{"id":"image-jais","mimeType":"image/jais","name":"JPEG AI Image Sequence","shortName":"JPEG-AI-Sequence","category":"Graphics \u0026 Learning-Based Compression","extensions":[".jais"],"description":"The image/jais MIME type represents a JPEG AI coded image sequence. Standardized as part of ISO/IEC 6048 (ITU-T T.840), JPEG AI is the first international image coding standard based entirely on an end-to-end deep learning (neural network) architecture. While the image/jaii MIME type is used for single still images, image/jais encapsulates multiple frames—such as animations, burst photography, or volumetric slices—within an ISOBMFF container. Like its still-image counterpart, it provides a compact compressed-domain representation that can be directly consumed by computer vision algorithms.","seo":{"title":"image/jais MIME Type | Complete Developer Reference for JPEG AI Sequences","description":"Complete reference for the image/jais MIME type: magic bytes, ISOBMFF sequence container, neural network compression, and JPEG AI specifications.","faqs":[{"question":"What is the difference between image/jaii and image/jais?","answer":"Both MIME types use the JPEG AI neural network-based compression standard. However, image/jaii (JPEG AI Image) is strictly for single still images, while image/jais (JPEG AI Image Sequence) is used for files containing multiple frames, such as animations, burst photography, or focal stacks."},{"question":"How does JPEG AI compress image sequences?","answer":"JPEG AI utilizes variational autoencoders with hyperpriors. For sequences, the format relies on the ISOBMFF (HEIF) container to manage timing, tracks, and multiplexing of the individual deep-learning-compressed frames, allowing efficient storage and AI-driven machine vision processing."},{"question":"What are the magic bytes for a JPEG AI Image Sequence?","answer":"Because it uses the ISOBMFF container, a valid image/jais file begins with the standard 'ftyp' box, followed by the 'jais' major brand. The ASCII sequence 'ftypjais' will typically start at byte offset 4."}]},"index":{"compression":["Lossy (Variational Autoencoder / Neural Network)"],"browserSupport":{"score":1,"label":"None"},"commonUses":["Machine Vision and AI Video/Burst Workflows","Cloud Storage Animation Optimization","Focal Stacks and Medical Volumetric Slices","Visual Surveillance Sequences"]},"technical":{"magicBytes":{"hex":".. .. .. .. 66 74 79 70 6A 61 69 73 (Size + ftypjais)","ascii":"....ftypjais"},"container":"ISOBMFF / HEIF Image Sequence","compressionAlgorithm":["Deep Neural Network (Variational Autoencoder with Hyperpriors)"],"colorDepth":["8-bit","10-bit","12-bit","16-bit","Wide-gamut \u0026 HDR"],"transparency":true,"animation":true,"metadata":{"exif":true,"xmp":true,"iccProfile":true},"alphaChannel":true,"progressiveLoading":true,"hdrSupport":true},"history":{"developer":"Joint Photographic Experts Group (ISO/IEC JTC 1 / ITU-T)","released":"2025","specification":"ISO/IEC 6048-5 (File Format) / ITU-T T.840.5"},"browserSupport":{"chrome":false,"chromeVersion":"None","firefox":false,"firefoxVersion":"None","safari":false,"safariVersion":"None","edge":false,"edgeVersion":"None","internetExplorer":false},"softwareSupport":[{"name":"JPEG AI Reference Software (T.JPEG-AI-RS)","logo":""}],"conversion":{"convertTo":["gif","webp","avifs"],"convertFrom":["gif","webp","avifs","heics"]},"developer":{"httpHeaders":[{"header":"Content-Type","value":"image/jais"}],"htmlAccept":"image/jais, .jais","fileDetection":["Check the standard ISOBMFF header. The first 4 bytes indicate the length of the 'ftyp' box.","Verify bytes 4-7 are exactly 'ftyp' (66 74 79 70).","Verify the major brand or one of the compatible brands within the 'ftyp' box contains the 'jais' brand string (6A 61 69 73)."]},"prosCons":{"pros":["Extends the state-of-the-art compression efficiency of JPEG AI to multi-frame image sequences and short animations","Allows machine vision algorithms to process temporal sequences (like surveillance bursts) directly in the compressed latent domain, saving immense computational power","Utilizes the robust, industry-standard ISOBMFF container for metadata, timing, and multi-track encapsulation"],"cons":["Lacks native decoding support in current web browsers, operating systems, and video players","Decoding neural-network-compressed sequences requires significant compute overhead on hardware lacking dedicated AI accelerators (NPUs/GPUs)","Adoption is hindered by competition with established sequence formats like AVIF, HEVC (HEICS), and modern video codecs"]},"specifications":{"url":"https://jpeg.org/jpegai/","rfc":"","mimeAliases":[]},"developerSnippets":{"python":"import magic\n# Libmagic may identify the sequence merely as a generic ISOBMFF container (e.g., 'video/mp4' or 'application/octet-stream')\nprint(magic.from_file('sequence.jais', mime=True))","nodejs":"const fs = require('fs');\n// Manually detecting the JPEG AI Sequence brand in the ISOBMFF ftyp box\nconst buffer = Buffer.alloc(12);\nconst fd = fs.openSync('sequence.jais', 'r');\nfs.readSync(fd, buffer, 0, 12, 0);\nfs.closeSync(fd);\nconst isFtyp = buffer.toString('ascii', 4, 8) === 'ftyp';\nconst isJais = buffer.toString('ascii', 8, 12) === 'jais';\nconsole.log(isFtyp \u0026\u0026 isJais);","go":"import (\n    \"bytes\"\n    \"net/http\"\n)\n// Custom ISOBMFF detection for the JPEG AI Sequence brand\n// func isJPEG_AI_Seq(buf []byte) bool {\n//     return len(buf) \u003e 12 \u0026\u0026 string(buf[4:8]) == \"ftyp\" \u0026\u0026 string(buf[8:12]) == \"jais\"\n// }","php":"// Standard PHP setups may default to identifying the ISOBMFF container rather than the specific image/jais MIME.\n$mime = mime_content_type('sequence.jais');\necho $mime;"},"relatedFormats":[]}