Do you know if darktable has a lightness control tool?


(Mica) #62

@R_O – you’ve given us the sales pitch, you’ve tested us, you’ve given us assignments about shadow manipulation. You’ve talked about how great this feature would be, a lot. You’ve shared videos.

You’ve been asked by a developer for sample images and you haven’t provided them. After 57 posts in this thread, now would be a good time to provide those images.

How about a mock up of what the tools panel would look like in darktable? How about some of the math or technical techniques necessary to make such a tool?


(Roger) #63

This explanation goes for gadolf too:
Imagine that you have to go from point (A) to point (B), a distance of 100mts.
You can walk or you can use a bicycle. Indeed, each one is a different paradigm (with their own learning curve), but, in a simplistic approach, both do the same because they go through the field, you use your legs, etc.
What paradigm is faster and effortless?
Do you think that the best trained walker can beat a poor trained cyclist? Is it possible to make new things with the bicycle than are almost impossible by walking?
Having this in mind, Darktable is walking and ACDSee Light EQ is cycling, BUT NOT BECAUSE IT ADJUST THE LIGHTS WITH A CLICK, because it manages the exposure the same way than your camera. This is why you take 40 seconds instead of 40 minutes making the same, as you use seconds instead of minutes covering the 100mts in your bike.
However, Light EQ is not an exposure tool, but tone mapping, it is juts a proof of concept that shows how it is possible to adjust the lightness emulating shadows instead of by a gamma correction as all the RAW editors do (DT and ACDSee by itself).
Now, and thank to this post, my Exposition Zones proposal is to create a new software that uses the exposure management (tone curve) based in shadows as its core, taking advance of the possible benefits that this approach will offer. I want to make an electric bike.

Of course, you may use DT or the SW that you prefer, as you can walk or use the bicycle. Your choose depends of your purpose, but neither they are the same or the result is based on your skills.


(Roger) #64

I guess you are right, but this post is no longer suitable, as you say, it is now large and I may add confuse, the worst is it is getting bigger with topics that already were explained from people that only read an small part, plus the garbage made by those who posts things out of context. This trend will grow as more people is attracted by the number of comments.
I guess the best is freezing the post by avoid answering new comments, this way we may build a clean timeline with the newest comments, but I would prefer to use a managed forum to have more control.

In the meanwhile I am going after someone in the sw development team of Fuji to know their point of view. If the concept is valid, it would be great getting them involved.


(Mica) #65

To be quite honest, using Harry Durgin’s video to prove that editing in darktable is slow is very disrespectful and disingenuous. He’s making a video to teach people how to edit, he talks through his decision making, often pauses to explain things, and shows the viewer the effects of a filter he is working with. Because he is making an educational video, it is at a slower pace.


(Roger) #66

No, I am making an factual comparison, because making the same with the Ligh EQ without explanation takes 10 seconds instead of 40 (WITHOUT USING THE AUTO BUTTON). I tested it by myself in case a comment liker yours would come.
I also verify if the video uses a common method and at least in YouTube, most people do the thing the same way, no matter what tool they use (Even ACDSee suggests such method to promote its Editor).
The method is time consuming by itself, no matter what SW you use, specially because you get new results with every adjustment instead of watching them all together.
How long may you take? Give a chance to Light EQ in mode STANDARD (including learning curve) to note the difference by yourself.


(Mica) #67

@R_O I think you missed the most important part of my last post:

I’d say two of those three things would be bare minimum to getting this discussion back on track, otherwise I think you’ve lost everyone’s attention somewhere in the paradigm void.


(Gustavo Adolfo) #68

@paperdigits correct me if I’m wrong, but what I see in this lighteq tutorial are the good old luminosity masks in action.


#69

There have been a few points I’d like to answer to, but since they are scattered all over the thread I’ll just have a single post.

