Image MIME Reference
HSJ2
JPEG 2000 Image Sequence (Obsolete)
image/hsj2
VS
JPEG AI Image Sequence
image/jais
JPEG-AI-Sequence
A complete technical comparison of JPEG 2000 Image Sequence (Obsolete) and JPEG AI Image Sequence — covering compression, feature support, browser compatibility, magic bytes, and when to choose each format.
Equal browser support
Feature Support
| Feature | HSJ2 | JPEG-AI-Sequence |
|---|---|---|
| Transparency (Alpha) | Yes | Yes |
| Animation Support | Yes | Yes |
| Progressive Loading | Yes | Yes |
| HDR Support | Yes | Yes |
| EXIF Metadata | Yes | Yes |
| ICC Color Profile | Yes | Yes |
JPEG 2000 Image Sequence (Obsolete)
image/hsj2
Extension
.hsj2
Container
HEIF / ISOBMFF (ISO/IEC 15444-16 Obsolete)
Compression
Lossy / Lossless
Algorithm
Motion JPEG 2000 (ISO/IEC 15444-3)
Color Depth
8-bit, 10-bit, 12-bit, 16-bit
Developed by
ISO/IEC JTC 1 / ITU-T
Released
2019
Magic Bytes
XX XX XX XX 66 74 79 70 68 73 6A 32
JPEG AI Image Sequence
image/jais
Extension
.jais
Container
ISOBMFF / HEIF Image Sequence
Compression
Lossy (Variational Autoencoder / Neural Network)
Algorithm
Deep Neural Network (Variational Autoencoder with Hyperpriors)
Color Depth
8-bit, 10-bit, 12-bit, 16-bit, Wide-gamut & HDR
Developed by
Joint Photographic Experts Group (ISO/IEC JTC 1 / ITU-T)
Released
2025
Magic Bytes
.. .. .. .. 66 74 79 70 6A 61 69 73 (Size + ftypjais)
Browser Support Comparison
| Browser | HSJ2 | JPEG-AI-Sequence |
|---|---|---|
|
|
✕ | ✕ |
|
|
✕ | ✕ |
|
|
✕ | ✕ |
|
|
✕ | ✕ |
|
|
✕ | ✕ |
HSJ2
HSJ2 Strengths
- Historically attempted to combine Motion JPEG 2000 sequences with the flexible metadata capabilities of the HEIF container
- Supported mathematically lossless compression and scalable wavelet decoding
Limitations
- Officially deprecated by the standardization body (ISO/IEC JTC 1)
- Completely superseded by image/hej2k, rendering hsj2 files obsolete for any new development or archiving
- Zero web browser compatibility and negligible modern software support
JPEG-AI-Sequence
JPEG-AI-Sequence Strengths
- 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
Limitations
- 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
When to choose which format
HSJ2
Use HSJ2 when…
- Legacy Archival
- Deprecated Workflows
JPEG-AI-Sequence
Use JPEG-AI-Sequence when…
- Machine Vision and AI Video/Burst Workflows
- Cloud Storage Animation Optimization
- Focal Stacks and Medical Volumetric Slices
- Visual Surveillance Sequences