I get round it by using ACR to get a very basic, low contrast file. Both the noise reduction and sharpening options in Photoshop are in my opinion a lot more subtle and less destructive than those in ACR. This is why I don't use Lightroom because I wouldn't use 99% of it. If you have Photoshop, try this. Open your raw file in ACR. Turn off all sharpening and all colour noise reduction. Its even a good idea when you start doing this to ignore CA removal as well. Turn highlight and white faders all the way to the left (highlight recovery) and shadow faders to the right (lightening shadows) Keep contrast relatively flat.
You should now have a very low contrast, very flat and pretty dull looking file in Photoshop. Bring the contrast and colour back by using the controls in Photoshop and experiment with the different sharpening options. I use a radius size of 0.5 pixels and anything from 100-500% in terms of amount. If you have to remove any colour noise, which from what I've seen is where I suspect the smearing watercolour effect comes from, use the remove noise option in Photoshop filters adding only enough to get rid of the noise and then move up the sharpen details fader until you get a nice crisp image. It takes some experimentation, but once I got what I wanted I saved most of this as an action.
Photo Ninja certainly produces sharper files with a different demosiacing algorithm but they are also noisier and I'm not that impressed by PN's noise reduction or CA removal. Getting all this to suit what you do is a huge balancing act and requires juggling things around until you get what works. It took me a while to get it right.
I suspect that all of the above and the way in which I do this explains the fact that I see an improvement in the latest ACR update and others don't. I am now getting files completely free of these smearing and watercolour effect problems. I do however have to accept that I'm never going to get the ultra crisp detailed images I can get effortlessly from my Sony A7 and A7r, but I can get my Fuji X-files pretty close. But I think we have to accept that both Fuji and Adobe are favouring noise-free images over sharp detailed ones that have some noise remaining.
I don't know any of this for a fact, but I believe Fuji have decided that high ISO performance and the ability to have ISO 400 and 800 as usable everyday settings with all the benefits of stopping action, using long telephoto's etc. is more important than low ISO ultra sharp high resolution detail. Leica used to prefer the other extreme, but even they seem to have succumbed to noise paranoia to a certain extent. And it's not hard to understand why. When a new camera is announced everybody wants to know 'How does it perform at ISO 1600, ISO 3200 etc.' I see very few queries about what a sensor produces at ISO 100. You only have to look at any OOC jpg. to see that noise reduction is a priority. And Fuji raw files can be very noisy. Try one in any basic dcraw programme and you will see what I mean.
Now I do think that both Adobe and Fuji are making some efforts to get round this. I believe that every ACR update and those endless Fuji firmware updates are slowly tweaking the files to get this noise / sharpness equation better. Adobe have been altering ACR parameters for cameras without any announcement for years. There was a demosaicing problem using certain wide-angle m-mount lenses on the Panasonic G1 years ago, creating little grid squares in the image. It did however completely disappear after an ACR update.
Plus there is this. When I asked Jim Christian, the developer of Photo Ninja to write some comments on what the thinking was behind his approach he wrote this:-
'Frankly, much of the academic literature has to be taken with a big grain of salt. The algorithms are usually tested on a limited set of synthetic images that are created from images that have already been post-processed. That is, they don't actually test on RAW files. Moreover, most fixate on achieving a low signal-to-noise ratio, which I've found to be a poor indicator of image quality. A few researchers seem to be realizing this (mostly the ones who work more closely with industry practitioners). But there are a number of algorithms out there that employ sophisticated math and look effective on paper, but don't seem to hold up very well in a more realistic setting.'
I think Fuji don't hold this view and do in fact 'fixate on achieving a low signal-to-noise ratio' and I also believe that Adobe do as well. I could be wrong, it's not unknown! but to a large extent I think we are stuck with what we have and can expect only marginal improvements from the 'official' Fuji / Adobe collaboration. For more 'adventurous' solutions I think we will have to look to the likes of PhotoNinja and Iridient developer, but we will have to accept that they aren't perfect either.'
Please Respect That
N.B. to see more on the cameras and lenses featured in this post click on the relevant labels (tags and keywords) at the bottom of this post.
For commenting, discussion, posting your pictures, links and articles - join the Soundimageplus Blog Readers Group on Google+
follow by email
follow on twitter
follow on facebook
follow on 500px
follow on pinterest
soundimageplus blog readers pictures group on flickr
soundimageplus on YouTube
soundimageplus on Vimeo