Coffee culture and nostalgia on the street
For the "Augmented with Artificial Intelligence" photo contest, I reinterpreted a shot from a vintage car photoshoot at Lake Constance using artificial intelligence.
I used ChatGPT, which utilizes DALL·E for image editing. Step by step, a real photograph was transformed into a cinematic scene that could easily have come from a gangster movie set in the south of France.
This was my first attempt at developing a photo further using AI in this way. Previously, I had only worked with AI-generated masks - a completely different approach in terms of both technique and creative execution.
The original image was particularly well-suited for this: the goal was to move the vintage car from its setting in a Lake Constance vineyard to an environment that matched its character and era - a brief journey through time made possible by artificial intelligence. I was curious to see what is possible with AI and to what extent my image concept could be realized.
Since I had no prior experience with this type of image editing, my prompts were very simple. It took several iterations to achieve the desired result. First, the background and the road were replaced: the original landscape gave way to a cobblestone street scene typical of an old town in southern France. Then, the German license plate was swapped for a period-appropriate French one.
In the next step, the image was extended to the left, adding two bistro chairs and a small table. Finally, a suitably dressed model was added to complete the scene's cinematic atmosphere.
In this way, a documentary-style shot became a fictional movie scene - not entirely reinvented, but expanded step by step based on the original photograph.
I think it's clearly evident that AI was involved here. Nevertheless, I find the result quite impressive.
4 Comments
The starting point
I absolutely love this, but the perspective of the table and chair with that curb really take me out of it. Otherwise, I'd have given this an Excellent.
Thanks a lot Marian for your constructive input! I’ve also noticed that the work still has a few weak points. Since I don’t have much experience with generative AI yet, this is really helpful for me.
This will be the next step in the AI learning curve for me as well. I ran out of credits before I could try making further edits to the initial AI image. In the case of your picture here, I'd want to instruct AI to place all four of the legs of the chair on the sidewalk... and see what it would do.