This week, OpenAI released a groundbreaking new feature for its wildly popular ChatGPT platform. In addition to working with text, ChatGPT can now process and chat about images.
Up to 70% of the content on the Internet today is visual, not textual. For example, people are more willing to take thousands of photos than write text every year, and for example, the major self-media platforms are basically visual content is more popular with people.
The new features will be rolled out to select ChatGPT Plus users in the coming weeks. At present, some people have already qualified for the first batch of closed betas.
Here's an introduction to ChatGPT's new features.
Interact visually with ChatGPT
Visual interaction with ChatGPT is straightforward. Users who have been granted access can now see a small icon next to the Send Message field of the ChatGPT interface.
In addition to entering text hints, you can now upload images at the same time as you upload text. This allows you to ask questions about the image, or have ChatGPT analyze the visual content of the image and chat with you.
Three case studies to demonstrate the new features of ChatGPT:
1. Describe the picture as a whole
After uploading an image, ask ChatGPT: What did you see? Take this image as an example.
The result of ChatGPT output is shown in the following figure:
As we can see, ChatGPT doesn't just describe the container and the tube on the wall, it also provides the details of the image (the color of each cleaning solution, their relative position from left to right, and the text on each cleaning solution), which is simply useful for people who don't understand what pictures mean, and is also useful for those with color blindness.
2. Detailed analysis of the picture
Take the following figure as an example, through a picture of the spice rack, see how good ChatGPT is to analyze the picture, but there is a premise, your picture must be clear enough.
The analysis results are as follows:
From the results, we can see that ChatGPT can not only list the name of the seasoning, but also describe the brand.
3. Analyze the breed of dog.
Ask ChatGPT: Please analyze the breed of this puppy.
The reply was as follows:
Despite ChatGPT's major breakthrough, there are some limitations at the moment.
1. It does not process or introduce character images. There are a lot of biases in visual AI systems. It doesn't accurately describe the outcome you want.
2. It will not answer medical-related topics about pictures, which is also a good thing, after all, the development of artificial intelligence in the medical field is not mature enough.
There are more gameplay waiting for us to dig, but do you think the cost of more than 150 yuan per month is worth it?
#Brilliant Flower Creation Challenge ##白城头条#