Issues Masking Layers in LIE

Tamuna

I've run into a confusing issue here. I'm adding a new color layer in LIE and attaching an image mask to that layer so that only the unmasked shape is visible - eye makeup options for Genesis 9. But when I render in Iray, I can clearly see that the whole head material has changed a little bit. Mind you, the mask does work mostly, I can see the makeup properly laid on eyelids. I've checked the mask itself, to make sure the masked areas are fully black and they are.  Strangely enough, the same doesn't happen with my LIE lipstick and eyeliner options, even though I'm making them exactly the same way.

Any ideas?

FenixPhoenix

It's hard to say with hhst that information. Can you show a screenshot of the LIE screen? Maybe you can also check to see if the mask has any blendmodes applied to it too.

Tamuna

Here, I've tried to capture everything related. The LIE is used for Base color and Translucency color. Like I said, in some cases it looks fine, but in others it does this weird color change thing. You can probably notice in the viewport render the head and neck look too pink. To clarify even more: I'm making a set of makeup options. So this eyeshadow you see in the first screencap has several color options. The way I went about creating them was creating one shadow, say green, saving that one. Then just loading it and changing the color of the color layer in Base Color LIE and Translucency Color LIE. There's no reason why green and blue should be fine and red and pink be rendering wrong, if it's a mistake on my end. I think there's something I'm not taking into account? a bug, or some strange quirk of Iray's related to specific colors? I don't know.

FenixPhoenix

OK, everything looks okay in that one. Can you save the problematic LIE and the LIE that's working and share the duf file here? I don't need the textures, I just want to take a look at the code to see if I can spot what's going on. It could be that the gamma is different in one of them.

Tamuna

Sharing the LIE file. Saved it right after a problematic render. I've removed the mask textures and saved it without compression. Hopefully it shows something.

Tamuna

Sharing the special temp folder, actually.

FenixPhoenix

Tamuna said:

Sharing the special temp folder, actually.

Okay, so these two duf files have no mask, meaning that the color you are applying as LIE will be spread across the entire surface. I'm not sure why only one would be more glaringly obvious than the other, though both would create a seam. It was my understanding that you were using a mask to control where the color was applied, correct? Is the mask used on both examples the same one?

Tamuna

FenixPhoenix said:

Tamuna said:

Sharing the special temp folder, actually.

Okay, so these two duf files have no mask, meaning that the color you are applying as LIE will be spread across the entire surface. I'm not sure why only one would be more glaringly obvious than the other, though both would create a seam. It was my understanding that you were using a mask to control where the color was applied, correct? Is the mask used on both examples the same one?

Yes, exactly the same. I only changed the colors. Another thing I've noticed is, no matter what color these problematic versions are, they all affect the rest of the skin the same way - give it that exact pink tint you see in the first screenshot - the green layer, the blue layer, makes no difference. The white areas of the mask are the correct color, but the rest of the skin gets pinkish. Was there anything amiss in the code?

Tamuna

Would it be better if I shared the versions with the mask?

FenixPhoenix

Yes, please share the versions with the masks to see if there is something amiss in those. In the ones you shared, the code is exactly the same, so both should behave the same. In the case or the mask, is the mask used the same one?

Tamuna

I've included the actual file hierarchy now, so you can add the main folder as a temporary library and look in the Genesis 9 characters > TMDesign directory.

I've been experimenting some more and I'm more confused than ever. See, there were 7 problematic LIE eye makeup files and I've managed to fix 3 of them - by loading nonproblematic versions, changing the colors and resaving them. Which is how I created the problematic ones in the first place, but for some reason they got saved better today, I guess. But it only worked on 3 of the 7. The other 4 were resaved in the same problematic manner, even though the process was exactly the same on my side. 

And here's another thing. I've resaved a NONproblematic one without changing anything. Just loaded it on a clean slate and resaved it. And it became problematic too.  I'm at a complete loss here. Do you think it's a bug? It has to be a bug, right?

In the folder: in shades directory Green is the formerly nonproblematic one. Just try them all out on some G9 and see for yourself.

FenixPhoenix

I will not be at my computer for most of today. But I will take a look tonight or tomorrow to see if I can spot what may be causing the issue.

crosswind

As far as I tested with the presets ( the green one, which is wrong... ) you attached, the culprit should be the baked LIE file names in temporary folder... After applying LIE presets you provided, the files names of the baked LIE maps are too long as well as including many "code like characters"... (in screen shot 1). I renamed them to shorter file names and assigned to Base and Translucency Color slots, the pink color issue is gone.

