With all those generative AI tools available, do we still need human designers?
Artificial Intelligence (AI) is changing our life left, right and center at unprecedented speed. When it comes to graphic design and images, many platforms are popping up to provide text to image feature. Just like ChatGPT, you can just type in your request in text, the app or platform will generate an image in seconds.
In addition to providing all sorts of templates, filters, stickers, style and effects, generative AI Text to Image is an interesting feature that saves you a lot of time coming up with a draft of an imaginative picture or graphic.
I have used three different platforms to generate the same image requirement to see the results. Here's what I've got.
DeepAI doesn't require that you open an account before you try it. This is what I generated on the homepage. You may select different style and enhance the image after. However, I don't see much difference in enhancement. At the end of the day, it's a free quick trial. There was no watermark and the image was 2048 px x 2048 px, 72dpi file.
Picsart requires that you create an account to try. There is only 1 image that that it generates. You may add text, effects, stickers and even create an animation from the image. if you want to download or remove the big watermark in the middle, you will have to subscribe a plan.
You will need to sign in to Adobe to use Firefly (Beta). This image by far is the best of the three Generative AI platforms that I've tried. I don't know the image source of the other platforms, but Adobe has its own library and commercial images which are of higher quality for this Beta.
Generative Fill is another Firefly feature that allows users to add and subtract elements in the photo. From the image generated from text, I have added a crown, jewelry and changed the eye color of the dragon cat. The Generative Fill is also a "text to image" function but I can put it in specific area on the photo.
All I had to do is to identify the area using the brush, type the requirement in English, then it will generate 4 optiosns for me to choose.
All images generated can be downloaded. File size of this image is 1048 x 1024px. There is an Adobe Firefly logo and a message saying that this image not for commercial use at the bottom left corner. This leads to an important question: Intellectual property. It is still unclear who woul be the owner of the final generated image because it may have incorporated some copyrighted images.
Not yet. Text to Image feature is great for visualizing a creative concept. For example, if I were to create a far-fetched character like the dragon cat with wings, I can see some possible options in seconds, saving a lot of time sketching and coloring. However, if I wanted a realistic image, there isn't any good options yet. It comes down to the database that the software/app has.
Back in the good old days, the creative director would come up with a concept, sketch the idea, or hire an illustrators to visualize the concept, then present to the client. With the generative AI tools, we can simply put the concept in words then see what it looks like. It is potentially an efficient and cost effective tool to conceptualize an initial idea.