Topic: Curvemeister's beta little bugs ?

[Frits]

Mike,
I was busy with the horse on page 66 of the Color lab book.
I had set a neutral on the street.
also I had set the info on from the neutral.
You can set then the neutral on and off.
When I go for instance  for example to the A channel I can see the neutral.
When I select Reset this curve the neutral vanish from channel A.
that is correct.
But the info says the neutral is on. Indeed he is on in channel B, but not in A.
I must set the info selecting swich off  and on again to have the neutral back again in the A channel.
Is that correct or perhaps a little bug!

Frits

[curvemeister]

Yes - this can be thought of as a bug. 

Resetting a particular curve always sets it to a straight line, even if there is a neutral in effect.  Moving the neutral, or selecting the color space will re-establish the same neutral.

The reason I say "can be thought of as a bug" is that I can't think of a way to improve the behavior without making things more complicated.  If you, or anyone else reading this, has any thoughts, let me know.

[Frits]

Yes - this can be thought of as a bug. 

Resetting a particular curve always sets it to a straight line, even if there is a neutral in effect.  Moving the neutral, or selecting the color space will re-establish the same neutral.

Mike,

the best thing you can do, is tell it in the manual. Not that it is a sort of bug, but the way you must handle it.

Frits

[Frits]

Mike,

i have found a new bug.
I had found it yesterday, but I could not first redo it.
But now I know how I did it.
I  am still busy with the horse.
And I was repeating doing what is in the book on another way.
I make use of the pin grid lower left  on the same time in A and B.
then I use the slider to make the upper curve steeper.
suddenly the curve near the 127 mark goes walking.
This happened on both channels A and B.
Strange. I could correct that with the hand, but it is not mentioned so.

Frits

[Frits]

Again a bug !

When I was in Lab mode and the hue clock was in RGB- display, there is a error.
He gives the wrong measurements

Frits

[curvemeister]

This looks correct to me.

Red with a small amount of green added is orange, and dark orange is brown.

[Frits]

This looks correct to me.

Red with a small amount of green added is orange, and dark orange is brown.

I am forgotten i have it about the brown color.
the RGB values must be R=129 G=96 B=82, but the clock said  R=50,6   G=37,6 and B=31,8.

[Frits]

I am forgotten i have it about the brown color.
the RGB values must be R=129 G=96 B=82, but the clock said  R=50,6   G=37,6 and B=31,8.

ADDUM,
When you check it with others colors, the same error, wrong numbers!!
(The clock displays in RGB -mode, but you are in LAB-mode.)

Frits

[curvemeister]

This is actually a feature - click on the menu button on the hue clock that appears in the curve window, and you can set its options separately.

Mike

[Frits]

This is actually a feature - click on the menu button on the hue clock that appears in the curve window, and you can set its options separately.

Mike

mike I have made a picture of all errors with this.
When I am in a mode for instance lab, you  can set the clock to another diplay.
That is very nice. But the values are not good.
ThaT I can set the clock display also to the mode, I knew this from the beginning.

See the picture speaks for himself.

Frits

[curvemeister]

Hi Frits,

I've looked over your images carefully, and I thank you for the information.  You have checked these more thoroughly than any of the beta testers did, and found at least two new bugs!

All of the color values, except two, are correct. Part of the problem is that I have not explained adequately.

In some cases, where the % sign appears, the color values are displayed from 0(light) to 100(dark) instead of 0(dark) to 255(light).  This option is controlled in two ways: by clicking the diamond shape in the margin of the curve, and by setting it in the Units options.

In some cases there may be a qualtization error that can cause a color value to be off by one, and I believe this is what happened when you displayed an RGB value with the curves set to Lab mode.

The RGB value of 50,37,31, and the similar wgCMYK values are incorrect.  This happens because the values are mistakenly displayed as percentages, instead of the correct 0-255 range.  This will be fixed in the next release, and I thank you once again for finding this.

In your last example, you refer to the hue clock remaining, even after the option is changed to display no hue clock.  This will also be fixed in the next release.

Mike

[Frits]

Mike,

again a bug.

difficult to explain.

See the picture and numbers.

I was experimenting wit Targetting the whitepoint RGB ( 255,255,255)
When you not change anything, there is no problem.
The whitepoint spot is pulled up to 255, 255 ,255. See 1 and 2
--
Now I  delete the old white point and make the target 222, 222, 222. and click on apply and ok! See 3.

You see on picture 4 that nothing changed. It stays 255,255,255.

Only when I go out of Curvemeister and start Curvemeister again , then the whitepointvalue is changed to  222,222,222 in the info of the Whitepoint! see pict 5

Frits

[Frits]

Mike,

I am always curious about why, how and what about things.
Reading the book Photo lab Colors  I was curious about the exactness of Curvemeisters.
Just because there is speaking of Decimal places, witch suggest a high precision.

I made with the computergenerated chart what you can find on http://www.efg2.com/Lab/Graphics/Colors/ColorCharts.ZIP a output in RGB with photoshop elements 4.0 and Curvemeister.
Photoshop gives only RGB output.  Curvemeister gives also RGB output and also in 2 decimal places.
Very good.
But when I set on the Lab mode and also output with 2 decimal places , then I get a curious output.
The a and b are nice, xx,00 but the "L", changes.
When I compare this with a precision CIE Calculator with you can find on :
http://www.brucelindbloom.com/ColorCalculator.html
there is also a difference on all output compared to what Curvemeister found.
Not much , but there is.
Only the red ones are equal.
I have chose another whitepoint or colourspace model , but then it is worse.
Is this a small error or not ?
Till now I don't use the numbers with Curvemeister. But you never know.

Can you explain the how or what ?

Frits

[curvemeister]

Hi Frits,

First, thanks for finding the problem with the shadow/highlight values not updating.  This bug has been around for a long time.  I'll fix this in the next release.

Re the Lab values being different between the two color spaces.  Curvemeister is working internally at 8 bits per pixel.  The Lightness values will therefore vary from 0 to 100.00 in increments of 1/256.  Most Lightness values will therefore have a fractional decimal component.

The a and b channels are 8 bit values mapped from -128 to 127.  Since the increment is 1.0. the a and b values will be even integer values.

If your image is in 16 bit mode, you will see values closer to those of the Lindebloom calculator.

This was a particularly complex part of Curvemeister, and it's good to see you looking at it in so much detail.

Mike

[curvemeister]

BTW - Curvmeister will also copy the colors, as text, of any selected sample points.   

Select the samples you are interested in, or ctrl-A to select them all.  Then right click on one of the sample palettes and click on the "Copy Color" menu item, or use ctrl-alt-C.

Example:

   RGB(126.7,132.7,117.9)
   RGB(0.0,15.2,1.5)

Mike