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 photo was taken in an Oxfordshire sculpture park in 2023. The ‘warped’ hand is reflected ‘correctly’ in the central metal tube!
The photo as described by Chat GPT 4 in 150 words
The image depicts a sculpture that is an intriguing blend of abstract and figurative elements. It’s set outdoors, as evidenced by the green grass and a paved path in the background. The sculpture appears to be a stylized, oversized banana, but upon closer inspection, it integrates human hands in a manner that suggests interaction or manipulation. The hands emerge from the reflective surface which seems to be a mirror or polished metal, possibly stainless steel. The hands are detailed and realistic, contrasting with the smooth, flowing form of the banana-like structure.
The sculpture’s surface has a lustrous sheen, indicating it’s made of a material like bronze or polished stone that has been finished to a high shine. The reflective quality of the surface captures elements of the surroundings, creating an interplay between the sculpture and its environment. The artwork may be open to interpretation, but it likely explores themes of nature, human influence, and perhaps the surreal coexistence of the natural with the man-made.
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:
- Perspective and Angle: The first image is taken from a side angle and appears to be at eye level, giving a side profile of the sculpture. The second image is taken from a lower angle, looking up at the sculpture, showing more of the underside and less of the top surface.
- Background: The background in the first image includes a grassy area and a paved walkway, while the second image has a grassy background with a building far in the distance and a clear sky.
- Reflections: The reflective surface of the sculpture in the first image shows distorted reflections of what appears to be a nearby environment, possibly including the photographer. The second image shows clear blue sky and the surrounding landscape reflected on the sculpture’s surface.
- Color and Lighting: The first image has a warmer tone, with the sculpture appearing more golden due to the lighting or the time of day. The second image has a cooler tone, with the sculpture appearing more silver or chrome-like.
- Detail and Texture: The texture of the sculpture, especially on the fingers, is more pronounced in the second image, likely due to the different lighting conditions and the closer perspective.
- Image Composition: The composition of the first image includes more of the surrounding environment, with the sculpture offset to the left side. In the second image, the sculpture is centered and occupies more of the frame, with less environmental context.
- Image Quality: The second image appears to have higher resolution or sharper focus, as the details of the sculpture, like the creases in the fingers, are more defined.
- Distortion: There is a noticeable distortion in the first image, likely from the curvature of the reflective surface, which is not as evident in the second image due to the different angle of the photo.
These observations are based on the visual information available in the provided images.
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.
