A few months ago, I tested the abilities of text-davinci-003 and ChatGPT to create SVG images. The results were promising but not all that impressive.
Enter GPT-4.
After conducting multiple experiments and reviewing the examples provided above, it’s clear that GPT-4 has significantly improved the quality and capabilities of SVG generation compared to its predecessor, ChatGPT. The intricacies and precision showcased in these examples demonstrate the immense progress achieved in a relatively short time.
Test 1: Sunrise Over Water
Prompt: <!-- SVG: a sunrise over water -->
Prompt: Write the code for a highly detailed SVG of a sunrise over water:
Test 2: A Cat’s Face
Prompt 1: Here is an example of a highly detailed SVG of a cat's face:
Prompt 2: Write the code for a highly detailed SVG of a cat's face, including it's ears:
Test 3: A Dog
Prompt: Write the code for a highly detailed SVG of a dog:
Results
GPT-4 has made remarkable strides in the realm of SVG generation. Its ability to produce high-quality, intricate, and visually appealing images has far surpassed the capabilities of previous models. This progress showcases the rapid development of AI technology and hints at the possibilities and applications for AI-generated art and design in the future. As we continue to refine and enhance these models, the creative potential of AI-generated SVG images is bound to grow exponentially.