Spent $15 in DALL·E 2 credits creating this AI image (pub.towardsai.net)

Article summary

The article discusses the author's experience with DALL-E 2, a tool for generating images using AI. The author spent $15 in credits creating an AI image, but the outcome was often strange or useless. The discussion reveals that users have to be specific about what they want without over-stuffing or adding redundant words to get desired results. The tool has limitations, such as generating images that are too small or not being able to produce vector images.

Main themes

• AI image generation
• DALL-E 2 limitations
• Prompt specificity
• Tool comparisons
• Creative applications
• Technical challenges

What commenters say

• Using NLP to generate input for DALL-E 2 could be a useful approach.
• The tool's limitations, such as image size and lack of vector image support, hinder its potential.
• Iterating between image generation and description can lead to interesting and unexpected results.
• Some users find that combining tools, like Midjourney and DALL-E, can produce better outcomes.
• The need for specificity in prompts can be challenging, especially for creative work.
• Other tools, like Stable Diffusion, may offer better results and more flexibility.
• The potential for code generation tools to have similar limitations and challenges is noted.
• Comparing results from different tools, like Craiyon and DALL-E, can provide insights into their strengths and weaknesses.