DALL-E
Contents
About
DALL-E is a program by OpenAI's that creates images from textual descriptions. To use it go to: labs.openai.com, create your account and type away. OpenAI developed DALL-E and DALL-E 2, which are deep learning models that can create digital images from natural language prompts. DALL-E was first introduced in January 2021, and utilizes a modified version of GPT-3. In April 2022, OpenAI released DALL-E 2, which has the ability to create more realistic images with higher resolution. The source code for DALL-E 2 is not yet available and access to the model was limited to a select group of users for research purposes, with a beta phase starting in July 2022. However, on September 28, 2022 it was made available to anyone without waitlist requirement.
Some Examples
Here are some fun examples:
Merging Two Concepts
Always fun - just take any two things and merge them.
Attempting Humans and Photo Realism
Trying to render photo-realistic images often falls down, especially if you want faces in there. It's got nothing on Midjourney right now.
Classic Art
Notice it still gets the faces wrong.
Graphic Design
Honestly these results don't hold a candle to Midjouney.
See Also
- Midjourney - Similar thing, but better in my humble opinion.
- Stable Diffusion - Similar thing.
Links
- openai.com/dall-e-2 - official website.
- Wikipedia DALL-E - wikipedia article.
Acknowledgements: The wonderful Lauren Zou from Google for introducing me to DALL-E, which she used in her fabulous mock website: "Ball Regulatory Commission"! |