OpenAI's DALL-E 3 is a text-to-image model built into ChatGPT that handles complex prompts better than rivals, generating readable text and realistic hands. It integrates with ChatGPT for prompt refinement and rolls out in October, though it faces copyright lawsuits and restrictions on public figures and living artists' styles.
So OpenAI has just released Dall E3, the latest version of its text to image tool that can create amazing images from natural language descriptions. This is a huge deal because Dall E3 is much better than Dall E2 at creating images that closely follow complex prompts. For example, Dall E3 can accurately represent a scene with specific objects and the relationships between them. It can also generate text within an image and render human details like hands more realistically. And the best part is you don't need any prompt engineering to use Dall E3. You can just type in a simple sentence and get stunning results. No hacks or tricks required. So what is Dall E3 and how does it work? Well, Dall E3 is a 12 billion parameter version of GPT3 trained to generate images from text descriptions using a data set of text image pairs. It receives both the text and the image as a single stream of data containing up to 1280 tokens, and is trained using maximum likelihood to generate all of the tokens one after another. A token is any symbol from a discrete vocabulary. Like for us, every English letter is a token out of the 26 we have in our Alphabet. But for Dall E3, its tokens can represent both words and parts of images. Dall E3 is built natively on ChatGPT, which means you can use ChatGPT as a brainstorming partner and refiner of your prompts. Just ask it what you want to see in anything from a simple sentence to a detailed paragraph. Chat GPT will automatically generate tailored, detailed prompts for Dalle 3 that bring your idea to life. If you like a particular image but it's not quite right, you can ask it to make tweaks with just a few words. Dalle3 will then update the image accordingly. Dalle3 is now in Research Preview and will be available to ChatGPT and enterprise customers in October via the API and in labs later this fall. As with Dalle 2, the images you create with Dalle 3 are yours to use, and you don't need OpenAI's permission to reprint, sell or merchandise them. Now, when looking at Dalle 3 and other text to image models, honestly, it blows them out of the water. It produces images that are more detailed and lifelike than those from Midjourney. The images from Dalle 3 have brighter colors, clearer shapes, and a better overall look. In contrast, Midjourney's images appear blurry and not as clear. Another popular model is Stable Diffusion xl, which is designed to generate images from text prompts. It says it can make images with fewer words and can even put text into the images. But when you look at what Dalle3 can do, stable diffusion XL doesn't quite match up. Dalle3's images are of higher quality with clearer text and a more attractive design. The images from Stable Diffusion XL seem grainy and have too many tiny details that sometimes seems just unnecessary. Then there's deepfloid if, which is a new model that says it can cleverly put text into pictures. But when we see what it can do next to Dall? E3, it's clear that Deepfloid if does not match the ability of Dall? E3. The way OpenAI's new model combines text and pictures is smoother and looks more real. And the images from deepfloid IF don't look as good and feel fake. In conclusion, Dall E3 is leading the way in terms turning text into images. It's a big step up from Dall E2 and is better than other models available. It makes great images without needing extra tweaks. Plus it works with ChatGPT, making it even more versatile and strong while also being easy to use. And to be honest, the ease of use of an AI tool is crucial, which is why ChatGPT remains the top AI chatbot worldwide. While some chatbots might be better for specific tasks, ChatGPT is preferred because of its convenience. Now, while Dall E3 stands out as the latest marvel from OpenAI, it's essential to understand its journey. The original Dall? E was a revolutionary innovation when it emerged in January 2021. By April 2022, the world witnessed a remarkably advanced sequel that reshaped the arena of AI generated imagery. The technology behind these models, known as latent diffusion, progressively refines noise into images that the system recognizes from its training data. This same technique later paved the way for other models like the Open weight stable diffusion. But OpenAI's commitment to refining AI tools for text to image synthesis doesn't exist in a vacuum. Several formidable players are striving to perfect their image generating models. These competitors bring unique offerings to the table, with some even having distinct advantages in specific niches. And of course, Dall? E 3 is not perfect. It still has some limitations and challenges that need to be addressed. As we all know, the rise of AI generated images hasn't been without its share of controversies. With AI scraping vast datasets of human artwork, artists worldwide are concerned about the potential undermining or unethical replication of their styles. This fear has been so profound that there have been protests, lawsuits about copyright infringements, and even rulings from institutions like the U.S. copyright Office. Recently, a U.S. district Court judge even made a ruling regarding the copyright nature of AI generated art. Moreover, OpenAI is currently facing lawsuits from a group of US authors, including renowned names like John Grisham and George R.R. martin, accusing the company of using their works without permission to train ChatGPT. That's why OpenAI has taken steps to limit Dall? E3's ability to generate violent, adult or hateful content. It has also implemented mitigations to decline requests that ask for a public figure by name to prevent the generation of images that could be used for propaganda or misinformation. And it has designed Dall? E3 to decline requests that ask for an image in the style of a living artist to respect the rights and creativity of other artists. However, these steps are not enough to ensure the ethical and responsible use of Dall? E 3. There are still many unresolved issues and controversies surrounding AI image generation, such as the question of who owns the rights to the images generated by AI, how to protect the originality and authenticity of human made art, and how to prevent the misuse or abuse of AI generated images for malicious purposes. So OpenAI is trying to find solutions. They're developing a tool called a provenance classifier to determine if Dall E3 made a particular image. They hope to use this tool to better understand the ways generated images might be used and to inform their future policies and practices. So what do you think of Dall E3? Do you think it's useful for creating art or could it affect the value of human made art? Let me know in the comments. I hope you found this video interesting. If you enjoyed it, give it a thumbs up and think about subscribing for more AI videos. Don't forget to hit the notification bell to stay updated on new videos. Thanks for tuning in and I'll catch you in the next one.
