AI as a support tool
Recently, I discovered a feature that could reshape how we view Artificial Intelligence in the context of photography. While AI is often associated with the creation or manipulation of images, this post explores how it can provide value in a different way.
This discussion focuses on the ChatGPT app, as it’s the tool I currently have at my disposal.
In this post, I’ll highlight what I see as an interesting use for this tool: its ability to analyze and provide feedback on an image. This is achieved by simply uploading the image into the ChatGPT prompt and asking that it perform an analysis.
The offer of an unbiased critique
Sometimes, as photographers, we’re too close to our work to see it objectively. Emotional connections to our images can prevent us from viewing them through the eyes of detached viewers - an audience that may not grasp the story we aim to convey. Asking others for meaningful feedback can also feel risky, as that is reliant on the skill of the viewer and their ability to set aside personal tastes or biases.
This is where I’ve found ChatGPT to be somewhat helpful. It can serve as an unbiased critic, providing insights into composition, storytelling, and technical execution. I’ve tested this by uploading several of my own images - ranging from street photography to unconventional ICM (Intentional Camera Movement) work—and challenging ChatGPT to evaluate them.
Although it might be expected for the AI engine to operate on an established set of rules, I have been surprised by its uncanny ability to offer thoughtful and accurate evaluations beyond simple composition.
Here are a couple of examples of what ChatGPT offers, along with the prompts I used.
Example #1
Prompt used - “Analyze image and include style”
In this example the most basic of prompts was used in order to challenge the software
Response - The response touched on several in-depth key topics, but for brevity, I’ve included only the most relevant feedback to demonstrate the tool’s capabilities.
Overview:
‘This black-and-white image captures a dynamic scene on the Brooklyn Bridge, featuring a group of people engaged in various activities, such as taking photos or walking.’
Notice how the location was identified without any prior context, and how the content of the image was accurately assessed.
Style and Artistic Approach:
‘The photographer employs a street photography style, which is spontaneous and observational. The inclusion of everyday individuals in a public setting aligns with the genre’s focus on candid moments and urban life.
There’s an emphasis on human behavior and storytelling, which aligns with social documentary photography.
The image’s clarity and sharpness balance well with the chaotic energy of the subjects, suggesting a deliberate effort to preserve detail while capturing spontaneity.’
It’s remarkable how the tool has correctly identified the genre (“Street”) and the intended purpose of the photograph. What the software did NOT identify was the woman crouched in the center of the image. This for me is an integral part of the image, but to be fair to the software peer photographers seem to miss this key point also.
Example #2
Prompt used - “Perform a brief analysis of this image”
In this example, I wanted to truly challenge the software. The image I chose is one of my better-known works and serves as a prime example of using ICM (specifically the 'followICM' technique) to create a highly impressionistic scene.
I wondered whether the inclusion of multiple abstract characters and non-distinct umbrellas might pose a challenge to the AI engine.
Would the software be able to extract meaningful information from such an abstract composition? And how would that influence its analysis?
So, how did it perform?
Response - Once again, for the sake of brevity I have extracted from the analysis only some key points that are most relevant to this discussion.
Overview:
“This image is a captivating example of Intentional Camera Movement (ICM) photography, where motion blur is used to create an impressionistic effect. The photograph combines vibrant colors, fluid shapes, and the suggestion of urban life to evoke a sense of energy and dynamism.”
Motion and Abstraction:
“The blurred figures and umbrellas suggest movement through a rainy city street. The lack of sharp detail emphasizes form and color over individuality, turning the subjects into abstract shapes.”
Story and Emotion:
“The image captures a fleeting moment of urban life, evoking themes of transience, anonymity, and connection. The interplay of umbrellas and blurred figures hints at shared human experiences, even in the rush of everyday life.”
Overall Impression:
“This image is an evocative and artistic interpretation of a city street scene. By blending vibrant colors, soft textures, and dynamic motion, it transcends documentary photography to become an expression of mood and emotion. It effectively uses abstraction to invite viewers to project their own narratives onto the scene.”
Eerily, the software once again delivers an accurate assessment of the image. The evaluation is impressive in its identification of the style (ICM), interpretation of the blurred elements, and ability to determine the overall mood and intent.
Suggestions for improvement
The offer of critique is a bonus, but as an extension to this ChatGPT can offer suggestions on improvement if requested to do so directly.
As an example, here is an image I questioned due to crop (image #1), and the results suggested. Of course, the decision to crop or not remains firmly with the photographer, but sometimes suggestions can add something not formerly considered.
Prompt: ‘Visualize crops that may make this image more compelling’
Conclusion
Test cases have demonstrated that ChatGPT can deliver a thoughtful and competent analysis of images, even offering actionable suggestions for improvement when necessary. This makes it a potentially valuable tool for photographers seeking structured feedback and unbiased advice on enhancing their work.
Have you explored AI in a similar way or discovered other creative uses (beyond image creation) to support your craft?