Elle Stone's tutorial, Autumn colors

gimp

(Bill Martz) #1

I am new to editing photos and am wondering if anyone has followed the steps in Elle Stone’s tutorial, Autumn colors: An Introduction to High Bit Depth GIMP’s New Editing Capabilities, https://ninedegreesbelow.com/photography/high-bit-depth-gimp-tutorial-edit-tonality-color-separately.html#edit-for-Chroma, and reproduced her results? I have gone through the steps numerous times, but do not end up with anything near what she ends up with.


#2

I have followed many of her tutorials. The thing is that they are quite involved, so they might take some time to understand even though they are fairly well-written. Another thing is that she was either using her own custom CCE GIMP or the dev version, neither which are supported presently.

I don’t think she frequents discuss anymore. I suggest that you contact her via the email address found on her website.


(Bill Martz) #3

Thanks, Afre.

Yes, the tutorials are quite involved. I could try emailing her with my questions. The worst that could happen is that she says she is too busy to respond, which I would understand.

I am using GIMP 2.10.2, which is not the same as her modified version, but with 32 bit linear light I hope it can handle the steps of the tutorial.


#4

Keep at it and say hi for me. I am learning all the time. :slight_smile:


(Bill Martz) #5

Will do!


(Bill Martz) #6

afre,

I couldn’t find Elle’s email address. Is it still available?


(Morgan Hardwood) #7

https://ninedegreesbelow.com/about/copyright.html

Keep in mind that Elle took a hiatus precisely because she needed the break from never-ending discussions for various reasons, so maybe emailing her about this relatively unimportant stuff is in your best interest but not in hers.


(Bill Martz) #8

Understood. Thanks.


(Mica) #9

Your choice to email her, she still posts to the GIMP dev stuff, so she’s not completely offline.

Her email is ellestone (at) ninedegreesbelow (dot) com, though the footer of her websites suggests coming here :smile:

Here color tutorials are thick and if you’re completely new to color science and all that jazz, I’d suggest you work though them several times to try and understand all the conceptual information. If you cannot elaborate to her in pretty good detail what is not working, you won’t get anywhere!


(Bill Martz) #10

Some of the problem may be the difference in the image she set up for download and the one she worked on; it’s lower quality. Other questions I have involve why when I do what the tutorial says, the image in GIMP doesn’t seem to change. I may not be using GIMP correctly, or a step was left out on the instructions. That’s why I was looking for someone who had worked through the tutorial and had gotten the correct result.

The first issue doesn’t show up until quite a number of steps into the tutorial, so posting an explanation here would be rather involved versus saying go to step C.C1.3.6 and compare her result with this picture of mine.

To produce results that came close to hers in later steps, I had to duplicate layers entirely differently than her instructions. I worked through the instructions many times, saving files at intermediate stages as I went. Yet, I am still very puzzled. Has anyone actually worked through the tutorial?


#11

Like I said, I have. It requires a lot of patience, questions and figuring out the compatibility. Consider sharing your issues here. We the forum are here to help. :slight_smile:


(Gustavo Adolfo) #12

Yes, I did, a couple of months ago.

It resulted in this:

I recall having some questions and I’m not sure if I completed the tutorial :stuck_out_tongue:, but here it goes the xcf file: scene-referred-plus-masks-16f.xcf (40.7 MB)
Finally, I’m newbie.


(Bill Martz) #13

gadolf,

To begin, compare your image against this screenshot of Elle’s final image. Elle's%20finished%20Autumn%20image%20

There’s quite a difference. However, up to finishing step C1.3.6, we get the same results.

Although I see that your LAB B channel is not assigned a V4 RGB working space profile from Elle’s profile pack that has the “LAB L” companding curve. These profiles all have file names ending in “-labl.icc”, I can’t see that it makes a difference.

I was uncertain here: C1.3.4 “Go ahead and delete the originally extracted layer.” I took that to mean delete “scene-referred for LCH Chroma group” layer.

I was also uncertain here: C1.3.5. Make a layer mask using “Grayscale copy of layer” and load this mask as a selection. For “load this mask as a selection,” I right-clicked on the mask and selected “Mask to Selection.”

Again at C1.3.6. “Add the selection as a layer mask to the original “Visible + Channel mixer to add Chroma” layer, and then delete the modified extracted layer.” I right-clicked on the layer and selected “Add Layer Mask” and in the “Add Mask to Layer” dialog box selected “Selection.” Then I deleted the LAB B channel layer.

