Topic: CurveMeister, mask, transparency

[derekfountain]

I've been having trouble with CM when I present it with images which have masks and layers containing transparency. It's proving hard to reproduce, but I've seen it a few times with different images.

I've put a sample image which shows the problem here (47MB file):

http://www.derekfountain.org/misc/img_1396.zip

Leaving it with just the top layer visible and selected and opening CM, I get a result as shown in the screenshot below: the whole image is visible in CM (not just the visible layer) and the mask seems to have been converted into a couple of while blobs. Clearly not right.

When I've tried to reproduce this with a simple image created from scratch, CM always seems to do the right thing. Hence pointing people at the rather enormous image that reliably demonstrates the issue.

[curvemeister]

Well, the darndest thing happened.  I loaded your image, Derek, and as is often the case, the image came up perfectly.  No white areas.  Then when I clicked OK, another bug I've been chasing occurred: the Photoshop window refused to come to the front.

So - this is an important clue to a new bug.  I will take a crack at improving the handling of layers that overlap the curvemeister window.  There are no guarantees that I will be able to fix this problem before it is time for me to move on to Curvemeister 3.

OTOH several other bugs that you found are addressed in the 2.3 release.

[derekfountain]

Well, the darndest thing happened.  I loaded your image, Derek, and as is often the case, the image came up perfectly.  No white areas.  Then when I clicked OK, another bug I've been chasing occurred: the Photoshop window refused to come to the front.

So - this is an important clue to a new bug.  I will take a crack at improving the handling of layers that overlap the curvemeister window.  There are no guarantees that I will be able to fix this problem before it is time for me to move on to Curvemeister 3.

OTOH several other bugs that you found are addressed in the 2.3 release.

That's annoying. I've been seeing this bug all the time - it was happening constantly yesterday while I was working on my 'lukas' CM challenge image and I've been trying to find ways to work around it. For me, it's 100% reproducible with the image I left for you to look at.

When you say "the image came up perfectly" do you mean just the (faint) visible layer came up in CM, or the whole image? I was expecting just the visible layer - maybe that's the wrong expectation? Certainly the white areas aren't correct!

It's not really that important and I wouldn't want you to prioritise this in favour of CM-3.0. I'll get a copy of 3.0 when it comes out, and if the bug persists maybe we can have another look then.

[curvemeister]

Hi Derek,

I tried using Curvemeister on several layers, and got the expected behavior.  There is a limitation in Curvemeister right now in that it will honor only the transparency of the layer you are working on.  This means that layers under that layer will look OK, but layers above the currently active layer will look strange.

Thanks for your patience.  It's not unusual that a bug maifests itself differently, or does not show up at all, on my system.  I'll spend some more time tomorrow trying to reproduce the error using a different system.  I tested on XP, using Photoshop CS2 - perhaps you are using a different combination of OS and Photoshop version?

[Frits]

Derekfountain,

I have used your  picture in my Photoshop elements 4.0.
Trying loading the picture it said he is converting the colormode.
PSE4 can only use RGB, indexed colors, bitmap and gray-values.

So, converted, I get the same results you have.
P.S.E 4.0 is derived from P.S so I can expected such same result.

Frits

[derekfountain]

I tested on XP, using Photoshop CS2 - perhaps you are using a different combination of OS and Photoshop version?

Nope, that's what I've got. I suppose in a program as complex as Photoshop there's any number of permutations of settings that might cause things to behave differently. Frits has got a result that's weirder than mine. )

[Frits]

Mike,

I have done it again, loading the picture and starting regmon.exe and then Curvemeister.
It does a lot reading in the registry and settings.
So when you like it , I can make a log file of it and send you, for when you making Curvemeister update.

frits

[curvemeister]

Thanks Frits,  I don't think this is a registry related issue, but your post makes me think that going back to older versions of Photoshop and/or Elements might "smoke out" this bug.

[curvemeister]

Aha!  I'm reproducing the problem perfectly - all I had to do was keep the layer visibility settings as Derek had them in the psd file.

Now - all I have to do is fix the problem!

[Frits]

Now - all I have to do is fix the problem!

Yeah....lots of small problems. You have much to do. So we can beta-testing the solutions.
In witch language is CM written ?
Assembler ?

Frits

[curvemeister]

Hey, I love this stuff.  Writing code - even chasing bugs, is what I do best.  Curvemeister is written in C++.

[Frits]

It is for me years ago, say 1979 1980-81 that I have done something in a prgr-language.
never C++. I did somewhat in BASIC assembler and forth. ( not forthran).
I have tried a little in Pascal.
So, nowadays I think I could n't.
These days assembler was relativ easy. I programmed the 6502 with 50 opcodes.

frits

[Greg Groess]

Mike, you need a hobby...maybe... oh, I don't know Photography or something...it takes the mind off PC's and software....

Greg

[Frits]

"However, Billions of electrons were terribly inconvenienced. "
 
No they are not, they like the trip.
But they return always.

Frits

[curvemeister]

It is for me years ago, say 1979 1980-81 that I have done something in a prgr-language.
never C++. I did somewhat in BASIC assembler and forth. ( not forthran).
I have tried a little in Pascal.
So, nowadays I think I could n't.
These days assembler was relativ easy. I programmed the 6502 with 50 opcodes.

frits

I started off in Basic too, way back when.  The 6502 was used in the early Atari video game, and I did quite a lot of programming on that processor.  It had a fun instruction set because there were so many different combinations of address modes and operands that there was always a way to squeeze out another instruction or two.