Show HN: A Dalle-3 and GPT4-Vision feedback loop (dalle.party)

Article summary

The article discusses a project that uses a feedback loop between Dalle-3 and GPT4-Vision to generate images. The loop involves uploading an image to GPT-4 to generate a prompt, which is then used to create a new image with Dall-E. The process is repeated, creating a series of images. The project allows for experimentation with different prompts and themes, resulting in varied and sometimes surreal outcomes.

Main themes

• AI-generated art
• Feedback loops
• Image generation
• Creative collaboration
• AI limitations
• Artistic experimentation

What commenters say

• The feedback loop can produce interesting and humorous results, such as the 'descent into Corgi insanity'.
• The ability of GPT-4 to understand and generate prompts based on images is impressive, but can also lead to nonsensical or surreal outcomes.
• The use of custom prompts can influence the output of the feedback loop, but may not always produce the desired results.
• The project highlights the limitations and quirks of AI image generation, such as the inability to spell correctly or understand certain contexts.
• The feedback loop can be seen as a form of artistic collaboration between humans and AI, with the AI generating new and unexpected ideas.
• The project has implications for the development of AI-generated art and the potential for new forms of creative expression.
• Some users find the default UI of chat.openai to be limiting and frustrating, and appreciate alternative interfaces that allow for more experimentation and sharing.
• The feedback loop can be used to explore different themes and ideas, but may also result in unpredictable and unstable outcomes.