Subscribe to this thread
Home - General / All posts - Image channels after georeference
jarrah15 post(s)
#09-Sep-21 02:22

When I import an image with RGB channels, it looks like the original image, with

R=Ch 0

G=Ch 1

B=Ch 2

After georeferencing it, the new image has the wrong colour mapping, with

R=Ch 2

G=Ch 1

B=Ch 0

Is this a bug? Should the channels be preserved through the georeferencing process?

Dimitri


6,713 post(s)
#09-Sep-21 03:38

No, that's not a bug, and yes, a departure from the default would be nice.

The georeferencing process creates a new image. It doesn't alter the old image. New images are created using default channel order, which is BGR, as it is for most image manipulation programs (what most people inaccurately refer to as "RGB"). If your image was the default BGR to start with then the channel order of the new image will be the same. If the image used some other channel order, as images often do, then the channel order of the new image will be different and you'll have to click the channel order button in Style (one click) to pick the ordering you want.

I'd prefer that transforms, georegistration, and other processes took a look at the starting image and altered the result from defaults to use whatever the starting image does. There are pluses and minuses to that approach, though. On the one hand I think it would be more convenient for casual use. On the other hand, instead of downstream processes (if you're automating workflow) being able to expect that all new images have a default channel order, they'd first have to take a look at input images to know what to expect.

lionel

769 post(s)
#13-Sep-21 22:08

Is there a SQL function that SPLIT the global image into 4 images : R,G,B, alpha ?

In a sense we consider that the Values in R ,G ,B alpha are in fact gray values to which we attach a color (RGB or alpha transparency)!

to modify the values of the selected pixels is to use what is called adjustment in photoshop

***

Is it correct to say that :

-projection is manage /store inside vector and raster component ?

-a pixel contains four channels = x4vector values

.................................................................=VectorMakeX4(<val>, <val>, <val>,<val>)

-an image contains four layers

no transform mathematic algo is use since each value are render on the each RGB diode !

Attachments:
adjustement.png
affinity_adjustment_curve.png
affinity_photo_adjustment.png


union, doc , APIand most important deepl & keyboard shortcut

Dimitri


6,713 post(s)
#14-Sep-21 10:06

Is there a SQL function that SPLIT the global image into 4 images : R,G,B, alpha ?

That's something you'd do using a combination of functions and SQL. To learn how, use the Transform pane, set to the Copy command to use "channel". Pick a channel like channel 0, choose new table for the result, provide a new image/table name, and then press the Edit Query button. That shows you the SQL that is used.

You can then adjust that to do all four channels in one big query.

to modify the values of the selected pixels is to use what is called adjustment in photoshop

See the examples on processing images with filters, such as this one. Those examples use filters, but they show the SQL machinery for splitting out channels, processing channels separately, and then putting the results back together.

I respectfully note, by the way, that your odds of getting useful advice would be better if you started a new thread, since your post isn't about duplicating the results of channel order after georeferencing, the topic of this thread.

Manifold User Community Use Agreement Copyright (C) 2007-2021 Manifold Software Limited. All rights reserved.