy in a week. They posted obimy videos on TikTok, drawing additional users. We've also resumed U.S. video promotion.

We gained 2,000,000 new members with less than $100 in ads, primarily in the U.S. and U.K.

TikTok helped.

The figures

We were confident we'd chosen the ideal tool for organic growth.

• Over 45 million people have viewed our own videos plus a ton of user-generated content with the hashtag #obimy.

• About 375 thousand people have liked all of our individual videos.

• The number of downloads and the virality of videos are directly correlated.

Where are we now?

TikTok fuels our organic growth. We post 56 videos every week and pay to promote viral content.

We use UGC and influencers. We worked with Universal Music Italy on Eurovision. They offered to promote us through their million-follower TikTok influencers. We thought their followers would improve our audience, but it didn't matter. Integration didn't help us. Users that share obimy videos with their followers can reach several million views, which affects our download rate.

After the dust settled, we determined our key audience was 13-18-year-olds. They want to express themselves, but it's sometimes difficult. We're searching for methods to better engage with our users. We opened a Discord server to discuss anime and video games and gather app and content feedback.

TikTok helps us test product updates and hypotheses. Example: I once thought we might raise MAU by prompting users to add strangers as friends. Instead of asking our team to construct it, I made a TikTok urging users to share invite URLs. Users share links under every video we upload, embracing people worldwide.

Key lessons

Don't direct-sell. TikTok isn't for Instagram, Facebook, or YouTube promo videos. Conventional advertisements don't fit. Most users will swipe up and watch humorous doggos.

More product videos are better. Finally. So what?

Encourage interaction. Tagging friends in comments or making videos with the app promotes it more than any marketing spend.

Be odd and risqué. A user mistakenly sent a French kiss to their mom in one of our most popular videos.

TikTok helps test hypotheses and build your user base. It also helps develop apps. In our upcoming blog, we'll guide you through obimy's design revisions based on TikTok. Follow us on Twitter, Instagram, and TikTok.

Could this happen?

Medium isn't like other social media giants. It wasn't a dot-com startup that became a multi-trillion-dollar social media firm. It launched in 2012 but didn't gain popularity until later. Now, it's one of the largest sites by web traffic, but it's still little compared to most. Most of Medium's traffic is external, but they don't run advertisements, so it's all about memberships.

Medium isn't profitable, but they don't disclose how terrible the problem is. Most of the $163 million they raised has been spent or used for acquisitions. If the money turns off, Medium can't stop paying its writers since the site dies. Writers must be paid, but they can't substantially slash payment without hurting the platform. The existing model needs scale to be viable and has a low ceiling. Facebook and other free social media platforms are struggling to retain users. Here, you must pay to appreciate it, and it's bad for writers AND readers. If I had the same Medium stats on YouTube, I'd make thousands of dollars a month.

Then what? Medium has tried to monetize by offering writers a cut of new members, but that's unsustainable. People-based growth is limited. Imagine recruiting non-Facebook users and getting them to pay to join. Some may, but I'd rather write.

Alternatives:

• Donation buttons

• Tiered subscriptions ($5, $10, $25, etc.)

• Expanding content

and these may be short-term fixes, but they're not as profitable as allowing ads. Advertisements can pay several dollars per click and cents every view. If you get 40,000 views a month like me, that's several thousand instead of a few hundred. Also, Medium would have enough money to split ad revenue with writers, who would make more. I'm among the top 6% of Medium writers. Only 6% of Medium writers make more than $100, and I made $500 with 35,000 views last month. Compared to YouTube, the top 1% of Medium authors make a lot. Mr. Beast and PewDiePie make MILLIONS a month, yet top Medium writers make tens of thousands. Sure, paying 3 or 4 people a few grand, or perhaps tens of thousands, will keep them around. What if great authors leveraged their following to go huge on YouTube and abandoned Medium? If people use Medium to get successful on other platforms, Medium will be continuously cycling through authors and paying them to stay.

Ads might make writing on Medium more profitable than making videos on YouTube because they could preserve the present freemium model and pay users based on internal views. The $5 might be ad-free.

Consider: Would you accept Medium ads? A $5 ad-free version + pay-as-you-go, etc. What are your thoughts on this?

Original post available here