Radelta Web
Fast, lossless, microscopy image compression in your browser with local (no upload) processing
TIFF to Radelta
Operation
Input
Drop a TIFF file here
or select a file from your computer.
Maximum concurrent files. More workers use more memory.
Output folderNo output folder selected
Saving to a folder requires desktop Chrome or Edge, over HTTPS or localhost.
Output names: image.tif → image.tif.rdlt. Subfolder paths are preserved.
0 TIFF files
Showing the first 300 files. All selected files will be processed.
Lossless compression
Compare decoded chunks and pyramid images with their input.
Preparing…
Processed—
Output size—
Compression ratio—
Select an input file.
Output:
.rdlt
Processing takes place in this browser. Image data is not uploaded.
Supported formats and limits
- Compression input
- Unsigned 16-bit TIFF or BigTIFF volumes, including OME-TIFF and ImageJ hyperstacks with Z, channel and time dimensions, and SubIFD pyramids. Channels must be stored as separate grayscale planes. Plane order is detected from the metadata; TIFFs without recognized OME or ImageJ dimensional metadata are treated as Z stacks. Both byte orders, strips and tiles are supported, with uncompressed, LZW, Deflate or PackBits pixels and horizontal prediction.
- Multifile OME-TIFF
- By default, each source TIFF produces its own .rdlt file. Original paths and OME references are preserved, and companion XML is copied unchanged. Missing members produce warnings without blocking safe per-file compression. Keep the fileset together when decoding. Consolidation is an explicit option for complete acquisitions.
- Decoding output
- Lossless and calibrated-lossy Radelta files are decoded to uncompressed BigTIFF. Channel/time dimensions, known TIFF descriptions, calibration, plane order and nested SubIFD pyramids are retained. Reduced images are always stored losslessly. Files are compatible with the Radelta CLI and ImageJ plugin.
- Unsupported input
- RGB/interleaved samples, planar multisample pages, multiple image series, top-level reduced-image chains, and nested EXIF/GPS directories. Contiguous ImageJ stacks require uncompressed strips. Linked image chains inside SubIFDs are also unsupported.
- Memory limits
- The former 128 MiB plane/block limit is removed. A complete plane or codec block and its working buffers must fit in browser memory. Each individual WASM buffer must be smaller than 2 GiB; usable image sizes may be lower. Descriptive tags remain limited to 16 MiB and directory indexes to 32 MiB. Pyramid data is processed incrementally.
- Browser support
- Desktop Chrome or Edge over HTTPS or localhost supports direct file and folder output. Other browsers use a download fallback, limited to 256 MiB of uncompressed pixels for compression or 256 MiB of final TIFF output for decoding.