[gdal-dev] Question on building multi band composite and going back to RGB GeoTiff

[Joaquim Manuel Freire Luís]

Hi,

This is not a direct answer but maybe you’d be interested in this Julia package that addresses for example the automatic histogram stretching needed to produce nice true color images. Its target is Landsat and Sentinel data but there is nothing(?) that prevents it to work with data from other proveniences. Lots of GDAL work under the hood.

https://www.generic-mapping-tools.org/RemoteS.jl/dev/gallery/L8cube_img/remotes_L8_cube_img/

From: gdal-dev <gdal-dev-bounces at lists.osgeo.org> On Behalf Of Daniel Evans via gdal-dev
Sent: Monday, April 22, 2024 3:29 PM
To: Raley, Nathan <nraley at gaig.com>
Cc: 'gdal-dev at lists.osgeo.org' (gdal-dev at lists.osgeo.org) <gdal-dev at lists.osgeo.org>
Subject: Re: [gdal-dev] Question on building multi band composite and going back to RGB GeoTiff

Hi Nathan,

My initial suspicion might just be that the scaling the data provider did to go from the raw data to a human-eye-friendly RGB composite isn't the conversion you're assuming.

I know that with the data I regularly work with, it may be provided as Uint16, but the data range doesn't extend all the way to 65535.

If you compare the values in the separate R and G images to the RGB composite, do they appear to match the conversion you're assuming, or is there a different scaling (and possibly offset)?

Cheers,
Daniel

On Mon, 22 Apr 2024, 15:20 Raley, Nathan via gdal-dev, <gdal-dev at lists.osgeo.org<mailto:gdal-dev at lists.osgeo.org>> wrote:
I currently have a RGB geotiff composite image that has a Byte datatype.  I also have individual band images for R, G, Redge, and NIR that are UInt16 datatypes.  Since I’m missing the Blue band from the individual bands, I was attempting to extract the blue band from the RGB composite image, scale it up to the UInt16 datatype, and build a composite VRT with R, G, B, NIR, RE bands included in it.  I was then attempting to extract the RGB bands from the multispec VRT in order to see if the process was working as intended, but I’m getting an extremely blue image.

Can anyone shed some light as to what I may be doing wrong here?

I started by building a VRT for each band:
gdal_translate source_RGB.tif b.vrt -ot UInt16 -of VRT -b 3 -scale 0 255 0 65535
gdalbuildvrt -b 1 r.vrt source_R.tif
gdalbuildvrt -b 1 g.vrt source_G.tif
gdalbuildvrt -b 1 nir.vrt source_NIR.tif
gdalbuildvrt -b 1 re.vrt source_RE.tif

I then merged the VRTs:
gdalbuildvrt -separate multispec.vrt r.vrt g.vrt b.vrt nir.vrt re.vrt

I now have a multispec.vrt with the R, G, B, NIR, and RE bands, all with a UInt16 datatype.

Now, I attempted to rebuild a RGB GeoTiff from the composite VRT with something like:
gdal_translate -ot Byte -of GTiff -b 1 -b 2 -b 3 -scale 0 65535 0 255 -co PHOTOMETRIC=RGB multispec.vrt multispec.tif

Viewing the result in QGIS appears overly blue.  What am I doing wrong here?

Thanks,
Nathan

________________________________
This email transmission, including any attachments, is intended solely for the addressee named above, and may contain confidential or privileged information. If you are not the intended recipient, be aware that any disclosure, copying, distribution or use of the contents of this e-mail is prohibited. If you have received this e-mail in error, please notify the sender immediately by reply email and destroy the message and its attachments.
_______________________________________________
gdal-dev mailing list
gdal-dev at lists.osgeo.org<mailto:gdal-dev at lists.osgeo.org>
https://lists.osgeo.org/mailman/listinfo/gdal-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20240422/6a5322f1/attachment.htm>