Screenshot by David Gewirtz/ZDNET via DALL-E 3
We’ve been discussing DALL-E, OpenAI’s text-to-image generative AI tool, for a while now. It has been available for a long time (at least a long time in terms of generative AI) via its web interface at labs.openai.com. But recently OpenAI made it accessible from ChatGPT.
So I decided to get my hands dirty and see what it could do.
Activating DALL-E 3 from ChatGPT Plus
Like other ChatGPT add-ons, the DALL-E functionality is only available inside ChatGPT Plus. DALL-E 3 in ChatGPT Plus is currently listed as beta, but you don’t need to enable it in the settings panel.
You’ll need to choose it from the GPT-4 drop-down menu, and this choice eliminates access to Browse with Bing, ChatGPT Plugins, and Data Analytics.
Screenshot by David Gewirtz/ZDNET
Once you have checked it for your session, all you have to do is send it a few prompts.
Let’s see what we can make DALL-E do
Sending prompts is simple. Like ChatGPT, you must provide clear prompts. And like Midjourney and other text-to-image conversion tools, you need to provide stylistic information, if you have a style in mind.
For my first post, I requested a photorealistic Mac mini on a marble pedestal.
Screenshot by David Gewirtz/ZDNET
I got four renderings. I chose the third one by clicking on it, and got this larger image:
Screenshot by David Gewirtz/ZDNET via DALL-E 3
As you can see, it looks a lot like a classic Mac mini. The case is a bit more rounded and the ports aren’t exactly the same, but it’s definitely close. I was impressed. So much so that I tried to convince DALL-E to produce a line of MacBook Pros.
But after creating a near-perfect Mac mini, ChatGPT explained to me that there was an intellectual property problem.
Screenshot by David Gewirtz/ZDNET
So I tried to get DALL-E to produce laptops, without any brand pedigree. Here’s what I got in return:
Screenshot by David Gewirtz/ZDNET via DALL-E 3
In case you can’t make out the details, let’s zoom in on the last image.
Screenshot by David Gewirtz/ZDNET via DALL-E 3
For some reason, ChatGPT saw fit to provide two laptops from the year… 1900. Even though I’m working with DALL-E, ChatGPT’s ability to produce completely wrong answers is still there. . I felt like I was meeting an old friend again.
A handy PowerPoint maker
DALL-E in ChatGPT is significantly more limited than DALL-E itself, and than Midjourney. You cannot upload images there. You can’t pan around an image and ask DALL-E to expand the image. You cannot selectively edit parts of the image. But what it lacks in power it makes up for in convenience.
Take the following image:
Screenshot by David Gewirtz/ZDNET via DALL-E 3
If you need an image of someone working remotely for a slide, you can generate one very quickly. To be honest, the image doesn’t look photo realistic, but it’s certainly good enough to use in PowerPoint.
Screenshot by David Gewirtz/ZDNET via DALL-E 3
Let’s say you want to illustrate a meeting in which revenue increases. Here’s a quick prompt and four versions, rendered as if from a Pixar cartoon:
Screenshot by David Gewirtz/ZDNET via DALL-E 3
Of course, in true ChatGPT style, only one of the images actually features four people, but that’s okay. I actually prefer the fourth version, which has six people:
Screenshot by David Gewirtz/ZDNET via DALL-E 3
Note : Diversity can be a problem. Of the 22 people that DALL-E 3 generated, only two appear to be of color. If you create slides and other documents, you should be aware of this issue and guide DALL-E appropriately.
Some final thoughts on DALL-E 3 in ChatGPT Plus
There it’s done. It’s very easy to get started with text to image conversion with ChatGPT’s DALL-E add-on. I recommend you to try. I’ll keep my paid Midjourney account as I use some of the advanced features for projects, but I’ll definitely use DALL-E in ChatGPT for quick images.
What about you? Have you used text to image conversion tools? Which ? What was your experience? Let me know in the comments below.
Source: “ZDNet.com”