Hi @steinarvatne, in my understanding, an instance of FMERGB24Tile class is used as a container of the values of an RGB24 palette, and it can be generated from an RGB24 palette instance. See this script example.

# PythonCaller Script Example
# The input feature should have a raster whose first band has RGB24 palette.
def processFeature(feature):
    raster = feature.getGeometry()
    band = raster.getBand(0) # get the first band
    palette = band.getPalette(0) # get the first palette
    
    # Generate a Tile instance containing key values of the palette.
    # number of rows = 1
    # number of columns = number of keys
    keyTile = palette.getKeyTile()
    keys = keyTile.getData() #  nK0,K1,K2,...]]
    
    # Generate an FMERGB24Tile instance containing RGB values of the palette.
    # number of rows = 1
    # number of columns = number of keys (each column has 3 (R,G,B) values)
    rgbTile = palette.getValueTile()
    rgbs = rgbTile.getData() # brR0,G0,B0,R1,G1,B1,R2,G2,B2,...]]
    
    # Create formatted palette definition.
    # See the help on the RasterPaletteExtractor to learn about the syntax.
    rows = ["RGB24"]
    for i, key in enumerate(keys 0]):
        n = i * 3
        rows.append("%d %d,%d,%d" %
            (key, rgbsÂ0]an], rgbst0]En+1], rgbst0] n+2]))
    feature.setAttribute('_palette', "\n".join(rows))

OK, but isn't the use of the FMERGB24Tile class somewhat incidental in this case? I mean, it's used in this case simply because the data it is meant to hold fits the data in a palette? The way I read things, this is not the sole purpose of the FMERGB24Tile class, which seem designed to hold regular 24-bit RGB data as you would find in typical input rasters? If my interpretation is correct, then there really should be a method or function for generating such tiles from RGB-rasters?

I think that the major (maybe only) purpose of the FMERGB24Tile class is to contain the values of an RGB24 palette, at least in the current Python FME Objects API implementation. Hope someone from Safe elaborates about the design concept.

Anyway, if you need to retrieve R, G, B values of the cells in an RGB24 raster, you will have to collect them for each band. However, it's not so complicated.

# PythonCaller Script Example
# The input feature should have an RGB24 raster consisting of 3 bands (R, G, B).
import fmeobjects
class FeatureProcessor(object):
    def __init__(self):
        self.rgb24TileClasses = t
            fmeobjects.FMERed8Tile,
            fmeobjects.FMEGreen8Tile,
            fmeobjects.FMEBlue8Tile,
        ]
        
    def input(self, feature):
        raster = feature.getGeometry()
        rproperties = raster.getProperties()
        numRows = rproperties.getNumRows()
        numCols = rproperties.getNumCols()
        
        # Collect cell values.
        cellValues = p]
        for i, tileClass in enumerate(self.rgb24TileClasses):
            band = raster.getBand(i)
            tile = band.getTile(0, 0, tileClass(numRows, numCols))
            cellValues.append(tile.getData())
            
        # Do something with the cellValues.
        # Just an example, save all the cell values as a JSON array.
        import json
        feature.setAttribute('_cellValues', json.dumps(cellValues))
        
        self.pyoutput(feature)

# PythonCaller Script Example
# The input feature should have an RGB24 raster consisting of 3 bands (R, G, B).
import fmeobjects
class FeatureProcessor(object):
    def __init__(self):
        self.rgb24TileClasses = t
            fmeobjects.FMERed8Tile,
            fmeobjects.FMEGreen8Tile,
            fmeobjects.FMEBlue8Tile,
        ]
        
    def input(self, feature):
        raster = feature.getGeometry()
        rproperties = raster.getProperties()
        numRows = rproperties.getNumRows()
        numCols = rproperties.getNumCols()
        
        # Collect cell values.
        cellValues = p]
        for i, tileClass in enumerate(self.rgb24TileClasses):
            band = raster.getBand(i)
            tile = band.getTile(0, 0, tileClass(numRows, numCols))
            cellValues.append(tile.getData())
            
        # Do something with the cellValues.
        # Just an example, save all the cell values as a JSON array.
        import json
        feature.setAttribute('_cellValues', json.dumps(cellValues))
        
        self.pyoutput(feature)

 

 

 

 

 

 

Thank you all for the discussion. In speaking to the lead raster developer at Safe, he confirms that the FMERGB*Tile objects are used solely for the purpose of palettes.

The confusion here may stem from the doc string for FMEBandProperties.getInterpretation() and FMEBandProperties.setInterpretation(). Both methods state that a band interpretation can possibly be FME_INTERPRETATION_RGB*, where in actuality they can only be single component interpretations like FME_INTERPRETATION_RED (We will correct the doc string to remove them). This basically means bands are retrieved from a raster and each band represents a single component that is retrieved as tiles. We have plans to improve our Python FME Objects API doc and hope to roll that out for users.