New tool "Capture Sharpening"

The new tool “Capture Sharpening” in the RAW tab (available in the development build) works great and the “Corner radius boost” does a really good job. For me, the standard sharpening is obsolete. :grin:

grafik

Thanks @heckflosse

6 Likes

May you be so gentle to indicate more precisly where i can download this tool?

THX Giuliano

1 Like

@Julianus it’s in the latest development builds, dev branch.
http://rawpedia.rawtherapee.com/Getting_Started#Get_RawTherapee

1 Like

Thanks!!

We still need the standard sharpening for non raw files.

I know. :wink:

And for the case you want to use the “blur” function. I find that Capture sharpening works best when you have little noise in your photo. I shoot MFT and even at base ISO I often see noise. Also the Auto Threshold sometimes (mostly in photos with large OOF areas I think) is too low so that noise gets sharpened.

I think it’s an awesome tool, especially for Landscape! Thanks a lot, Ingo!

2 Likes

@heckflosse
Just a but of feedback :+1:
I was quite happy with my X-Trans 2 raw processing using Demosaicing 3-pass + fast, RL deconvolution and auto matched tone curve.
But now with same demosaicing + Capture Sharpening + Haze Removal, it’s taken them to a new level.
My Capture Sharpening on 5.7-59-g1e27156 hasn’t the Corner radius boost, so still more to look forward too.

Thanks :+1::+1::+1::+1::+1::+1:

1 Like

Thanks for the feedback :slight_smile:

I guess it will be in next nightly build.

Some details to the Corner radius boost:

Assume you set the radius to 0.6 and the boost to 0.15. Also assume the image is in landscape format and consists of 7x5 tiles. Then the following radius values would be used for capture sharpening:

0,75	0,72	0,69	0,68	0,69	0,72	0,75
0,73	0,69	0,66	0,64	0,66	0,69	0,73
0,72	0,68	0,64	0,60	0,64	0,68	0,72
0,73	0,69	0,66	0,64	0,66	0,69	0,73
0,75	0,72	0,69	0,68	0,69	0,72	0,75

This is a simplified example. In fact, each tile has a size of 194x194 pixels, giving for example roughly 30x20 tiles for a 24 MPixel raw.

3 Likes

Are negative values possible with corner radius boost? In other words, can I get less sharpening in the corners such as for a portrait where I want the background to be out of focus anyway?

Not at the moment, but that’s easy to implement.

Thanks. If others concur, I would suggest it it be added.

@chaimav

It’s in dev now :slight_smile:

3 Likes

@heckflosse Was always wondering how apps could have a radius of less than 1. In your example radius seems to be the center value of the kernel and boost + radius the corner…

@heckflosse forgive me for not checking, but can the gamma value be 1,00 now, as it theoretically should be as insisted by Aurelien?

The gamma is still changeable but defaults to 1.0

2 Likes

Radius is in fact the wrong name. It’s the sigma for gaussian kernel.

So what boost is doing is adding weight to the edges of the kernel. A Gaussian blur of a point fades out at the edges but boost in your example does the opposite increasing values with distance.

It does not add weight to the edges of the kernel. It adds an offset to the sigma of the kernel used for the outer regions.