Major changes in PhotoFlow "stable" branch

Can PhotoFlow open CIELab images?

Using the stable branch (just built) I opened an image in the LAB color space (from Lindbloom’s Profile Evaluation set - Welcome to Bruce Lindbloom's Web Site). The image just looks solid black, was opened “as if” it were sRGB with no profile. Assigning a LAB profile from disk didn’t make any difference. Krita opens the same image as a LAB image and it looks correct, so I’m guessing there is no problem with the image file itself.

This happened to me a while ago but I don’t remember in which context.

The packages from today should completely solve the issue with opening old .pfi files with straight conversions to Lab colorspace… please let me know if this is still not fixed for you, or if you still experience other backward compatibility issues…

Thanks for doing this, I’ll check it out when the OSX version is available.

It seems that the underlying VIPS library has some troubles reading such type of image… I will investigate.

Just checked again and whilst there are appimage and windows versions dated 29th there is no OS X version. The one dated 28th does not include the fix I presume as it does not open the direct to Lab files correctly.

Correct. I have some issues with the automated OSX building, hopefully something that will be sorted out today…

The OSX package is finally available for download! It is tagged with the date of today 30th of November.

1 Like

The ‘direct to Lab’ files open fine with this version. Thanks again.

1 Like

@Carmelo_DrRaw If I embed perspective cor + crop within a Group Layer,

image

the edges of the image get distorted; e.g.,

image

This remains in exported images. Outside of the Group Layer, they work as expected.

Set the group layer blend mode to passthrough to disable any blending, otherwise it tries to combine two images with different sizes.

1 Like

Thanks. Are there other uses for passthrough?

I don’t know if you prefer issues reported via github or here, but… I just tried the latest appimage of PhotoFlow (./PhotoFlow-20171129_0741-git-stable-070c6d3e7be6e654b13c83e980bb5c2213742b7f.glibc2.17-dbg.glibc2.17-x86_64.AppImage) and I get a segfault if I try to set a custom DCP profile in the RAW developer module. The crash happens as soon as I select “DCP from disk”…

Thanks! I will have a look. Generally I prefer issues reported on github 'cause they are easier to follow, but here is fine as well…

Confirmed, I am looking into it…

It was just a missing initialisation of a pointer… the latest AppImage should not crash anymore when switching to DCP profiles.

Thanks for discovering this!

@Carmelo_DrRaw When I export the image to a TIF (Rec.2020) and import it back into PF, why is the input profile embedded (sRGB)? When I import into GIMP, it recognizes the ICC. Also, GIMP asks me to select a page (of 3) in the import dialog.

embedded (sRGB) means “read the embedded profile, and fall back to sRGB is no profile is found”. If the colors of the image imported back into PhF are OK, it means that the Rec2020 profile is loaded and used correctly.

Concerning the 3-pages issue with some images, it looks like there is some EXIF data which is not removed correctly. I have plans to look into this as well…

Organizing lots of layers - makes it easier to figure out what each group of layers is supposed to accomplish, by allowing to group them without having the actual layer group do anything other than “contain” the layers.

@Carmelo_DrRaw - is my explanation correct? I think I asked Ell (one of the GIMP devs) this question about GIMP’s passthrough blend (new in GIMP-2.9), and this was the answer, I think . . .

@afre - try setting up a passthrough group, and if my understanding is correct, moving a layer in or out of the passthrough group shouldn’t make any difference.

I think this is indeed the main use of the passthrough mode. I might even consider setting the blend mode to passthrough by default for group layers. At the moment I cannot think of some other typical usage…

I have to think about it more carefully, but most likely you are right… at least in the case in which the group only contains one layer. Otherwise, moving a layer in or out could actually change the order in which the layers are applied.