This is the latest entry in our ‘State of the art-ful AI’ series, where we upload a photo to Chat GPT 4 and ask it for a photo description, then ask DALL-E 3 to generate an image using the description. How similar will they be?
Why this workflow?
Less than a year after its launch, ChatGPT and DALL-E have already seen multiple iterations. In October 2023, ChatGPT 4 gained the capability to describe uploaded photos. To track the progress of generative AI, we uploaded these AI descriptions to the Dall-E 3 image generation model to see how faithful its output is to the original photo. A variety of photos have been selected which we’ll bring you throughout November and December.
Our original photo
This image was taken at the Longleat festival of light, Longleat, Wiltshire, England around 2017.
The photo as described by Chat GPT 4 in 150 words
The photo captures a vibrant and colorful light sculpture of a dragon at a night event, possibly a lantern festival. The dragon’s body undulates in a sinuous form, creating the illusion of movement as it stretches into the distance. It is adorned with a multitude of illuminated bulbs, with the colors transitioning from a fiery orange at the head to a sequence of cool blues and purples along its body. The dragon’s head is raised, mouth open, with bright yellow lights suggesting fierce eyes and a roaring mouth, giving it an animated expression. Additional decorative lights, resembling flora or flames, accent its spine.
The scene is set against the backdrop of a darkening sky and silhouetted bare trees, which contrast starkly with the vivid colors of the installation. The surrounding area is dotted with other illuminated objects and strings of lights, contributing to the festive atmosphere. Spectators are visible at the bottom edge of the photo, adding a sense of scale to the large installation.
Photos generated by Dall-E 3 using the AI description
Spot the difference
Using a separate chat, we asked Chat GPT what differences it finds between the original photo and the FIRST image generated above:
- Positioning of the Dragon: In the first image, the dragon’s head is facing upwards, whereas in the second image, it is oriented towards the left side of the frame with its mouth open as if roaring.
- Color Scheme: The dragon in the first image features a predominance of pink hues with some purple and blue on its body, and warm orange for the mane. In contrast, the second image shows a dragon with a more diverse color palette, with vibrant blues, oranges, yellows, and reds throughout its body and mane.
- Lighting and Time of Day: The first image appears to be taken during twilight or early evening, giving the surroundings a darker shade, while the second image is captured at night, which makes the lights stand out more prominently.
- Background Activity: There are visible groups of people in the second image, suggesting an event or exhibition, while the first image has no discernible crowd.
- Surrounding Decorations: The second image shows more elaborate and varied light decorations in the background, including trees and other structures, as opposed to the simpler, less colorful background decorations in the first image.
- Detailing on the Dragon: The dragon in the second image has more intricate patterns and a texture that appears like scales, while the first image has a smoother texture with less visible detailing.
- Composition and Perspective: The first image is taken from a side angle, showing more of the dragon’s body length, whereas the second image is taken from a frontal angle, focusing more on the dragon’s head and upper body.
- Ambience: The ambience of the second image is more festive and vibrant due to the variety of colors and the presence of a crowd, compared to the relatively subdued ambience of the first image.
- Clarity and Quality: The second image appears to have a higher resolution with clearer details compared to the first image, which seems a bit less sharp.
Example workflow video
Here’s our workflow in action, using an ice cream truck image. This is not an explainer video – It shows the main steps of the above ChatGPT 4 workflow in November 2023.
