Licensing options for this format begin with FME Professional Edition.
FME provides reader and writer access to data in ER Mapper’s Compressed Wavelet (ECW) format.
The ECW image format is used to compress very large images, without using intermediate tiling or intermediate disk storage. There is no inherent limitation to image size in the compression technology. However, any image saved in this format must be at least 128 x 128 pixels. Any non-eight-bit source data will be rescaled to eight-bit.
The ECW technique also takes advantage of CPU, L1 and L2 levels of cache to do its linear and unidirectional data flow through the Discrete Wavelet Transformations (DWT) process.
Many less common coordinate systems may not be mapped properly when writing to ECW files. Coordinate system and georeferencing transformations are read, but in some cases coordinate systems may not translate.
GCPs (ground control points) present along with a projection in an ECW file being read can either be applied to the data as an affine transformation, or stored as properties of the raster geometry. GCPs cannot be written to ECW.
The ECW writer supports the following option:
Compression Level: This option sets the desired compression level of the output image. This is expressed as a percentage of the original file size; for example, setting a compression level of 75 means that the output image will be approximately 75% smaller than an uncompressed image (i.e. it will be ~1/4 of the size).
This value only represents the targeted compression level, and this may not actually be achieved. Also note that ECW does not support lossless compression; that is, even if a compression level of 0 is specified, the output ECW file may not be identical to the original image.
Repeatedly reading and saving files may decrease the quality for this format. In addition, Workbench may not know which format data comes from, and whether it used lossy compression. If you know that source data comes from files with lossy compression, you should also set the compression level to 0. Although this compression still does not guarantee the exact copy of the source file, the output will be very close to the original.