What Is The Big Deal About AI Imaging?

Nearly two years ago, I first read about a technology called Dall-e, which would help create content using Artificial Intelligence (AI). This was around the time when quite a few projects in the space of AI driven content were in news. These included sites or apps that could change pictures into caricatures, write short paragraphs or generate headlines, convert text to speech, or its opposite, i.e. speech to text. Captions for Video, or generating Video from a blog post, were other ways in which this idea evolved. There was a common factor in these tools. Majority of them used some form of AI enabled content creation technology or another.

This was also the period when I came across a site called podbot.ai. they use Deep Learning for creating podcast episodes. On this site, has to type a subject or a topic, and the site creates an entire podcast episode. This includes the audio, intro and outro music, choose a feature image from a free stock photo site, and even generate the transcript! (Note: as per their FAQ, this project used an older algorithm of GPT, i.e. GPT-2.) As a podcaster, I should have seen this as a threat. Instead, I was enamoured by the sheer possibilities AI driven content presented.

In the current stage of evolution of this technology, AI generated content + Human expertise can result in amazing outcomes.

As a side note, you can listen to the episodes of two podcast shows I created on podbot.ai. they are called, Weekly Seasonal, and The Experimental Stuff, respectively. I do not believe the developers are updating the back end technology for this site at the time of writing, but the idea is awesome indeed.

Over the next few quarters, many tools such as background removal, re-rouching old photos, and others like thispersondoesnotexist made a splash.. Or its variations, including thisartwork, thiscat, thishorse...doesnotexist, etc) They all were the pre-cursors to something big that was about to happen in second half of 2022.

#Fast Forward to Second Half of 2022

Around August 2022, AI driven image generation really took off. This was the time period when Openai removed the waiting period for trying out Dall-e-2. Stable Diffusion was open sourced a few weeks later, and Midjourney was publicly accessible as a beta. How quickly have things changed in the past few months!

Stable Diffusion is on to version 2.1 at the time of writing, and Midjourney is on V4. There are atleast one hundred different tools, sites and apps that can help you with one or more of the following:

-AI Image generation

-Image enhancement

-Image to Text Generation

-Creating variations of a specific image

-AI enabled Video generation

-Generating Avatars from profile pictures

The list of applications and possibilities for AI Imaging is ever expanding. This area excites me a lot because creating images using artificial intelligence opens up multiple opportunities and possibilities. From the hobbyist user to the newbie artist, to the digital marketer and even the scientists modelling climate change, and from medical professionals, game developers and satellite imagery. The possibilities are immense. We will get into some of the applications and use cases of AI Imaging in the subsequent chapters of this book

I believe AI based content generation is a Big Thing. Image generation is an important part of this new wave.

#Some applications found a readymade market

Among the different applications of AI based imaging tools, background removal tools became a big deal since mid 2021. And they still are, particularly for e-commerce sites for travel sites that need new and fresh creatives regularly for b2c marketing. Likewise for social media influencers who post new images and content on a very regular basis.Image enhancement is another popular application, in which the resolution and quality of images are improved using AI techniques. The changes made to the images may include removing stains from old all pictures, converting grainy to sharp pictures, and increasing the resolution (breadth and height of image in terms of number of pixels). With so many projects coming up practically every day, the field has really exploded. So much is the popularity of AI Imaging now, that even well established, graphic designing tools like Canva started introducing background removal and text to image using AI.