From the course: OpenAI API for Python Developers

Text to image: Introducing the DALL·E model

From the course: OpenAI API for Python Developers

Text to image: Introducing the DALL·E model

“

- [Narrator] Over the past few years, image generation models have become more sophisticated, leveraging the power of deep learning to generate original, realistic, and high-quality images. Before the release of the ChatGPT application, along with the OpenAI's APIs, deep learning networks like GAN, which is short for generative adversarial network, and other machine learning systems, were already capable of generating very realistic images. Some of these generative models are even behind some deepfakes representing the likeness of some celebrities. So with the advanced technologies of today, AI-generated images look so realistic that you couldn't tell the difference between deepfakes and real ones. In 2021, OpenAI releases DALL-E, an image generation model capable of generating more realistic images and art with a greater resolution based on a simple text inputs. So, what about we create our own image generator? So for that, we're going to integrate the image generation model right here, that you're going to find under documentation, image generation. And you'll see that the syntax is very simple. We're going to look at the usage for generation, and we basically need to, here, define this function and specify the model, which will be DALL-E 3. Then, it's going to be easy to generate a new original image based on a simple text inputs. So, this is as simple as that. So, let's look at the project. We're going to see the requirements, and you're going to see also a definition of the different packages that we use for this project. We even use Pillow, which is a Python library, to handle images within your application. So after you go through all the instructions, which is to install the packages, set up your API key, you'll be then ready to start your app. So, let's look at the project a little bit. What we want is to allow to generate an image that we're going to do after we submit the button. And also, we're going to add another feature which is going to allow us to then download. Every file that we generate with AI, it's going to be downloaded and saved within this directory, which is media. And after that, the final part will be to create a gallery, so we're going to display all the images in the gallery. So, we're going to complete this project together. Sounds like a very exciting project where GPT and art meet together to create original artworks with the power of AI.

Contents