My experiments with AI stuffses

AI is blowing up right now, and it's only getting better. I'm trying to document my experiments with AI in this new series of posts. Come along for the ride as I hope to explore the possibilities of this amazing technology.

Using Bard to understand a photograph, and then trying to recreate it using AI

Used the following photo and asked Google Bard to explain what the photo is about.

This is what it came up with

// Image Description

The image depicts a red Chevrolet Opala classic car parked in front of a two-story brick building with a pitched roof.
The car is in good condition and has a license plate number of FU 325 17. 
The building is located on a gravel road surrounded by trees.
The image was taken on a sunny day with a blue sky.
The focus of the image is on the car, and the background is blurred to draw attention to the vehicle.
The colors in the image are warm and inviting, and the overall mood of the image is positive.
The image is well-composed and features a diagonal perspective that gives the image a sense of movement.
The shadows on the car and the building are well-defined, which helps to create a sense of depth in the image.
The texture of the brick building is visible, which adds to the realism of the image.
The image is a visually appealing depiction of a classic car parked in front of a brick building.
It would be suitable for use in a variety of contexts, such as a car dealership website, a classic car magazine, or a travel blog.

**Additional Details

The car is parked in a diagonal position, which gives the image a sense of movement.
The shadows on the car and the building are well-defined, which helps to create a sense of depth in the image.
The texture of the brick building is visible, which adds to the realism of the image.
The image was taken on a sunny day with a blue sky.
The image is well-composed and features a diagonal perspective.
The overall mood of the image is positive.

**Keywords
Chevrolet Opala
Classic car
Brick building
Gravel road
Trees
Sunny day
Blue sky
Diagonal perspective
Sense of movement
Shadows
Texture
Realism
Positive mood

After this, I asked Bard to create a prompt for to recreate the image in an AI model.

✒️Bard Prompt: A red Chevrolet Opala classic car parked in front of a two-story brick building on a sunny day with a blue sky. The car is in good condition and has a license plate. The building is located on a gravel road surrounded by trees. The focus of the image is on the car, and the background is blurred to draw attention to the vehicle. The colors in the image are warm and inviting, and the overall mood of the image is positive.

Using this prompt in Stable Diffusion XL using clipdrop.co, here's some of the results

Tweaked the prompt a bit, and made my version of the same prompt

✒️My Prompt: A red Chevrolet Opala classic car parked in front of a two-story building in Copenhagen on a sunny day with a blue sky. The car is in good condition. The building is located on a gravel road surrounded by trees. The focus of the image is on the car, and the background is blurred to draw attention to the vehicle. The vehicle is in side view. The colors in the image are warm and inviting, and the overall mood of the image is positive.

Observations

1. Bard's ability to correctly identify the model of the car and describe the location of the photo is impressive. This shows that Bard has a good understanding of cars and the world around it.

2. Bard's incorrect interpretation of the graffiti as the license plate number is a reminder that AI is still under development. AI models can make mistakes, and it is important to be aware of this when using them.

3. The ease with which a prompt could be generated from Bard's description of the photo is a promising sign. This suggests that AI could be used to help people generate prompts for AI models, which could make it easier for people to create realistic images and videos.