I’m looking for the best way to brighten shadows in RT which gives a natural-looking result.
The Shadows/Highlights tool seems OK for slightly brightening deep shadows, but stronger values quickly lead to ugly artefacts.
While looking for solutions I stumbled upon a post in the old forum by DrSlony, where he talks about a “Blackwave” technique: http://rawtherapee.com/oldforum/viewtopic.php?f=1&t=5525&p=43903&hilit=black#p43903
Apparently it makes use of negative values for the black slider, plus curves and wavelets, but unfortunately the uploaded PP3 in the thread is no longer available.
I experimented a bit with negative values for the black slider, then added a tone curve to push deep blacks back down, then increased contrast for wavelets level 7&8 and brightened shadows in the “residual image” section. This is a first try, I’ve no idea whether I’ve used the tools in a sensible way: Brighten.pp3 (2.6 KB)
What are your thoughts on the best way to brighten shadows in RT?
Hey @Stefan. That would be me and my post. There are many ways of brightening a photo’s shadows, each has its own pluses and minuses. Blackwave is more of a special technique (or tool combination) which can work great in some cases (I’ll put writing a post about it on my TODO list), but the most simple and most often-used (by me) approach is to use the two tone curves from the “Exposure” tool - one curve to lift the shadows and reduce global contrast, the other to put back some punch where needed. Being able to use different curve modes for each curve gives you much room for artistic expression - one curve could be set to reduce saturation (Saturation and Value Blending) while the other could be set to add it (Film-Like). I may sometimes use the “Shadows” slider of the “Shadows/Highlights” tool with a very low value, between 5 and 10, to lift the darkest areas. It tends to work well if one of the tone curves is set to push down the shadows (make shadows darker) in combination with the Shadows slider lifting them a little bit. Perhaps that sounds a little strange, but try it, the result looks natural in moderation.
I use the black slider and usually a curve. A similar thing can be done by moving the lower left point on a curve so I sometime do it there as the effect on contrast can be seen. The default straight line represents contrast over the tonal range. Increasing the slope increases contrast and reducing it decreases contrast. It’s usually not a good idea to create small humps over a limited tonal range but it can be done. So in effect raising a point on the line increases brightness at that level in the tone range which in turn increases contrast below the point and decreases it above.
I then use the other sliders to titivate the images. The highlights might get pushed out for instance to keep contrast in them but can be pulled back in with the recovery slider.
Some times it’s best not to recover all of the dark stuff. It depends on the shot. The black slider or curves can be used to chop some proportion of the dark end off completely by using positive values on the black slider. This can be used to remove low level noise. It can also leave more room for playing with contrast in the final image and sometimes that is more important. In other words it can help get round the problem of fitting a 12 or 14 bit raw image into the 8 bit sRGB colour space as some is just thrown away.
The preface to a shadows issue is what I try to do to almost every file: apply this excellent curves tutorial to the Lab-L curve:
The tutorial is written in an RT setting.
If I have not raised shadows quite right, next I try Shadows/Highlights or I turn on Tone Mapping. I don’t know why, but the change from no tone mapping to Tone Mapping at strength 1 is dramatic, and further strength increase is incremental. Whichever of these two I try, I usually pair it with an increase in the Black slider.
Tone mapping can be done another way as well as in RT. It’s possible to generate 2 different exposures from a single raw file. Exposures isn’t really the correct word but will usually be involved but other steps can also be taken on each of the images. They can then be merged with MacroFusionEnfuse which doesn’t produce the typical HDR look. I posted a silly example on Flickr. The eventual water could vary considerable from what I chose to use. I hope this package is still maintained. Linux only I think but there is a low cost windows version available - I think, This should link to an album.
I tend to use which ever package is best for the shot. The GIMP can be used as well. Shadow recovery is pretty simple. Duplicate the layer, desaturate and invert the top one. Then try various layer modes on it. Several duplicates of this layer can be placed on top of the original image. Better results are obtained by using lower levels of opacity on each one rather than 100% on all.
Enfuse is libre software, it is alive and well and comes with Enblend. It has only a command-line interface. MacroFusion is a graphical user interface for Enfuse. It is also libre software, written in Python by @Dariusz_Duma.
Seems that there is another way - selective and it will do a number of other things. Using NIK plugins via the GIMP. This can even be done on Linux via wine. There are other pages about on the same thing in win. Maybe mac too.
That probably means that we can all now produce nice sterile nothing like reality photoshop images easily.
I have gained the impression that play with linux may be the easiest way to get wine with a good selection of dll’s. I’ve been using Q4wine for launching but only use it for some very old utilities that do bizarre none photo things. Mostly if not all engineering things,
I work with darktable and certainly the shadows/highlights module does, in general, a fine job.
Recently however I ventured into ‘new territory’ for me … as I played with the ‘lowpass’ module.
Lowpass is not an easy module to use and it require blend lighting to accomplish the final result.
What is, in my opinion, unique in this module is that not only will it brighten the shadows it will, like magic, enhance shadow detail.
As the lowpass radius is changed downwards, emphasis of detail is moved from larger to smaller objects in the image and a negative contrast value pulls valuable detail out from the shadow areas.
It is worth while looking at the darktable manual and some of the on-line videos to get an appreciation of how helpful, despite its complexity, this module can be.
There is an interesting shadow recovery task some one posted here recently. I’ve tried to note what I did to a few tries. The tough part is getting the colour of the dog more correct and the rest reasonable.
I’d emphasise the point Morgan made about the curves. They are what the shadow and etc recovery use but with less control. GIMP’s level controls are a sort of in between as far as simple use is concerned but also have the advantage that colour balance can be changed. That can be more tricky to do with curves but that facility isn’t available in the exposure curves in RT and it doesn’t have the same style of levels control either.
As requested by Morgan … here are before and after parts from an image shown at 100% scale.
This image had very strong shadows and I had used shadows/highlights to open us as much as was possible but there was no contrast in the shadows and consequently no detail.
Using darktable with the following parameters made a significant change IMO.
Radius=14
Contrast -2.50
No change to brightness or saturation
Blend mode=vividlight
Opacity=20%
I use darktable for 100% of my image processing and file management.
The blue horse is a bit more tricky. Am I correct in thinking that the body is a little dappled in places and not pure white? Speckled might be a better way of putting it.
I doubt it. I may have been influenced by the tone mapped lookish one that was posted. Trees to gritty type problems for me tend to be down to my use of RT. Like many I usually settle that aspect when an image has been reduced while it’s being viewed full size. A general sharpening type route is a very mild one while the image is full sized and then a final one after reduction. The first one is just guilding the lilly really, a 1 pixel rad. Last one varies.
The tree lighting is wrong in places and could easily be fixed in the GIMP. Depends on how big the image is to me.
Just a thought so don’t get this wrong. My monitor is calibrated to exactly where it out to be in terms of brightness and colour - as close as my calibrator is able to get it that is. Supremely accurate ones are rather expensive so I use a colormonkey that can also measure ambient light levels.
Thought I would try an unusual one on the horse. No idea what it should look like and might change my mind about general light levels. Done too quickly.
2 Images from RT. Then the GIMP. One used as a desaturated inverted overlay layer with a layer mask to restrict it to the horse. I seem to have got away with some rather quick painting.
Only difference in the PP3 for them was lab brightness.
That probably means that we can all now produce nice sterile nothing like reality photoshop images easily.
No idea what it should look like and might change my mind about general light levels. Done too quickly.