RAW Converters Comparison
2. Defaults
In this RAW converters comparison this section’s goal is to establish exactly the default behaviour of the various converters. We will see whether lens profiles are applied, any sharpening or noise reduction takes place, and see some very different interpretations of auto white balance.
A RAW file starts out as a bunch of 1s and 0s representing a 2*2 Bayer colour filter matrix of 4 photosites – 2 green, 1 red and 1 blue. This has to be interpolated, or demosaiced, into a standard RGB image. The algorithm used for this process varies between RAW converters. Once converted into an RGB file the data requires further interpretation via camera profiles to tell the software what colours things should be in various lighting conditions. RAW converters will contain at least one profile for a camera, sometimes more than one, and you can usually create your own (not a simple and easy process unfortunately) and load that in too.
When a camera produces an image it embeds in that image information about the lighting conditions at the time; this is the white balance of the scene. If you shoot RAW then this is not baked into the file and the majority of people shooting landscape images will leave their camera set to ‘auto’ white balance. The camera usually gets this in the ballpark, especially in outdoor daylight conditions, and this can be fine tuned later in the RAW converter.
The default colours that a RAW converter produces are a combination of choices about the demosaicing algorithm, the camera profile the software applies, and the software’s interpretation of the information the camera has embedded in the file about the lighting conditions it recorded at the time.
As well as making these decisions some RAW converters go further when opening an image for the first time, and apply denoising and sharpening, and/or automatically correct lens defects, and practically all of them make exposure adjustments. The application of these modifications to the image is not always made explicit by the software.
The following image was taken with the Olympus E-M1 with the 12-40mm lens (at 40mm) at ISO 200 (base ISO for this camera), 1/1000s and f/5.6. It is slightly underexposed. I have chosen it because everybody has a fairly good idea of what colour grass, the sky and the sea should be (this colours in the image will change a little of course depending on how you have your monitor calibrated). [You can click on the images to bring up a larger version in a lightbox.]
The Conversions
The table below shows how each application interpreted the ‘auto’ white balance in the image, whether or not the lens has been corrected, whether or not noise reduction and sharpening was applied, and the resulting file size (these are 1576 * 1182 pixel files, i.e. double sized for retina displays, and have been produced by exporting a full size 16 bit uncompressed TIFF from each RAW converter and then using Photoshop’s Save for Web at 70% quality to output them).
App | Kelvin | Tint | Lens | NR | Sharp | Size |
---|---|---|---|---|---|---|
ALR | 5350 | +10 | Yes | Yes | Yes | 489k |
AP | 5733 | 21 | Yes | Yes? | Yes | 517k |
C1P | 5637 | 0 | No | Yes | Yes | 566k |
CAP | 4977 | 21 | No | Yes | No | 521k |
DXO | 5047 | 11 | Yes | Yes | Yes | 504k |
ID | 4980 | 14 | Yes | Yes | Yes | 565k |
PN | 5300 | 2 | No | Yes | Yes | 521k |
RPP | N/A | N/A | No | No | No | 477k |
Looking at the white balance of the images themselves it would be hard to say that any of them look particularly ‘bad’ or ‘wrong’. Our eyes and brains are easily fooled by slight differences in exposure levels that make images appear more or less saturated or cooler or warmer than in fact they are. The numbers in the kelvin and tint columns in the table are a little more informative (Raw Photo Processor supplies RGGB values). There is some considerable variation here and in the absence of very careful colour profiling and calibration it is impossible to make any judgement as to whether any of the software is ‘getting it right’. For the landscape photographer white balance is typically judged visually according to subjective preference, and in any case the image will often have its colours enhanced, or otherwise altered, in subsequent post processing.
