RAW processors

I can see the difference there too - I don't think the PSE method has worked as well - you can see where the tones have been extracted from the branches and not all of the magenta channel has been removed.

The Lightroom examples above have dealt with the colour and fringe defects, but the issue is the highlight areas remain blown (it appears - Smeggy, what are the white values on your clouds there?) - RawTherappe looks to have managed contrast control much better, there's a smoother blend between the very bright whites and very dark under-branches, which I noticed in my own version using RT as well.

Smeggy, you mentioned the adjustment brush as an extra step to control the application of the effect - but is this tool entirely practical in this scene with so many fine intertwined branches to work with?

I've no idea I didn't even try and recover any highlights from the image. I'll try and remember to take a look later


It can be. The Lightroom Adjustment brush has an auto mask feature which when enabled will try to confine the mask to the general colour-luminosity your brush stroke was first clicked. not perfect but very handy

I've hade a play around with RawTherapee and this is about the best that I've been able to do. One problem is that the best highlight reconstruction result screws up other parts of the photo. I think I'll have to spend a lot longer learning RT to be sure I've got the best out of it.




Andrew

I've had a look at this and I don't think this particular effect is a fault of any of the raw processors but instead a natural limitation of the sensor (and probably also lens resolution). Colour rendition of the larger branches is generally fine until the branch diameter falls below pixel or bayer pixel group size. At this point the sensor is recording a light value from an area of white or blue sky with a black/brown line across it and hence the only reasonable value that pixel can have is light grey or blue respectively hence the apparent colour bleed.

The colour bleed can be removed with selective desaturation in the blue sky areas and the pale blue branches then become pale grey BUT on close inspection it is still looks bad because we now have a light grey branch that is unnaturally thick until it dissappears i.e. the pale grey branch is one pixel wide for a very extended length just reducing in lightness rather than tapering in thickness. To remove this effect I think you will need a higher resolution sensor and probably better lenses, although of course it won't actually remove the effect but just restrict it to even finer details.

That seems like a plausible explanation, although there is a definitely a difference in how the various RAW processors handle this, ACR(8.2) doing the best job. The best final image that I am happy with is not all that different from the in-camera JPEG - sans PF, CA and with just a bit of exposure etc tweaking.



Full size here.