---
tags: ai
---

# DALL·E 2

* [link](https://openai.com/product/dall-e-2)

## description
* Generic description:
* Another OpenAI project, which together with ChatGPT is responsible for kick-starting the current wave of consumer interest in generative AI. This one takes text prompts and transforms them into computer graphics (images, photos, drawings, paintings, etc.).
* User notes

* Very good image creation, can make anything with any prompt, the more detailed the better
* It seems like generally if you give it very general prompts it will give back very wierd, surreal stuff, which can be cool if used intentionally
* Generally, the more specific the prompt the more it will look like real things instead of these kind of strange, surreal emalgamations of images that blend together (example below in "examples of creations" section)
* Very cool ability to use key words to make the art in a certain style (i.e. "impressionist", "pixel art", "basquiat")
* It seems pretty important to include a style like this in your prompt, and you can really try out almost any
* It is very easy to play with prompts, add adjectives, change around words to see what different things get you
* You can also expand images beyond their original bounds by generating new content around the borders
* You can also edit preexisting images and change them in many ways, either with specific sections like clothes or the entire style and content
* Each search uses a credit and you start with 50 credits. Once you run out they replenish in like a month
* You can buy 115 more for 15 dollars

## tutorials
Image editing: (https://help.openai.com/en/articles/6516417-dall-e-editor-guide)
## prompting tips
Prompting tips + examples (https://www.kimp.io/dall-e2/#g357463b31ac5)
* Generally: be specific but not overly complex, and be clear about style
## examples of creations
* when using a relatively general prompt like "surrealist painting of man and woman on the porch of a house in a desert" the images come out like this with features blended and mixed together in a way that is kind of distinctive to ai art
* this could be a cool practice if used intentionally because the images do have an interesting, unsettling vibe but for a lot of pedagogical uses more specificity can create images with less of this blended quality


* With a more specific prompt like "a surrealist painting of an old man and old woman on rocking chairs on the porch of a wooden house in a vasst, flat desert watching a huge sun set" I got these images
* These still have interesting and surreal qualities, but they don't have that less of the mixed up, blended faces and make more sense to the eye
* Personally, I like these images more because they are closer to my goal with this art, and I think for creative project uses more specificity is usually more interesting because it is clear what elements are intentional, and the images seem more human because they don't have the wierd blended quality

## pedagogical use-case