How do you use artificial intelligence to synthesize better images?
Much of the excitement and fear that AI generates focuses on text, but systems that generate images are also experiencing rapid and dramatic advances. In many cases, these systems share a similar syntax to generative, text-based AI, but can be more engaging and lend themselves to some fun creative activities.
Creative Intelligence Innovations
Image generators are trained on billions of images, allowing them to create works that were once the preserve of illustrators and artists. Experts sometimes can’t tell the difference between AI-generated images and real photos (a situation that opens the door to both dangerous disinformation campaigns and hilarious discoveries).
Compared to products like ChatGBT, AI tools that generate images are still not at the same level of development because they still have some challenges to overcome and cost more money. But if you’re interested in learning to use these tools, there’s no better time than now.
Photoshop with artificial intelligence
Adobe recently added a creative artificial intelligence feature to a new beta version of its standard graphics program Photoshop, and the feature has caused a stir on social media platforms like TikTok and Instagram.
When we tested the new Generative Fill feature in Photoshop, we were impressed with the speed and professionalism with which it performed artificial intelligence tasks that would have taken hours to complete ourselves. In less than five minutes and a few clicks, we’ve used the feature to delete some things, add others, and change the background.
(To test these tools, start by signing up for a free trial of Adobe Creative Suite, then download the new Adobe Photoshop beta trial, which includes the Generative File feature.)
After downloading the copy, you can import any image and try the following tricks:
– Click the “Select Theme” icon (it looks like an arrow pointing to a box) to change the background. Next, go to the “Select” options menu and click “Invert” to choose a background, then click the “Create File” box and type the command (pumped) or leave it blank to let Photoshop create a new background for you. .
We used these steps to create an image of a corgi, typing “Dog Fancier” into the prompt box and clicking “Create” to change the background.
– Use the lasso tool to remove objects. In the picture of a motorcycle, we wanted to remove a tractor behind a fence in the background. We drew a line around the tractor, then clicked on the “Create File” box, and then clicked “Generate” without entering the instructions in the induction box. The program succeeded in removing the jars and filling the background without affecting the fence.
New York Times photo editing professionals don’t use artificial intelligence to enhance, modify, or create images, but our first impression after trying the new “Adobe” feature is that photo editing professionals working in other fields like marketing will soon be out of a job. When I shared this theory with Eli Greenfield, Adobe’s chief technology officer, he responded that the feature would make editing more accessible and accessible to people, but he seemed optimistic about continued demand for humans.
Greenfield added: “I can make better pictures with this tool, but my pictures are still boring. When I look at the content created by artists and compare it to mine, I feel that their work is more interesting because they know how to tell a story better.
We admit that what we were able to do with “generic fill” was far less impressive than what others were posting on “social media.” Lorenzo Green, who always tweets about artificial intelligence, published a poster for the covers of popular music albums like Michael Jackson’s “Thriller” and Adele’s “21” which he expanded using artificial intelligence, the results of which must be acknowledged. It was really interesting.
(A note: If you feel like downloading Photoshop is tiring, you can test Adobe’s artificial intelligence on the Adobe Firefly website, where you open the Generative File tool, upload the image you want to adjust, click “Add” and draw a line around it. Target Object, then “Background ” and type “beach” at any command prompt.)
Many tools to create images
Tools like DALL-E and Midjourney can create new images in seconds. These tools work like chatbots: you type instant text, and the more accurately you type your text, the better results you’ll get.
To write a quality text, start with the medium you want to follow, followed by the main topic and additional details. For example, typing “photo of a cat wearing a jacket in a bright room” into DE’s prompt box will get you the photo you requested.
Owned by OpenAI, the developer of ChatGPT, DE is one of the first, most popular, and easy-to-use AI image generation tools. The program gives you 115 credits for $15, and one credit can be used to create a set of four images.
Another AI tool for image creation, Medjourney, is still in development, which means it doesn’t deliver as good and accurate results as its predecessors. The service costs $10 per month, and the onboarding process is somewhat complicated because it relies on a separate app called Discord. However, this program can produce promising, high-quality images.
To use it, join the Discord app and then request an invite to the Medjourney server. After connecting, type “/imagine continue prompt” in the dialog box. During testing, we typed “/imagine a manga magazine cover featuring a corgi dressed as a Ninja Turtle,” and the tool produced a compelling set of images.
Normal line printing is fine, but some people find that ambiguous stimuli produce unusual results. Notably, Lance Wheeler teaches his students at Columbia University how to use artificial intelligence tools including “Midjourney” to create works of art.
Whatever tool you use, remember that it is your responsibility to use these tools responsibly. Technology experts warn that image-making tools can increase the spread of persuasive fakes and misinformation, but they can also be used in positive and creative ways, such as creating great family photos and brainstorming artistic principles.
“Professional coffee fan. Total beer nerd. Hardcore reader. Alcohol fanatic. Evil twitter buff. Friendly tv scholar.”