AI Weekly Digest #23: Dalle3, Midjourney, Stable Diffusion and Adobe Firefly

Hello, tech enthusiasts! This is Wassim Jouini and Welcome to my AI newsletter, where I bring you the latest advancements in Artificial Intelligence without the unnecessary hype.

You can find me on LinkedIn, Twitter and Medium! Let’s connect!

Now let's dive into this week's news and explore the practical applications of AI across various sectors.

Main Headlines

Midjourney, Stable Diffusion, Dalle3, Adobe Firefly? Confused about the different models out there? Let’s look at their differences and how to choose the right one for you!

In my previous newsletter, I mentioned the release of Dalle3. It wasn’t available to everyone yet a week ago, and that has changed!

#1 Dalle3 - Access & First Impressions

#1.1 Activate it: Dalle3 is available to all ChatGPT Plus subscribers. To use it, simply activate it on the interface (see screenshot)

#1.2 Generate your first image

What I love about Dalle3 is that you don’t need to think about the prompt to get started ; simply write the few important elements you’d like to see in the image. ChatGPT will reformulate your sentence into vibrant scenes before generating the images through Dalle3. It is also particularly good at embedding text within the generated images!

Here’s a example!

• Enter a simple prompt, e.g., "I love AI" 3D halloween vibe in paris cartoon

• Let ChatGPT imagine the scene: 3D cartoon representing a haunted Halloween in Paris. The Eiffel Tower dominates the skyline, emanating a ghostly luminescence. Streets are festooned with cobwebs and adorned with artistically carved pumpkins. A ghostly figure drifts above the Seine, presenting a sign that boasts 'I love AI'. The presence of bats swirling in the sky deepens the chilling effect.

• Then Dalle3 generates the image!

#2 Yet Another Image Generation Tool? Which One Should you Choose?

Stable Diffusion, Midjourney, Dalle3, and Adobe Firefly are all amazing tools, but each excels in different scenarios. So the question isn't "Which one is the best model?" but rather, "Which one is best suited for your application?" Let's dive into it!

2.1 Open Source, Fine-Tuning and API Access? Stable Diffusion

Stable Diffusion stands out as the only Open Source model among the ones listed. With access to the model's checkpoints, users have the flexibility to fine-tune it to acquire new concepts and styles, including the ability to generate images in any desired style.

Such fine-tuning has produced remarkable results even with the early versions of Stable Diffusion, such as v1.5 and v2.1. A testament to this is Civitai, which boasts a vast library of models that have been fine-tuned based on Stable Diffusion.

Additionally, Stable Diffusion models are the only ones accessible via APIs. This opens up the avenue for developers to seamlessly integrate image generation into their products or to automate the image generation process.

The latest and most extensive model from this lineage is named SDXL. It is on par with Midjourney v4 in terms of capabilities (Midjourney is at v5.2 as of today). For those interested in trying it out, it is available for free testing at https://clipdrop.co/stable-diffusion.

2.2 Best Realistic Images? Midjourney

We’ve all seen this image: “Pope Francis wearing a long, white puffer jacket inspired by Balenciaga.” (source). This image was generated by Midjourney!

Midjourney v5.2 offers state-of-the-art realism today, with impeccable skin, teeth, and hand features. It remains the go-to solution for realistic portraits or images of any kind.

2.3 Vibrant Illustrations with Text? Dalle3

As a already mentioned, Dalle3 allows users to simply provide key elements and ChatGPT will transform them into vibrant scenes, with the added capability of embedding text within the generated images. Here are a couple more examples!

2.4 Image & Video Editing? Adobe Firefly

This week, Adobe announed it’s next-gen AI-powered tools! And they significantly stand out by enabling image and video editing, as well as vector image generation, a feature not commonly found in others like MidJourney, Stable Diffusion, and DALL-E. Specifically, Adobe Illustrator now has a Text to Vector Graphic feature, allowing easy creation of editable vector graphics from text prompts, enriching creative workflows​

#3 Image to Video generation is coming along nicely - Meet Gen2 by RunwayML

Upload an image and let the AI animate it. This is what Gen2 offers today as one of its multiple Image and text based video generation!

The best way to explain it is probably to let you watch this short video with several examples of images animated by Gen2.

You can test Gen2 for free here.

This is it for Today!

Until next time, this is Wassim Jouini, signing off. See you in the next edition!

Have a great Sunday and may AI always be on your side!