Having looked at the default conversions in our RAW converters comparison we can now evaluate the ability of each application to resolve small, high frequency, details in an image.
The photograph chosen here is of Hoyd Tamir Gol in central Mongolia. This is about as far as it is possible to get from the sea on our planet, which leads to very low humidity and negligible atmospheric distortion. The Nikon D80 and 18-200, equipment of very average capabilities, manage to produce a very detailed capture in these conditions. The image was taken at ISO100, 1/80s, f/11 with an exposure bias of -2/3 eV.
The original image (rendered at Lightroom defaults):
In the first instance, what I am interested in here is to discover if there are any differences due to the demosaicing algorithm used in each RAW converter. To this end in the first set of 1:1 image crops all sharpening, noise reduction, micro contrast and lens corrections have been disabled in each RAW converter (this doesn’t prove that the RAW converter is not surreptitiously doing any of these things of course).
Secondly, I am interested in finding the maximum possible detail that each RAW converter can extract from the image, using its sharpening and micro contrast enhancement tools (where available). In processing the RAW files I am looking to produce a capture sharpened image, not an image for final output, so I want to avoid, as far as possible, the creation of artefacts, obvious halos and general ‘crunchiness’. Where deconvolution sharpening is used by a RAW converter it is hard to completely avoid any tiny artefacts, especially where specular highlights are involved, and when unsharp mask (USM) sharpening is used there will always be some haloing as this is what this process relies on. Nevertheless, I will try to minimise these things and produce files that would be suitable for further processing without amplification of introduced problems. This is necessarily a subjective judgement and I am sure others would do things differently.
I have placed the second set of 1:1 image crops illustrating this directly under the first set to facilitate comparison.
[Click on each image to bring up a larger version in a new tab; make sure you do this if you are a retina display user.]
The Unprocessed RAWs
At the outset I think we can lay to bed the notion that any of the RAW converters has a demosaicing algorithm that is notably superior or inferior to the others. The level of detail rendered in the files is almost identical. If I had to pick a winner I could just about convince myself that RAW Photo Processor produces marginally more detail in this particular image. Just about.
Processed Raws: Settings & Notes
Sharpening – Amount: 45; Radius: 0.6; Detail: 70; Masking: 0
Notes: A small increase from the defaults. The details is very fine in the image hence the low radius. Raising the detail amount increases the balance towards deconvolution sharpening, rather than USM, if I understand how the slider works correctly.
Edge Sharpening – Intensity: 0.55; Edges: 0.8; Falloff: 0.5
Notes: RAW Fine Tuning, which includes some capture sharpening, was reenabled to defaults. The Definition tool in the Enhance panel had too large a radius to help with the detail in this image.
Capture One Pro
Clarity – Algorithm: Neutral; Clarity: 10; Structure: 70
Sharpening – Amount: 60; Radius: 0.6; Threshold: 1.0
Notes: As noted in the previous section, C1P’s defaults are very aggressive; the sharpening applied here is around a third as strong as the defaults.
Local Contrast – Strength: 20; Sensitivity: 5
Sharpening – Amount: 120; Sensitivity: 6
Notes: It’s good to see a local/microcontrast adjustment with the a radius adjustment. It is unfortunate that it doesn’t really help the level of detail in the image very much.
DXO Optics Pro
Lens Softness – Global: 0.4; Details: 90; Bokeh: 50
Notes: It is very hard to see much of a change when moving the global slider. Adding even small amounts of USM dramatically increased the deconvolution artefacts produced by the lens softness tool, so it was not enabled. The microcontrast slider applied too large a radius to help with this image.
Richardson-Lucy Deconvolution – Radius: 0.6; Iterations: 25
Notes: Irident Developer provides 4 sharpening algorithms. Richardson Lucy deconvolution was the best on this image. Unfortunately there is no micro contrast tool at the current time.
Sharpening – Strength: 80; Sensitivity: 0.6
Notes: Photo Ninja’s detail slider appears to raise both microcontrast and midtone contrast. It works well at low values. The sharpening tool offers very fine control.
RAW Photo Processor
Focus Magic: 2 Pixels, faded to 80%
Color Efex Pro Detail Extractor – Amount: 20%; Contrast: 6%; Saturation: 6%
Notes: Focus Magic creates a few artefacts and had to be faded back. A little more detail can be extracted in Color Efex Pro without causing any additional damage.
As with the unsharpened images the differences between the second set of files are minor with this image. For me, the winner by the tiniest margin is Irident Developer, but it is so closely followed by Photo Ninja, Capture One Pro, RAW Photo Processor and DXO, that it would make no difference to anybody in practical usage. Others would doubtlessly call it differently. Irident Developer manages, to my eye, to just bring out a little more detail in the central band of trees. RAW Photo Processor and Capture One Pro appear to bring out a little more in the foreground stones, and Photo Ninja and DXO do a very good job on the dead tree. Irident Developer and DXO are both using deconvolution algorithms but DXO introduces more artefacts into the image and borders on being too crunchy in places in the trees. Irident Developer lacks any microcontrast adjustment at the current time; if it had it then it might produce a slightly punchier image. Nevertheless, the differences really are tiny between these five and one’s preference is probably driven as much by aesthetic appeal as judgement of absolute high frequency details resolution.
Of the other three RAW converters, AfterShot Pro is very obviously the weakest. Lightroom also appears very slightly more blurry than the leading five and Aperture’s sharpening algorithm is a little crude and introduces haloes and general crunchiness.
On this image, shot at low ISO, with very little noise and with no noise reduction necessary, there is not a great deal to choose between the RAW converters. Not all images are shot in such ideal conditions however, so we will turn next to seeing how well noise can be controlled and details retained in High ISO images.