Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Saul-goode Luminosity Masks not working in 2.10.14
#1
Has something changed between 2.10.12 and 2.10.14 that would cause the Saul-Goode Luminosity Masks script to not work properly?

   

The attached .png is a shot of my channels window after running the script. The LL and LLL are completely black, while the D is totally white. I would expect the normal graduation from one step to the next.

I have downloaded a fresh copy of the script, run it through Tim's Constantize Filter and get exactly the same results.

Any help and advice much appreciated.
david.
Reply
#2
Is the image in high-precision linear light or 8-bit gamma corrected? The script was written well before GImp got high-bit-depth
Reply
#3
Ofnuts,
Thanks for the reply. The image is RGB 8bit integer, so not high-bit-depth.
david.
Reply
#4
(11-10-2019, 09:33 PM)david Wrote: Has something changed between 2.10.12 and 2.10.14 that would cause the Saul-Goode Luminosity Masks script to not work properly?



The attached .png is a shot of my channels window after running the script. The LL and LLL are completely black, while the D is totally white. I would expect the normal graduation from one step to the next.

I have downloaded a fresh copy of the script, run it through Tim's Constantize Filter and get exactly the same results.

Any help and advice much appreciated.
david.

Downloaded latest version, and tested it with a photo of mine, and it has the same problem... Note that the median masks seem totally wrong.  But since it D is wrong and is used to compute many thing, that could be the root cause.
Reply
#5
I think it must be a bug. Is working in Gimp 2.10.12 but not in Gimp 2.10.14.

Same result as David, Channel D should be L inverted but it is white. That then gets passed on to L and LL which end up black.

The code is near the beginning of the script lines, 35 to 43 but beyond my meagre scripting talents
Reply
#6
I can reproduce the problem by playing the script manually... Looks like a Gimp bug



Reported: https://gitlab.gnome.org/GNOME/gimp/issues/4218
Reply
#7
rich & Ofnuts,

Thanks for confirming the problem. I originally thought it may be "finger-trouble" on my part, so I spent a few hours trying to understand script, but way beyond me!
I don't know what we would do without the help you give to us mere mortals!

david.
Reply
#8
Another version of the luminosity mask can be found here:
https://gimplearn.net/viewtopic.php/Incr...?f=10&t=81

I haven't installed 2.10.14 yet so can't test it in that.

Note the comment about black masks from the .scm file:

; following steps in tutorial by Pat David
; found at link: http://www.gimp.org/tutorials/Luminosity_Masks/
; This script will generate 9 extra layers named "L","LL","LLL","D","DD","DDD","M","MM","MMM"
; where L stands for Light-tones, D stands for Dark-tones, M stands for Mid-tones
; these layers will be masked as instructed in the tutorial
; so that the user can manipulate these layers at free will.
; Note: not mentioned in tutorial is how to create MM and MMM layers without them having black masks
; according to saul goode's script found at http://chiselapp.com/user/saulgoode/repo...99c37cda66
; MM is created by inverting LL then SUBTRACT DD
; MMM is created by inverting LLL then subtract DDD
; author: Tin Tran
; date: 2014
Reply
#9
Blighty,

Thanks for your input.
I have installed the Tin Tran Luminosity Masks (in 2.10.14) and it works as advertised.
It would appear that there is something in the Saul Goode code, not present in the Tin Tran version, which is causing the problem.

david.
Reply
#10
(11-11-2019, 01:31 PM)david Wrote: Blighty,

Thanks for your input.
I have installed the Tin Tran Luminosity Masks (in 2.10.14) and it works as advertised.
It would appear that there is something in the Saul Goode code, not present in the Tin Tran version, which is causing the problem.

david.

To create the first channel, the SG script doesn't create a channel from scratch but saves the selection. And there is a new bug with channels created that way.
Reply


Forum Jump: