Matthew Widdop
Matty is a Junior SEO Executive at Intelligency, helping our clients to improve their digital performance. Matty is currently studying for his Level 3 Multi-Channel Marketer apprenticeship and has completed a Sport Journalism Degree at the University of Huddersfield.
Chat GPT have released their latest features this week with 4o Image Generation being rolled out to customers. The latest feature allows users to generate images from scratch with a prompt, alter existing images or turn images into animations.
Improvements made to image generation
ChatGPT’s previous image generation capabilities had several limitations which have been addressed in this latest release including text rendering. For anyone who has tried to use AI Models to generate images in the past, it is commonplace for current models to confuse text and become incoherent which has been a major stumbling block for creators trying to generate images. 4o Image Generation comes with advanced text rendering which allows users to display coherent text on their images which is a massive step forward from previous models.
The latest model is also said to be much improved at understanding complicated prompts with multiple elements. Previous Dall-E models have struggled with “binding” in the past, fusing together different elements that have been given to it and often getting confused. 4o Image generation is much better at being able to understand how to combine elements and produce high-quality images to specification.
I tested out the text rendering on 4o Image generation by giving it Intelligencys brand designs and asking it to create me a superhero using our brand colours and logo. You can see that the text is legible and accurate to our existing brand colours and logo.
ChatGPT have admitted that the images do take longer to create, however believe the trade off is worth it for the images users are getting in return
What does this mean for artists?
Artists are concerned about the latest developments in AI technology with 4o Image generation. The latest release has seen a wave of users online recreating images in their favourite art styles with the famous Studio Ghibli style being used in the latest online trend. However, many users feel aggrieved by this saying these artists dedicated their lives to creating such art and it should not be available to be co-opted by anyone through LLM’s which they believe are essentially stealing the work. It also makes graphic design much more accessible to users without the skill which leaves designers who have been upskilling in the field for their entire careers on the back foot and leaves the future of the profession up in the air.
ChatGPT 4o Image Generation is available to all users including free users however, paid users can generate 5x as many images.
