Image Metadata Viewer
Introduction
Image Inspector currently supports PNG and JPG formats. It can read prompts and ComfyUI workflows from images generated by Stable Diffusion (note: must be SD-generated images).
JPG File Format
JPG format images are composed of a complex marker system, including:
- SOI (Start of Image) and EOI (End of Image): Mark the beginning and end of the image.
- DQT (Define Quantization Table): Defines quantization tables, crucial for image compression.
- SOF (Start of Frame): Describes basic image properties (width, height, color, etc.).
- DHT (Define Huffman Table): Defines Huffman coding tables for data encoding.
- SOS (Start of Scan): Marks the start of image or image component scanning.
- APP1: This marker is typically used to store EXIF (Exchangeable Image File Format) information, including but not limited to capture time, camera settings, thumbnails, and other metadata.
PNG File Format
PNG format images start with an 8-byte signature followed by multiple data chunks, mainly including:
- IHDR: Image header chunk, containing basic image information (width, height, bit depth, color type, etc.).
- PLTE: Palette chunk, used for indexed color images.
- IDAT: Image data chunk, containing actual image data.
- IEND: Image end chunk, marking the end of the file.
Additionally, PNG format supports text chunks for metadata:
- tEXt: Stores uncompressed text information in ISO/IEC 8859-1 encoding.
- zEXt: Similar to tEXt, but with compressed content.
- iTXt: Supports internationalized text chunks in UTF-8 encoding, can contain compressed or uncompressed text, and can specify language and translation.
When generating images through Stable Diffusion WebUI and ComfyUI, prompt
information is stored in the image metadata. In JPEG images, this information is located in the EXIF segment, while in PNG format it is stored in iTXt, tEXt, or zEXt chunks.
😉 腾讯云产品特惠热卖,戳我领取!