Blending modes don't work like they do in Photoshop!

Depends....

...using LAB mode in PS you get something similar.

Rô

Unfortunately, Photoshop blending modes aren't properly documented anywhere and while most of modes are described somewhere on the Net or in the books, Saturation, Color and Luminosity aren't. We did our best to guess the internal formulas of Photoshop for those, but we know that they are not identical, especially for different saturations -- looks like Photoshop "multiplies" saturation when doing HSY blending, but the exact nature of this "multiplication" is not known to us.

If you have some insider info or a proper description of the Photoshop HSY blending algorithms -- please share with us

As a side note, even Gimp implements these modes incorrectly and even "more" incorrectly than us

Well, I'd argue that RGB is very close to how our eyes work (not including the gamma correction and the additional Y "channel" present in our eyes) -- Lab is not closer

The problem of Lab is it's poor visual representation charasteristics -- HLS and to some degree RGB are much better at that (while L channel is pretty straightforward, a and b are not).

Printers are a wholly different story, while you may _assume_ RGB as device-independent color model, you certainly can't do that with CMYK, thanks to all those coating settings, subtle and not-very-subtle ink chromaticity and density differences

And yes, by "correct" I mean "just like Photoshop".

On topic: I can promise that we'll look further into improving blending modes compatibility with Photoshop, but cannot give any certain promises whether it will be fully compatible.

Quote
uberzev wrote: The Color blending mode in Photoshop changes the Hue and Saturation of the underlying layer while leaving the Lightness intact. I don't understand why that's so difficult to implement.

You can alway try to build your own blendmode by using the channel components.

Quote
uberzev wrote: The Color blending mode in Photoshop changes the Hue and Saturation of the underlying layer while leaving the Lightness intact. I don't understand why that's so difficult to implement.

It isn't.


The difficult bit is deciding how to represent the mixture that results.
As per you example, Luminosity is 0% or 100%, Hue is 0% and Saturation is 100%.

What Hue 0%, Saturation 100%, Luminosity 100% represents depends on each colour mode. (see image).

The conclusion.

Those used to working in Photoshop have a choice of modes to work in, sometimes because one is more suited to the job, more usually because of personal preference.
Filter forge does not have this option. It uses the HSY model - which, BTW, does not exist as an option in PS.
This means, as uberzev has demonstrated, that some operations may give "unexpected" results.

I see three options:
a) Learn how to work in HSY;
b) Ask onyXMaster and the guys to give us mode options;
c) As Quasimondo commented - make your own blenders.

My money's on a).

Rô

Yes, we've fixed all blending modes except for those you mentioned -- to fix them we must know the coefficients that the Adobe programmers used in their HSY calculations, and I don't think that reverse-engineering Photoshop is a good idea

Quote
uberzev wrote: I see your point. Still I think it should be changed so that it is at least somewhat similar

This is the closest we could get. For a full match, we would need to reverse-engineer Photoshop, which we obviously won't do.

Its not perfect by any means, but changing the internal math of color & luminosity to use HSL instead of HSY produces results much more similar to Photoshop's...

Sweet. 2.5 years later and you finally nailed it.

A date would be heartily appreciated.

Maybe's give me optimism.

Quote
uberzev wrote: Sweet. 2.5 years later and you finally nailed it. Wink

ROFL