Like a kid on Christmas morning, I clicked the “Get Started” button on my invitation email from OpenAI DALL-E 2.
Months ago, I joined their waitlist to beta test their powerful AI platform. For the past year, I’ve been experimenting with NightCafe Studio (NC), another AI art generator, with exciting but varied results.
Both AI art generators use text or keyword prompts to create visuals.
So, entering something like, “A wise owl wearing glasses, digital illustration” will result in a wide array of illustrations.
The more specific you are with the text prompts, the more control you can exert over the AI tool, provided you’ve trained the tool using multiple iterations.
After a few attempts, DALL-E 2 created a wide array of beautiful characters — something I couldn’t do with nearly as much ease in NightCafe Studio.
I entered the same prompt in DALL-E 2 and NC to see what both tools provided.
DALLE-2
NightCafe Studio
WOW! Right?
The DALL-E 2 illustration was one of 9 illustrations generated by the tool.
I spent 6 (!!) credits on NightCafe to get an illustration of bizarre-looking vegetables.
In order to get the NC version to look anything like the DALL-E 2 version, I would need to spend a lot more time and credits experimenting with various start images and prompts.
DALL-E 2 is incredibly, incredibly powerful.
More results from DALL-E 2
DALL-E 2 In the News
As I was experimenting with DALL-E 2, Cosmopolitan Magazine published its first cover using a DALL-E 2 generated image.
I tried the same prompt myself and the results were not as striking — proving that AI-generated artwork is revolutionary, but still inconsistent.
The inability to duplicate the same image from the same prompt proves 2 things:
1 – Nobody can really “copy” someone else, even if using the same text prompts
2 – Creating consistent images or visuals is extremely hard
The best combination, in my opinion, is leveraging all of the tools in your toolbox.
Use AI to generate a start image and then manipulate further in another editing tool like Photoshop.
Combining DALL-E 2 and NightCafe Studio
My favorite combination is to use DALL-E 2-created images as starter images for my NightCafe Studio prompts where I have greater control over the style and output.
Perhaps it is because I’m more familiar with NightCafe Studio, but I really love how I can control the finished result using my tested prompts.
Here are a few examples:
DALL-E 2
Post NightCafe Studio Edits
DALL-E 2
Post NightCafe Studio Edits
DALL-E 2
Post NightCafe Studio Edits
Ethical Considerations of AI Artwork
A lot of people are seeing the power of DALL-E 2 and freaking out saying that this software will replace artists.
AI can be used to improve our lives, but we are really uncomfortable when it starts replacing artistic creativity.
Employment, privacy, and AI becoming self-aware, are usually always cited as the top concerns for artificial intelligence (and rightly so!).
Sure, we are ok with AI in our cars braking for us in case a child runs into the street.
We are even ok with AI in healthcare to improve health outcomes.
But using AI to generate art?? Now we are in murky waters.
As someone who has been in the AI Artwork space for the past year and interacted with artists using Midjourney, NightCafe Studio, I can say with confidence that it does take a lot of experimentation, time, and effort to figure out how to use text prompts correctly.
A lot of artists say they use AI as their digital paintbrush.
Despite that, innovation and new technology should be approached with caution.
DALL-E 2 has strict limitations on creating artwork that resembles real people like celebrities and political figures.
Currently, DALL-E 2-generated art is for personal use only and cannot be commercialized.
AI can be a visual tool similar to how Billie Eilish and Kanye West experiment with autotune.
Remove the tool from the talented artists and they are still amazing.
Remove the tool from the amateur and they can’t produce the art.
Current limitations of AI Artwork
Currently, one major limitation of DALL-E 2, is that the AI generates completely different images every time.
One may struggle to create a cohesive narrative story with a main character using DALL-E 2 as it is not possible to replicate similar characters or scenes.
At the moment, the tool can be used for one-off visuals, inspiration, designing mood boards, or brainstorming ideas.
While DALL-E 2 is incredibly powerful, the novice user will have a hard time getting repeated results.
If one uses CLIP-guided diffusion tools and has sets of prompts that produce a similar style of artwork, it is very possible to create a book using only AI-generated artwork.
The Future of AI Artwork
DALL-E 2 is the most powerful AI-generator currently available and the results are stunning.
AI-artwork platforms are just finding their footing and they are becoming more powerful every day as users train the AI.
I don’t think AI-artwork will disappear and if anything, we will see more and more AI-generated artwork in our illustrated books, cookbooks, fairytales, magazines, and media.
What I love most about AI artwork is that it allows us all to play in the visual world.
If you want to sign up for DALL-E 2 beta access, you can do that here.
Since the technology is open to everyone, there are other interfaces that you can use to generate start images like DALLE-Mini.
Alternatively, you can always experiment with NightCafe Studio, Midjourney, StarryAI, WomboAI and others.
What do you think of AI-generated artwork?
Follow me on Insta!
If you love fae, elves, trolls, and more, be sure to follow me on Instagram @AIArtwork_Fantasy
Don’t miss one of our articles!
We highlight web3lit projects and creative pioneers every Friday and send a friendly email to our Gutenb3rg Readers.