Seite 1 von 2

Neue Testversion 22.90b4

Verfasst: Mi 03 Feb 2021 06:46
von Gerhard Huber
Hallo,

es gibt wieder eine neue Testversion 22.90b4.

Windows:
http://www.pl32.com/beta/pl2290b4.zip

macOS:
http://www.pl32.com/beta/plx2290b4.zip

Neues:
  • Panoramafunktion jetzt fertig für Betatest
  • Focus Stacking jetzt fertig für Betatest
  • Textebenen können wieder auf "dynamisch" zurückgeschaltet werden, in den Werkzeugeinstellungen, letzter Reiter
  • Diverse Fehlerbehebungen und kleinere Verbesserungen

Re: Neue Testversion 22.90b4

Verfasst: Mi 03 Feb 2021 12:28
von gmhofmann
Die Position der Bildübersicht ist jetzt stabil, danke.

Gibt's für die beiden neuen ß-Funktionen die Andeutung einer Anleitung?

Re: Neue Testversion 22.90b4

Verfasst: Mi 03 Feb 2021 12:34
von Gerhard Huber
gmhofmann hat geschrieben: Mi 03 Feb 2021 12:28 Gibt's für die beiden neuen ß-Funktionen die Andeutung einer Anleitung?
Im Moment nicht. Das Focus Stacking hat einige Parameter, die nicht selbst erklärend sind, die werden aber zur Release vermutlich noch verschwinden.

Re: Neue Testversion 22.90b4

Verfasst: Mi 03 Feb 2021 20:54
von maxwell
Panorama funktioniert jetzt bei mir. Probleme an den Übergängen habe ich noch nicht festgestellt.

Re: Neue Testversion 22.90b4

Verfasst: Do 04 Feb 2021 17:37
von shijan
Some nice bugfixes here!
Little icons in the lower right corner on text frame problem. In pixel mode if text size is small and document is zoomed-in, those icons became pixelated and distorted.
Problem fixed in 22.90b4

On macOS "Show Pixel Grid" always enabled by default for new documents and images, even if it was turned off earlier.
Problem fixed in 22.90b4

Stitch - if i use images that different in pixel size, filter report an error message and the only way to back things to normal is completely close PhotoLine and start it again.
Problem fixed in 22.90b4, no more global hang if something goes wrong in Stitch.

Re: Neue Testversion 22.90b4

Verfasst: Do 04 Feb 2021 18:23
von PhilM
Findings on Win 10 / Core i7 / 16Meg RAM

Stitching

Stitching 12 images 16Mpix (architecture) : Create sum layer
Takes approx 2min. But memory usage tops at 99% and data disk activity at 100%
The resulting image looks OK. But the horizontal features of the cathedral are bent/not rectilinear.

With the same images, Create layers > 12 aligned layers
Similar load
The resulting images are very badly aligned / unusable

An attempt to stitch the 12 aligned layers together chokes the PC forever. Mouse is frozen.
The first try, I had to shut down the PC.
The second try, I had to kill PL process after 10min.

Focus stacking

Attempt to stack 7 images 16Mpix (flower) failed after a few seconds : "Failed to match between image 0 and 3".


General comments :

1. I would like to be able to use the align algorithm to align a layers stack as a stand alone solution for further manual processing.

2. I would like to see an additional solution to reduce noise by stacking multiple shots of the same scene and apply "Median" on each pixel.
Median tends to reject stray pixels.

3. Would be nice to have an "Advanced" button to access the parameters of the HDR/Stitching/Stacking algorithm.

Best regards

Philippe

Re: Neue Testversion 22.90b4

Verfasst: Do 04 Feb 2021 18:58
von der_fotograf
But the horizontal features of the cathedral are bent/not rectilinear.
If the field of view (FoV) is larger than ~ 100° this is normal. You should install http://hugin.sourceforge.net/ (OpenSource) and create a panorama — the result would be identical. Hugin is the most precise panorama tool which gives you the most advanced features i.e. including manually setting control points. In addition, it offers a large variety of projection models.