  1. darktable, RT and all the other tools don’t use a gamma curve for exposure, they use the correct and trivially simple approach of multiplication. Of course that brightens the shadows, too, but that’s the very definition of exposure. So complaining that changing the exposure of an image doesn’t increase the contrast in the image in just silly.
  2. The proposed feature is supposed to be able to add new shadows on objects. For that to work a simple approach that takes the existing brightness in the image isn’t enough, thus it’s not a tone mapping operator. Instead you need a depth map of the scene, which regular cameras don’t give you. An exception being systems with more than one camera (stereo cameras or more) and lightfield cameras, and even for those the results are not working great in many cases.

(Mica) #70

More or less, yeah. Divide the image into N luminosity sections. I guess the interesting part is that it gives you both highlights and shadows, so you can increase contrast in a particular zone.


(Gustavo Adolfo) #71

@paperdigits So the point is how they define those zones. In conventional luminosity masks, a shadows slider applied to a pure highlights band would have no effect at all, correct?


(Mica) #72

I’ve never used ACDC, so I don’t exactly know, but I’d guess the shadows are relative to dynamic range of the luminosity mask, so the tool on some clouds would darken the darkest part of the luminosity band.


(Christian Kanzian) #73

The Look of this example reminds me on the Old Oak - A Tutorial

You can so the same in dt with two instances of the tone curve with different modes.
Here is an Image where tried to copy that Look


(Christian Kanzian) #74

I watched this video

But what is the difference between a tone curve in dt divided in 10 to 20 points compared to this light eq?


(Gustavo Adolfo) #75

@pk5dark Do you mean scale chroma automatic in XYZ for the first curve and automatic in RGB for the second?

I did this a saw some results:
Before:

After:

EDIT: But… what does it mean doing that?


(Christian Kanzian) #76

@gadolf

IRC one curve with L only in manuel mode and the second with automatic mode.

Tone curves looks like that for the example posted above.

tone_curve


#77

If I understand correctly, these two curves give you ability to control the amount of desaturation in shadows (automatic scale chroma decreases saturation in the shadows). If so, you can use just a single instance of tone curve with its scale chroma set to “automatic” and blending mode set to “color adjustment”. Then you can control the amount of the desaturation with the blending slider.


(Gustavo Adolfo) #78

I couldn’t get good results with that, not sure if I’m missing something. If you could provide the raw and sidecar file it would be great.
As for my results, look for Tone Curve and Tone Curve1. Turn them off, then on. It really makes the image pop up, similar to LightEq results. But, like I said before, I’m doing this blindly, I don’t know what I’m doing with that.

CRW_2887.DNG (19.4 MB)
CRW_2887.DNG.xmp (18.5 KB)
(licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License 2)


(Christian Kanzian) #79

Your are boosting the contrast by this S-shaped curve.
The midtones are stretched over a wider range in the histogram. By pulling down the curve on the left side dark tones getting darker (left hill in the histogram after). On the right side light tones are pushed a little bit to get lighter (right hill in the histogram).

histogramm_before_after

There is a nice thread here about histograms and tone curves were you can find much better explanations what is going on:

You can set the blend mode to difference to see what parts of the image are effected by the module and the changes.

Watching the last video I noticed the halos this tool could create. It looks like what shadows and highlights module in dt produces if pushed too far with the Gauß filter. So I guess there is some bluring filter around the tonal range controlled by the sliders to avoid harsh tonal breaks. Sorry, for this silly describtion, but I have no idea about the correct terms.


(Gustavo Adolfo) #80

I’m ok with S curves resulting in higher contrast. What I meant was that I don’t understand the meaning of tweaking the scale chroma for the L dimension on RGB and XYZ like I blindly did. But that it gave interesting results, similar to LEq, that is for sure (at least for this photo)

good tip, thanks.

I’m more illiterate in this terminology than you, that’s for sure, but your description didn’t sound different from what I’ve seen so far.


(Eduardo Battaglia) #81

Maybe a stack of “fill light” modules could work like op wants… although there are more controlled ways to work on an image like everyone here just explained