The rise of DALL-E 2's AI photo editing capabilities might just mean the end of the world as we know it. Just kidding – but seriously, DALLE-2 from OpenAI can generate completely novel images from textual input, and that finally includes images with faces. This provides helpful, real-world benefits to those who want to visualize virtually anything they want. OpenAI allowing users to edit faces with DALL-E will completely change the game of AI. OpenAI used to allow facial manipulation with DALL-E, but the company had raised deepfake & security concerns regarding this getting exploited. This technology is only going to become more advanced and ubiquitous in the coming years, so it's important to start thinking about the implications of AI-generated images now. What could go wrong? Well, for one, we might see a lot more "fake" images being shared online. It's already hard enough to tell what's real and what's not on the internet, and this will make it even harder. We might also see people trying to use AI-generated images to defraud others – for example, by creating fake pictures of celebrities endorsing products.
With this new update, OpenAI claims sexual, political, and violent content is more securely blocked in image generation. This means we can finally use DALL-E 2 for things involving people. The benefits to society are already becoming apparent – a reconstructive surgeon had been using DALL-E to help his patients visualize results. And filmmakers have told OpenAI that they want to be able to edit images of scenes with people to help speed up their creative processes. Some other use cases for the tool include:
- Seeing what you would look like with a different hairstyle or outfit
- Changing the couch type in a family portrait
- Trying on some funky outfits
- Visiting any tourist location and bragging about it on your Instagram
Facial editing with DALL-E works the same as any other inpainted image: upload the image, remove the borders and background, type your prompt in, and let the AI complete the rest. The results are always a little bit uncanny, but still fun to experiment with.
Honestly, DALL-E still seems to struggle with photorealistic images of nature and other more concrete images. Regardless, it's still really awesome to see the progress that's been made with this tool in such a short amount of time.
Who knows – maybe in a few years we'll all be able to convince everyone that we're actually on the moon. Just don't be too shocked when you start seeing some really weird stuff on your feed.
With the number of trials it took to get a few of these (between editing errors and production errors), inpainting in pictures with faces can get pricey. While we do recommend trying this new feature out if you have access to DALL-E, you should apply for the financial assistance program if you're a student, non-profit, or designer.
We now have the full array of tools to really create anything you're looking for. Inpainting and outpainting in conjunction with using photos containing faces lets you use DALL-E for it's raw, intended purpose. In just 5 months we went from a beta to having stable access to one of the coolest pieces of groundbreaking software that has been released over the last few years.
It seems like every week we get another step closer to the real-world application of artificial intelligence. And with tools like DALL-E, it's becoming easier and easier for people to use AI in their everyday lives. Whether you're a filmmaker or a fashionista, there's a good chance that DALL-E can help you speed up your creative process. So go forth and experiment – but keep your ethics and morals in the back of your mind. As AI gets better and better, it's going to be more and more important to make sure we're using it for good. What do you think about editing images with faces using DALL-E? Was this switched on too soon or does OpenAI have a good handle on things?