For years, I've had a great fondness for computer-based graphic design. Tasks such as building a presentation, preparing an event invitation, or creating icons for the knowledge management system I'm responsible for at work, have always been interesting and enjoyable for me. However, as someone without formal training in graphics design, when I needed to incorporate images or illustrations of characters, for example, until recently I would turn to stock image libraries offering such files. Finding the required images consumed a lot of time, and often, I found myself compromising on the representation that would appear in the final product, simply because I couldn't find graphics that perfectly matched what I wanted to present. To my great delight, all these problems were solved when I discovered Bing AI Image Creator.
In recent years, several tools have been created to assist in graphics creation using AI mechanisms, and I've examined quite a few of them. Each has its advantages and disadvantages. I found Bing's tool to be the most suitable for my needs, being simple and convenient, quick in producing results, allowing for learning that leads to improved outcomes, and additionally, free of charge.
Bing AI Image Creator was launched in March 2023, and it's based on the deep learning technology implemented in DALL-E, which gained significant publicity. After a few months, Microsoft rebranded these capabilities, along with improvements and additions, into a new tool as part of the Copilot toolkit, naming it "Copilot Designer". As of the publication of this article, both tools are available to the general public, and I find myself sometimes switching between the two due to the unique advantages of each.
Note: It is recommended to access these two tools through the Edge browser. When requesting to create graphics with tools it has developed, Microsoft has introduced several improvements for those accessing them through the browser it promotes.
Bing Image Creator
Access to the Bing Image Creator interface is through the address https://www.bing.com/images/create.
Upon registering for the service, the user receives several dozen "Boosts" (in the system's terminology). Each "Boost" means the ability to receive graphics according to the user's request within about 10-20 seconds. One Boost is deducted for each response. After the initial quantity is exhausted, the user will receive a daily amount of 15 Boosts. If all available Boosts have been used, the user can still request the system to create additional graphics, but they will usually take longer to produce the result. However, from my experience, often, when there are no Boosts left, no results are received from the system, even when patience is exercised. This probably relates to the utilization of servers used for AI tasks. When many users submit tasks to the system, the time allocated to each task for those without Boosts will be long. Don't worry -- there's no need to remember how many Boosts you've used. The number of Boosts available to the user is always prominently displayed at the top of the window, as depicted in the following screenshot:
The response given to each request, known as Prompt, will usually consist of four different graphic creations that meet the user's definition. According to Microsoft's statements, the graphics created by the system are based on learning from about a billion graphic files that were scanned, analyzed, and learned by the system. Nevertheless, from time to time, when the prompt is very complex, such as one that includes a demand for many details or elements that the system doesn't know many variations of, the user will receive fewer results. Thus, when I asked to create graphics including a Menorah, few results were received that met the need, and after improving the prompt, a message was displayed requesting to wait a longer time to complete the task, finally, to my disappointment, only one result was received.
Rerunning the same prompt will yield different results, sometimes even such that have no resemblance to each other. As each prompt will yield different results, the user must improve and refine their prompt to achieve the desired result. Be prepared for surprises! Sometimes many attempts will be required to achieve the desired result. In this tool, it's not possible to refer to a specific graphic result received and ask the system to change parts of it. Therefore, using an additional Boost to produce a result from the same prompt will always be a kind of gamble, which may not necessarily advance you toward your goal.
The system allows easy downloading to the computer of any graphic result received. A significant advantage of Bing AI Image Creator is the ability to save a history of past prompts within the system and create Collection folders of graphics received. Thus, you can save graphics according to topics or needs and return to the saved results, with the prompt that led to its creation saved alongside each design. Currently, there is no limit to the number of Collections that can be created. In each Collection, an unlimited number of creations can be saved, graphics can easily be moved from one Collection to another, and notes can be added about each creation, which will be saved in the Collection folder alongside that design. Access to the Collection folders is through the navigation menu located on the right side of the window, and on the window that opens, clicking on "Collections".
Copilot Designer
The ability to save history and preserve successful graphics in folders within the system does not exist in Copilot Designer. However, this tool is built in a way that allows dialogue with the system in a natural language. In addition to the option of typing text, one can use a microphone and give voice commands to the system. Similarly to the Bing AI Image Creator, the Copilot Designer will usually provide four results. This time, if you want to try to improve the results received based on the original prompt, you can refer to the results received, and communicate with Copilot to improve them.
Accessing Copilot Designer is done by clicking on the link to Designer on the side of Copilot's main page:: https://copilot.microsoft.com.
Beyond the ability to request a result similar to the one received, Copilot allows uploading a file and asking the system to create a prompt based on that uploaded image. The tool will analyze in detail the uploaded image and generate the wording of the prompt for an image with details similar to those in the uploaded image. Note that you can only upload a single image; it's not possible to upload multiple images and refer to them.
Both above tools create graphics with generic characters, according to the user's prompt. When asked to incorporate famous characters for which there are copyright restrictions, the graphics produced will not include those characters, and in Copilot, the system will even notify the user that it is committed to preserving copyrights. However, certain characters, such as Albert Einstein, Mickey and Minnie Mouse, and even Hello Kitty, can be incorporated into Copilot’s creations, as demonstrated herein:
The prompt for the design of Albert Einstein alongside the characters:
Albert Einstein in a research lab, playing cards with Hello Kitty, Mickey Mouse, and Minnie Mouse, hyper-realistic 3D still life painting, hyper-detail, ultra HD, soft features, realistic proportional lighting, highly detailed illustration. Please make sure that Mickey, Minnie, and Hello Kitty are sitting on chairs next to Einstein and that they are his size.
I won't finish before mentioning that a third Microsoft tool that is available at https://designer.microsoft.com under the name "Microsoft Designer" (without using the name "Copilot") and serves as a kind of competitor to Canva. We'll talk about that another time.
Try each of these tools, and I'm quite sure you'll be pleased.
Comments