Determined to steal OpenAI’s thunder, Google has released another image remix AI tool, Whisk. OpenAI’s “12 days of Shipmas,” has brought us a lot of different announcements but just as we were getting excited over Sora’s more widespread release and its AI video generation capabilities, Google has swooped in once more with its AI photo generator.
The Whisk tool is a fun way to take images you already have and combine them together in a unique way. For example, you can use an image of your pet and fuse it with an animated image of a beach, combining the two images to create an image of your pet in an entirely new format. It’s not groundbreaking necessarily, but it still adds a new layer to the AI tools we have at our disposal.
Image: Google’s new Whisk tool
Presenting the New Google Image Remix AI Tool
Google Labs’s latest experiment with AI has resulted in a new tool that essentially eliminates the need for long and elaborate natural language text prompts to generate images. Instead, it allows you to work with existing images—even AI-generated ones—where you pick a separate image in three categories, subject, style, and scene, and then combine all three in novel ways.
This new Google AI photo generator takes away some of the control you have over your results, but it provides you with unique inspiration each time. Because the text prompts make up only one part of the instruction of what you want the AI to pull from your images and what it should generate, the results can often differ from what you’re expecting. However, you can also view and edit the prompt at any time.
Hand’s On Experience with Google Labs’ Whisk Tool
The creators of the tool provide multiple warnings that the AI may be inaccurate to ensure you temper your expectations accordingly. You can test your luck and roll the dice with suggestions for images from the tool itself, or upload your own photos from your desktop to try the tool out. It is a little slow to respond to instructions sometimes, but the results are noteworthy.
The AI is highly sophisticated, starting right from its capability to decipher the contents of your image and produce relevant results. When we uploaded a screenshot from a video game, the generated description of the image showed us that it was able to easily recognize that this was a conversation between two game characters.
It also succeeded in highlighting the important elements in the scene to use for its own image generator. The results of the image generator were quite exciting as well, considering the AI tool took the characters out of their setting and placed them in an entirely new context with ease.
For prompts where we didn’t have an exact image to use, we were able to input a text prompt instead. The AI tool first generated an image for the prompt and then used the image to generate a central picture that combined all three image prompts. Truly astounding.
Image: The Whisk tool was able to look at a screenshot from Genshin Impact, distinguish what was happening, and then describe the surroundings. These descriptors are likely used to aid in generating new images and you can edit the details of each individual image before you generate your final one.
How to Access the Google Whisk AI Image Creator?
If you live in the U.S., the AI tool is freely available to test and play around with. Users can access Whisk on the Google Labs website, and provide the team with feedback on the tool and how it may dive deeper into generative AI. An elaborate privacy policy appears when you attempt to log in, but it’s nothing you wouldn’t expect from an experimental tool.
This tool does not appear to be a part of the company’s catalog of products and services and is only being released as a test service, so it’s unclear if the Google Whisk AI will launch globally anytime soon. It’s possible that the Google Labs team will collect feedback with regard to the tool and then explore the possibility of eventually integrating it into the organization’s other services.
This could be a very useful service for creatives to test out their ideas and build mockups of existing projects before they work on designing the images themselves, however, the potential for misuse continues to loom over all AI tools and steals some of the excitement surrounding them. Will we see a rise in people copying artists’ works and their art styles with a few simple clicks? That’s to be expected.
We were able to generate images of a realistic enamel pin from images of a comic we’re obsessed with, and the results were astounding. Could someone use this to come up with ideas for future products for their small business? Absolutely. Is there a chance someone will use such a tool to put up fake product listings to dupe customers? It’s absolutely possible as well.