My result: scene-referred-plus-masks-16f B C1.3.6 after deleting the modified extracted layer .xcf (17.5 MB)

You can compare my result to yours at this point by clicking on the eye icon for the Lightness group in your image to remove its visibility. As far as I can see, they are the same.

However, this is what Elle shows as the result at this point:Elle's%20Blended%20result%20after%20adding%20a%20suitable%20mask%20

You can get your image to be close to that by clicking the eye icon to remove visibility for the original, i.e., scene referred, layer (the very bottom layer).

You get that result because you did not delete the “scene-referred for LCH Chroma group” layer in the LCH Chroma blend group, which I thought, according to the instructions, we were to remove. Remove visibility for that layer, and you get an image with lots of transparency.

Click the eye icon again for the bottom layer (scene referred) to get back to where I am with my image. We both have pixels of bright oranges, yellows, and greens that are not in Elle’s blended result.

That’s as far as I am going to go with this particular post. I will cover more later.


(Bill Martz) #14

Here’s the blended result as an xcf:
Elle’s Blended result after adding a suitable mask .xcf (233.2 KB)

And the final result:
Elle’s finished Autumn image .xcf (468.6 KB)


(Gustavo Adolfo) #15

Hi @Underexposed

Please, give me some time to review this. It’s been some time since I did it, so I have to refresh my mind in the spare time.


(Bill Martz) #16

Take your time. Going back and getting up to speed is a slow process. I tried to make what I said as clear as possible, but if anything is confusing, let me know, and I will help you understand it.


#17

I haven’t touched this tutorial since it was first written, so I am a little rusty. And it has been updated since, along with GIMP, etc. … :man_shrugging: Here are my thoughts:

When I open the XCF, I immediately encounter an issue. It is one that I have noticed all along but haven’t asked about yet.

image

Let me tell you what it is in case you can’t tell by the image. To GIMP, this image has a perceptual gamma. This becomes apparent when GIMP confronts you directly with

image

But I don’t want to convert! My workaround would be to create an empty image with the same profile but set to linear light and 32f, and then open as layers the XCF. Is there a better way? (@paperdigits @Morgan_Hardwood please follow up my post wherever I write “my workaround”.)

Let’s get to where @Underexposed is stuck. Take a look at @Elle’s screenshot.

See how there is only one masked layer, besides the untouched scene-referred one? Well, that is where all the steps go toward in C1.

Step 3 is where it begins to be tricky. To my understanding, you can only apply profiles to the whole layer stack. When I extracted LAB B, I could not find a way to apply LABL to that one layer. My workaround would be to do ImageDuplicate to create a copy of the whole stack and then select and copy the layer’s image back to the original stack. Don’t forget to convert from LABL back to g10 before you do the copying. (Just remembered that I didn’t do that.)

What you delete is the extracted LAB B layer. You don’t need that anymore; you have the LABL LAB B now.

I am uncertain about step 5 but I think we are supposed to select the LABL LAB B layer and make a grey scale mask on it using its grey scale so that we are able to make selection from mask. Then you use this selection to make a mask for “Visible + Channel mixer to add Chroma”.

Now you delete the LABL LAB B layer because you don’t need that either. Now the result should be similar. It won’t be the exact same for the reasons that @Elle has stated in her article and website, and because our setup might be different in obvious or subtle ways.

Mind blown. :exploding_head: I feel my body contorting just by writing this post. I hope I didn’t mess up the explanation because of that.


(Gustavo Adolfo) #18

Why?


#19

This is because the profile is already linear gamma.


(Bill Martz) #20

I followed the steps as indicated.

Instruction B4.:

  1. Open the image and promote the precision to 32-bit floating point.
    If you convert to integer precision, many of the editing operations used in this tutorial will clip intermediate editing results. This “integer clipping” will completely defeat one of the goals of the tutorial, which is showing you how to work with out of gamut channel values.

  2. If you are using default GIMP 2.9, convert the image to GIMP’s built-in sRGB color space. If you are using CCE, make sure the image is in the linear gamma RGB working color space of your choice.
    The downloaded XCF file is in the linear gamma sRGB color space.

There will be very slight differences in the Chroma of the final image when processed in different linear gamma RGB working spaces because Channel Mixer is a chromaticity-dependent editing operation. Because of how Channel Mixer is used in this tutorial, the differences in the final image won’t be noticeable.