My sharpening workflow for base ISO mages with the new features in RT (aka extreme pixel peeping)


(Ingo Weyrich) #1

I implemented some new features in RT which help to get clean, but also sharp output.
Here’s the link to the raw I use in my example (licensed free for everybody)

1.) Open the file in RT
2.) Apply neutral profile
3.) Apply Auto Matched Curve ISO low
4.) add some detail windows
looks so:

As you can see in the detail windows it looks quite soft because no sharpening is applied

5.) enable Sharpening and use RL-deconvolution
looks so:

As you can see in the detail window which shows blue sky, there are some sharpening artifacts in blue sky

6.) Change demosaic method to AMaZE+VNG4
looks so:

Still some artifacts in blue sky,

7.) enable the new sharpening contrast mask display and adjust the new contrast threshold adjuster until the preview window which shows the sky is dark (as in this screenshot)

8.) Disable the sharpening mask display
Now the sky (and other flat regions) are free from sharpening artifacts
Looks so:


9.) Go to Microcontrast tool. There set the new contrast threshold adjuster to the value of the sharpening tool and enable Microcontrast
Looks so:

[PlayRaw] View from Fort Carré in France
(nosle) #2

Is this in dev? Looks interesting!

(Ingo Weyrich) #3

Yes, that’s in dev

(Ingo Weyrich) #4

@nosle it is in dev, but the mask view atm has not its own icon and no shortcut

(Ingo Weyrich) #5


(Sebastien Guyader) #6

@heckflosse I go, in your workflow do you also play with the contrast threshold under AMaZe+VNG4?

(Ingo Weyrich) #7

For my Nikon D700 the default of 20 works good. But for files from other cameras I use different settings.

(Sebastien Guyader) #8

Do you always use the same contrast threshold value for the three of Demosaicing, Sahrpening and micro-contrast?

(Ingo Weyrich) #9

If I use Microcontrast, I use the same value for Sharpening and Microcontrast. The value for Demosaicing is different.

(Heinrich Holtmannspoetter) #10

proposed procedure for tweaking contrast threshold in amaze+vng4 demosaicising:
First set sharpening threshold to a value, where sharpmask shows still marks in flat regions e.g. between 7 and 10. Then tweak contrast threshold of demosaicising to a value short above a point, where sharpmask doesn’t change any more. In the next step the obove mentioned procedure of heckflosse can follow.

Is a procedure like that ok?

(Ingo Weyrich) #11

Yes, that’s a good procedure :+1:

(Ingo Weyrich) #12

@hholt It should be possible to automate the process of finding a good threshold for AMaZE+VNG4. I will try that.

(Heinrich Holtmannspoetter) #13

If it’s possible to automate a good threshold for AMaZE+VGN4, perhaps it is also possible to use that process to automate a threshold for sharpening and microcontrast?

(Ingo Weyrich) #14

That should also be possible. But step by step :wink:

My first tests to find a good threshold for AMaZE+VNG4 are promising. It won’t be completely automated because the user still has to define the area where the threshold is calculated from. This will be implented as a tool like a spot whitebalance picker, means you click on a point in the image (for example blue sky) and the threshold will be calculated from the surrounding 36x36 square.

(Richard Regal) #15

How does one enable the sharpening contrast mask display? I cannot find any button to click on.

(Gabriele ) #16

Can someone explain me the colors in the mask? What they mean ??


(Ingo Weyrich) #17

The button is right of the focus mask button. Currently it has the same icon as the focus mask button. That will be changed soon.

The colours are the colours of the image. They mean nothing. They are just background for the mask.

The brightness counts. Bright parts are taken from AMaZE, dark parts from VNG4

(Richard Regal) #18


(Ingo Weyrich) #19

It turned out that it is possible. In my current tests I get an almost perfect auto-calculated contrast threshold for AMaZE+VNG4. Almost because currently the granularity of the slider is 1 and for that reason the auto-calculated threshold may be off by < 1. Changing the granularity of the slider to e.g. 0.5 would allow to get even closer to a perfect match.

The auto-calculation is done by an iterative process which at current granularity of 1 adds only < 1 ms to the processing time. Changing granularity to 0.5 would add < 2 ms to processing time.

Now I have to make the gui part. Damn, I hate gui coding…

(Stefan Chirila) #20

It’s posts like these that get me excited :slight_smile:
While Lightroom has its hidden-behind-the-hood demosaic process, RawTherapee has more than half a dozen and they are in constant development. I can’t wait to try out the new features. It’s so nice seeing a bunch of good folks working on pushing the limits of what we can get out of our cameras.