Calling all Pixinsight experts

[alan4908]

Over the past year I'm slowing finding that I'm transitioning from PS and CCDstack to Pixinsight for my image processing.  However, I only do this for a specific function if I can convince myself that I like the result better in PI that the PS or CCDstack equivalent. 

One item that I like in PS is the ability to boost colour in the LAB colour space.  

In PS, this can be done quite easily without impacting the colour balance by creating a symmetric "S" curve on either the a or b channel.  Unfortunately, I cannot see a way to perform the equivalent function within PI.  If I attempt to do the same thing in PI via the Curves Transformation process, then the colour balance of the image is impacted. 

So, how do I work in the Pixinsight LAB colour space for the purposes of increasing saturation without impacting colour balance ?

Alan

[Steve 1962]

I’m very surprised that applying a Curve Transformation in PI is affecting your colour balance. Are you sure that you’ve got the S (saturation) icon selected??

[RolandKol]

I am not a PI ninja... (yet!)

However, I found it very nice and handy to use Color MASK Script and adjust curves, including the Saturation, using mask per each color range.

Plus, if you feel, too many stars were included into the mask,

you can create star mask and deduct it from the Colour mask using PixelMath, 

so you will have Mask without stars, just for the nebula's particular color range.

There is a very nice star mask script here http://www.skypixels.at/pixinsight_scripts.html 

So you can adjust star colors and nebulae separately, without affecting the whole image.

[alan4908]

2 hours ago, Steve 1962 said:

I’m very surprised that applying a Curve Transformation in PI is affecting your colour balance. Are you sure that you’ve got the S (saturation) icon selected??

Many thanks for the reply. As I mentioned above I want to work in the LAB not the RGB colour space. This means that you select either the L, a or b channels in Curves transformation.  You can then adjust each channel by selecting either L, a or b and then adjusting the curve. 

If you select S and then adjust the curve you are working in the HSVL colour space to adjust the saturation levels. This works as expected, eg the  adjustments do not create a colour imbalance.

 

2 hours ago, RolandKol said:

I am not a PI ninja... (yet!)

However, I found it very nice and handy to use Color MASK Script and adjust curves, including the Saturation, using mask per each color range.

Plus, if you feel, too many stars were included into the mask,

you can create star mask and deduct it from the Colour mask using PixelMath, 

so you will have Mask without stars, just for the nebula's particular color range.

There is a very nice star mask script here http://www.skypixels.at/pixinsight_scripts.html 

So you can adjust star colors and nebulae separately, without affecting the whole image.

Thanks for the tip I shall investigate the color Mask script !  

However, I really want to work in the Lab colour space.

Alan

 

[RolandKol]

I am almost 100% sure, PI Curves are not the same as PS...

Different tools.

Never used PS in LAB mode, cannot compare, but PI actually has something similar,  - probably:

https://pixinsight.com/doc/tools/CurvesTransformation/CurvesTransformation.html

[RolandKol]

P.S.

Actually, saw quite a lot of people, who end up using both... Pi (NB or RGB combination, noise reduction + de-liner) and PS to finalize.

[alan4908]

16 minutes ago, RolandKol said:

I am almost 100% sure, PI Curves are not the same as PS...

Different tools.

Never used PS in LAB mode, cannot compare, but PI actually has something similar,  - probably:

https://pixinsight.com/doc/tools/CurvesTransformation/CurvesTransformation.html

Hi Roland

Yes - this is the tool documentation that I'm referring to in my original post.  As you indicate you can select the LAB mode in PI by clicking on the either the L, a or b buttons which are circled in your diagram.   If you read the documentation, then it does imply that the two tools are very similar to PS LAB mode. The only difference I can see is that in PI the x and y axis goes between 0 and 1, where in PS the x and y axis go from -127 to +128.   Obviously, there is a deeper difference which I cannot see, since introducing a symmetric "S" curve into the a or b channels introduces a colour imbalance into the image in PI but it doesn't in PS. 

Anyway, back to my central question - given that LAB mode has been implemented in PI - how do you use LAB mode in PI to increase colour saturation without creating a colour imbalance ?

Alan

[Scooot]

Hi Alan,

I don’t know the specific answer to your question but I wonder whether you’ve seen this and if it’s any help.

https://pixinsight.com/doc/legacy/LE/17_curves/curves_transforms/curves_transforms.html

If I understand it correctly using the S for saturation enhancement does also make an LAB transformation whilst preserving the colour balance “For each pixel, two color space transforms are carried out: to the HSV and to the CIE L*a*b* spaces.

 

[alan4908]

19 minutes ago, Scooot said:

I don’t know the specific answer to your question but I wonder whether you’ve seen this and if it’s any help.

https://pixinsight.com/doc/legacy/LE/17_curves/curves_transforms/curves_transforms.html

If I understand it correctly using the S for saturation enhancement does also make an LAB transformation whilst preserving the colour balance “For each pixel, two color space transforms are carried out: to the HSV and to the CIE L*a*b* spaces.

 

Thanks for that Richard. Yes, I've seen this documentation.

My understanding of the S button in CurvesTransform is that this increases saturation for unsaturated pixels but not saturated ones, which seems similar to the Vibrance colour function in PS.  As you say, PI have implemented this is to involve conversion into the HSV and CIE LAB colour spaces. So, this doesn't appear to be the same as increasing the saturation by a contrast curve in LAB mode.  To explain a little more, if you click on the b button and then click on the point 0.5, 0.5 and then pull down on the lower half of the curve you will end up with a symmetric "S" curve within the LAB  b space.  You can see the effect on the image by selecting the real time preview. If you are familiar with PS you can do a similar thing but this time you need to initially click on 0, 0.  Within PI you can see that the colour saturation of the image has increased but the colour balance is impacted, within PS it is not. 

Alan

[alan4908]

Just an update anyone that is interested 

I've had an answer from Juan Conejero to my original question - if you want to use a colour space that is designed to be perceptively uniform for human vision, then the recommendation is to use the LCH space rather than LAB space.  Apparently, this is equivalent to LAB, the difference is that it is constructed in a polar co-ordinate rather than a rectangular co-ordinate system. To use the LCH space within CurvesTransformation, you click on the "c" button and adjust the curve to increase/decrease colour saturation. This seems to work and preserves the colour balance of the image.

Alan