I tested the correct LIE Presets of Blue color, the baked LIE file names are much shorter and also looked correct... so there was no pink issue with it.

Besides, I manually set LIE on my side according to your LIE settings, there was no issue either. So I think... it should be a bug related to the long file names of baked LIE(s) maps.... especially with those strange characters in the file names...

But I don't know how they came on your side... Is there any special characters or something in the file path of the texture maps ... ? What is your DS version btw ?

(Edited by mod to remove attachments)

Tamuna

There are no special characters other than _

I don't think the name lengths are the issue either. Like I said, the problematic and nonproblematic LIE mats have the exact same mask file, and the names of .duf files themselves coudn't be the culprit, otherwise it makes no sense that blue is problematic but magenta is not.

The problem always occurs (if at all) directly after I save a LIE file. It's never visible before saving and never delays after saving. There's a bug there somewhere I'm sure.

crosswind

Tamuna said:

There are no special characters other than _

I don't think the name lengths are the issue either. Like I said, the problematic and nonproblematic LIE mats have the exact same mask file, and the names of .duf files themselves coudn't be the culprit, otherwise it makes no sense that blue is problematic but magenta is not.

The problem always occurs (if at all) directly after I save a LIE file. It's never visible before saving and never delays after saving. There's a bug there somewhere I'm sure.

No, the culprit is neither the names of *.duf files nor the names of texture maps but the names of baked LIE images, in both DS temp folder as well as within LIE Preset DUF files... You may clearly see it in my 1st screenshot above.

Well, you can check those image names yourself... in "TMD Shaely EyeshadowHL - Green.duf", by opening it with Notepad++, the "coded" image names are just there for the slot of Translucency Color: (in below screenshot 1). The long file names are also shown in the 3rd and 4th screen shots in your 2nd post above.

If you modifying "TMD Shaely EyeshadowHL - Green.duf" with Notepad++, just simply replace the string: (in line: 70, 71 and 160):

    Shaely_Body_TranslucencyColor.1001 3_6b5a272a-f968-56c0-affe-f5000177ab2a 10_6b5a272a-f968-56c0-affe-f5000177ab2a 2_6b5a272a-f968-56c0-affe-f5000177ab2a 2_6b5a272a-f968-56c0-affe-f5000177ab2a 2

with:

    LIE for Translucencly

Then save duf file, and apply the preset file to the figure, there'll be no pink issue. (screenshot 3)

(Edited by mod to remove attachments)

 

 

crosswind

Then I may speculate the reason why you got such an issue, probably ~

- if you configure those LIE for the first time with LIE Editor, then save an LIE Preset. There should be no such a pink issue after applying the save LIE Preset... but you'll find the image name getting longer in LIE Editor ~
- then if you save LIE Preset again, you'll find the image name getting much longer in LIE Preset DUF file than the firstly saved LIE DUF file ~
- then if you save LIE Preset for more times, the image name in DUF file will be getting longer and longer... till it bring you the pink issue ~

AFAIT, most likely if the length of the baked LIE image name is longer than a certain value, the issue will come... You may prove these yourself then ~ and it's certainly a bug, IMHO ~

PS: You may always change the Image Name in LIE, place a proper short name there before saving LIE Preset, then you'll never get similar issue again...before this bug is fixed.

(Edited by mod to remove attachment)

FenixPhoenix

I have not been able to get to my computer still. Fortunately, it sounds like crosswind solved the issue. Indeed, every time you apply a lie to a map, the name of it becomes long and longer. Saving a lie after applying it multiple times may result in those long names. If you are not comfortable solving via notepad++ (code), then you can solve it by applying the Lie, then opening the Layered Image Editior, and shortening the name there (not the layers, but the name at the very top). After that you can save the LIE and it will keep the short name.

Richard Haseltine

Filename length itself is limited, so even if the total path length is within bounds the OS may fail to access the file when asked.

Tamuna

I completely forgot to bring my ass in here and thank you guys. Recreating the problematic ones from the scratch helped. But boy, what an oversight that is. I hope future versions of Daz Studio fix it.