Wrong auto-scaling by 'Three Color (RGB)Image (Auto-scale)'

[hproe]

Hi -

Not sure whether things ovedone here, or these are bugs.

One of my email correspondents tried to emulate with 'Three Color (RGB)Image (Auto-scale)'
the DUST method of my 'RGB-Eumetsat' plugin (part of the the 'Miscellaneous' plugins). She found 2 issues that I can confirm with any 1.4beta:

1. The red and green channels are assigned to channel differences with the built-in formula 'Simple difference a-b'. I seem to remember that well before 1.4beta the 'a' and 'b' values would be requested immediately after okaying each call to the formula. However, now the channels for the 2 differences are called up in sequence only after the blue channel has been assigned as well. The RGB comes then out fine.

2. When applying the ranges and gammas for the Dust RGB, the resulting RGB (see attached RGBcomposite.png) is not what the 'DUST' is correctly displaying (see attached DustRGB). At first sight there is too much red. Also here, I seem to remember the somewhere in the past (when writing the RGB formuae) the display was correct.

br, HP

[tdrink]

HP,

I think you still have to set the ranges/gamma manually even with auto-scale according
to the product convention. You may also have to hit 'return/enter' on each text field
in the gui.

Tom

[hproe]

Tom -

I did this all ...

HP

[Rick]

Hi HP,

How are they trying to emulate the dustRGB formula using RGB auto-scale? Have they created their own formula? If so, please provide the formula.

Rick

[hproe]

Tom & Rick -

Point 1 of my original submission is solved. Not having used TomR's RGB method for quite a while, I completely forgot that the gammas used there are the inverse of the gammas given by in EUMETSAT's definitions of standard RGBs. If you look up e.g. the DUST method in my RGB plugin there is a gamma of 2.5 to be applied to the green channel, which then has to be 0.4 in TomR's layer control).

Point 2 still is valid, though. It is really vexing that the channel attribution dialogue for the 'simple difference' method comes only after having gone through all 3 colour channels, is not appearing immediately after having it chosen. This definitely has not been so in the past.

HP

[Rick]

Hi HP,

Using 1.2, I created the following formula:

a+b+c

From the popup field selector, for 'a' and 'b', I selected Formula, Miscellaneous, Simple Difference. For 'c' I selected the temperature value for a specific wavelength. After clicking 'ok', I was prompted to enter the values for the Simple Difference for parameter 'a' and then the same for parameter 'b'.

This is also what I saw using the latest nightly from 1.4. Let me know if I'm misunderstanding your question.

Rick

[hproe]

Hi Rick -

No, You understood my (rather minor) problem. My memory is obviously wrong, as you see the same behaviour already in 1.2. I have no very strong feeling for changes then. So, let it be.

HP