Screen Shot 2021-02-04 at 17.56.50.jpg
Failed to match between image 0 and 3
I could imagine that the deviation between image 0 and 3 have been too large.

Re: Neue Testversion 22.90b4

Verfasst: Do 04 Feb 2021 20:08
von maxwell
Wird aus Bildern ein Panoramabild generiert, ist dies bei allen Versuchen ok. Werden die Ebenen ausgegeben, so sind diese nicht optimal ausgerichtet. Wird aus den "ausgerichteten" Ebenen wieder ein Panorama erzeugt, so stimmt dieses nicht mit dem direkt erzeugten Panorama überein.

Re: Neue Testversion 22.90b4

Verfasst: Do 04 Feb 2021 21:05
von PhilM
der_fotograf hat geschrieben: Do 04 Feb 2021 18:58 Hugin is the most precise panorama tool which gives you the most advanced features
That is why I usually stitch using Hugin.
I could imagine that the deviation between image 0 and 3 have been too large.
The test images were assembled without problem onboard my Olympus camera.
They also get assembled using enfuse.

Re: Neue Testversion 22.90b4

Verfasst: So 07 Feb 2021 02:37
von shijan
Stitch still looks very strange sometimes. It always attempt to fix lens curvature even if there is no lens curvature.
Seems it needs some "Flat/Planar" mode and and "Lens Correction" mode. Seems similar separation used in many other apps.
Bild
Bild
Bild

Here is example when attempt to stitch of two simple screenshots:

Result in PhotoLine:
Bild

Result in other apps:
Bild

Re: Neue Testversion 22.90b4

Verfasst: Mo 08 Feb 2021 04:33
von shijan
I guess there is a mistake in color management logic between Adjustment Layers and Color Picker.

Here is how to reproduce:
- Switch image to Document mode.
- Set Document ICC profile to something like sRGB.
- Set Image ICC profile to sRGB with Linear Gamma.
- Add White Point Adjustment Layer (Setup Grey Point mode)

If White Point Adjustment Layer placed on top of image, it works in Document ICC profile as expected and pick colors based on Document ICC profile as expected.

:arrow: PROBLEM: If White Point Adjustment Layer placed inside image, it works in Image ICC profile as expected, but incorrectly pick color values based on Document ICC profile.
I see the same problem in Channel Mixer tool, so i guess same problem may exist with all other tools that use Color Picker.

P.S. The good news is that there is no problem if i create additional color managed Isolated Group with Linear ICC profile and put White Point Adjustment Layer inside that group. In this scenario it works in Group ICC profile as expected and pick colors based on Group ICC profile as expected.
Bild

Re: Neue Testversion 22.90b4

Verfasst: Do 11 Feb 2021 15:12
von Paul
Hi,
could it be possible to add presets to the Context Blur?
- Paul

Re: Neue Testversion 22.90b4

Verfasst: Do 11 Feb 2021 16:07
von shijan
Paul hat geschrieben: Do 11 Feb 2021 15:12 Hi,
could it be possible to add presets to the Context Blur?
- Paul
Nice notice! But it is not only about Context Blur. A lot of other filters also don't have presets. I wonder why?

Re: Neue Testversion 22.90b4

Verfasst: Do 11 Feb 2021 17:27
von Gerhard Huber
"Standard Filters" don't have presets because in 99.9999% of the usage you have to change one value only.
So you won't need presets.

Re: Neue Testversion 22.90b4

Verfasst: Do 11 Feb 2021 17:52
von shijan
I agree that presets for many filters are not too important, but soon or late sometimes presets may help... Anyway, probably it is not a huge problem because it is possible to save global presets for any Adjustment Layer(s) as global preset in Adjustments mini window. But it is not possible for legacy non adjustment filters.