[综合] ChatGPT Image Input: FAQ When Uploading Images

What are image inputs and how do they work in ChatGPT?
ChatGPT now has image capabilities and can understand and interpret images you add to a conversation as image inputs.

How should I use image inputs in a conversation?
Basic usage: Start by uploading a photo. Ask about objects in the image, analyze documents, or explore visual content. Add more images in subsequent turns to deepen or shift the discussion. Return at any time to check new photos.
Annotating images: To draw attention to specific areas, consider using photo editing/marking tools on the image before uploading. This can guide ChatGPT to focus on the elements you consider important.

Which memberships can use image inputs?
Plus and ChatGPT Enterprise.

Which models can accept image inputs?
GPT-4.

Which platforms can use image inputs?
All platforms, including web (chatgpt.com) and mobile (iOS / Android).

Will my images be used to improve your models?
OpenAI's approach to using content (including images) is the same for each product.
For ChatGPT Enterprise, OpenAI does not use content to train OpenAI's models.

How do I add image inputs in ChatGPT?
Make sure the model selector is set to GPT-4, then click the + icon in the prompt area to add an image input.

Do image inputs support video?
No, it cannot process video. It currently only supports processing still images.

What file types are supported?
PNG (.png), JPEG (.jpeg and .jpg), and non-animated GIF (.gif).

How many images can I upload at once?
The number of images you can add to a conversation depends on several factors, including image size and the amount of text accompanying the images. In general, if you encounter issues, consider reducing the number or size of images.

What is the size limit per image?
20MB.

How does the image feature handle blurry or unclear images?
If an image is blurry or unclear, the model will do its best to interpret it. However, the result may be less accurate.

What limitations should users be aware of when using ChatGPT with image inputs?
If you are using ChatGPT's new image input feature, be sure to note the following limitations:

Medical: The model is not suitable for interpreting specialized medical images such as CT scans and should not be used for medical advice.
Non-English: The model does not perform well on images containing text in non-Latin scripts (such as Japanese or Korean).
Large text: Zoom in on text in images to improve readability, but avoid cropping important details.
Rotation: The model may misinterpret rotated/upside-down text or images.
Visual elements: The model may have difficulty understanding graphics or text that vary in color or style (such as solid, dashed, or dotted lines).
Spatial: The model has difficulty with tasks requiring precise spatial positioning, such as identifying chessboard positions.
Accuracy: The model may generate incorrect descriptions or captions in some cases.
Shapes: The model has difficulty handling panoramic and fisheye images.
Metadata and resizing: The model does not process original filenames or metadata, and images are resized before analysis, which affects their original dimensions.
Counting: It can give an approximate number of objects in an image.