ideas

Ideas

• Lightweight, should be fast to start and open files
• Written in Python unless speed becomes an issue
• Look much like the Imagine Viewer. Other alternative is to look more like ENVI with its windows at multiple scales. Imagine more familiar to us, we usually have pyramid layers to do the Imagine stuff properly, and geolinking much easier in Imagine.
• Do we need to be able to layer up multiple images within a single viewer? Probably not, at least initially.
• Efficient and easy method of defining the image bands being displayed - particularly important for hyper-spectral or datasets with a large number image bands (e.g., texture filtered layers).
• Be aware of and display image band names
• Good options for tiling multiple geolinked viewers on screen (e.g., n viewers on an n x m grid)
• Be GDAL specific. Support all GDAL formats. Use some features (pyramid layers, RAT's) where available but still work when they are not. Allow plugins access to the GDAL dataset object.
• Make use of the gdal virtual raster pyramids not just erdas imagine.
• query value crosshair
• Support Raster Attributes properly. Have a sub window with the table, highlight rows when value queried. Highlight parts of image with value when row selected etc. Like Imagine. Editing? (Pete: editing should be for version 2)
• Do stretches properly (unlike QGIS!)
• Use a LUT for the stretches as it will provide a more flexible framework on which to build in the future - look into using QColormap and indexed pixels, might save memory also??
• Geolinking. Do this better than Imagine by making all windows at the same scale - more like the QGIS geolinking plugin.
• Plugins. Make the system easily extendible (and embeddable) by Python so we can experiment with new features etc. Maybe this should be version 2.0.
• Vectors. Allow overlay of vectors (maybe version 2) but not really much other vector functionality.
• Not try to be like Arc - do rasters properly but not processing, making maps etc.
• Rely on existing Qt infrastructure to - do things the 'Qt way'

Versions for Development

• • Possible development stages:**

Version 1: Single viewer which is fast to display images with all the stretching sorted etc.

Version 2: Geolinking, pixel values interrogation, spectral profiles etc, flickering between datasets

Version 3: Attribute tables.

Version 4: More advanced attribute tables - such as querying and editing.