The Olympus E-M1, like all micro four thirds cameras, supplies lens correction data to RAW converters in its EXIF files; there is no necessity to independently generate lens profiles as there is for almost every other camera. It is very easy to see whether these corrections have been applied here by scrolling through the images and looking at the truncated sheep on the left hand edge of the image. If the sheep gets further truncated then the lens correction profile is being applied. We will deal with lens correction in detail later in these articles but something to note here is that neither Lightroom or Aperture offer any way of disabling this automatic correction.
The final column containing the file sizes makes interesting reading, especially in conjunction with the noise reduction and sharpness columns. All other things being equal, file size is generally a very good indicator of how much detail is present in a file. A file with a large amount of high frequency detail, for example, an image of a beach where the individual sand grains are present, will be much larger than a file with an image comprised mainly of a blue sky. Here the image is the same, and unless significantly more detail is being revealed by the demosaicing algorithm used by each RAW converter, something else is going on.
What the file sizes reveal is that the image is being processed to a greater or lesser extent by the RAW converter at its default settings; most notably it indicates that sharpening and/or micro contrast adjustments are taking place. Sharpening and microcontrast adjustments introduce artefacts, artefacts that serve to (hopefully) make the image appear sharper and more detailed to our eyes, but these artefacts add to the file size. By contrast, denoising smooths over detail in files, and thus results in smaller file sizes, though this effect is outweighed by sharpening adjustments, especially in an image like this where very little noise is present to begin with.
Raw Photo Processor, where we can be reasonably sure from their documentation that nothing additional is being done to the image, has by far the lowest file size. Capture One Pro, where even a cursory examination of the image tells us that additional processing has been applied, has the biggest file size.
To clearly see what is going on with noise reduction and sharpening in these files they need to be viewed 1:1: [Note that clicking on an image will bring up an image in a new tab containing a matrix of the images, useful for comparing the conversions side by side, and essential for those with Retina screens who will be unable to see the details correctly in the images below.]
Comments
Lightroom has a notably small file size, considering that it is reducing colour noise only and is sharpening the file a little. It is applying its proprietary Adobe Standard color profile to the image, which looks OK with this image though, as we will see later in these articles, it can be improved upon. As noted, it automatically applies lens correction and this cannot be disabled.
Aperture‘s default noise reduction is a bit mysterious as we will see later, but it looks as if colour noise reduction is being applied, and we can see in its RAW Fine Tuning panel that some default sharpening is being applied, as is a contrast boost. This results in a medium file size and a default conversion that clearly has a little more contrast than some others. As noted, it automatically applies lens correction and this cannot be disabled.
Capture One Pro‘s rendition of the scene is clearly the most processed; it is darker, sharper and has greater contrast and saturation. It applies a custom ICC camera profile and also detects the lens correction data but does not apply it, which seems strange when its camera profile is so much punchier than anybody elses, and even more so because it actually applies some very heavy luminance noise reduction as well as colour noise reduction, and substantial sharpening, by default. Its documentation states that this is based on its profiling of the cameras but both settings are far too high in my opinion and yield results that are way too crunchy for default settings (we’ll come to this in more detail later). All this extra processing results in the largest file size.
Aftershot Pro applies RAW noise reduction, which appears to comprise a mixture of colour and luminance noise reduction, though the latter is more evident here, but no additional sharpening. This appears to borne out on examination of the image which appears softer than any of the other renditions but does leave one wondering why its file size is not smaller. The resulting image does not look very good.
DXO Optics Pro throws everything at the image with its sophisticated lens profiling, lens softness adjustments, which appears to involve deconvolution sharpening, and both luminance and colour noise reduction. No additional USM sharpening is applied which perhaps explains, in combination with the noise reduction, the relatively small file size.
Irident Developer has the second highest file size, just a shade under Capture One Pro’s, and this is accounted for by its lens adjustments, light colour denoising and sophisticated sharpening algorithms (which we will investigate later). Its colour rendition is quite flat and desaturated but this, arguably, gives a flat playing field for future colour enhancement with its powerful curve tools.
Photo Ninja‘s default rendition applies colour, but not luminance, noise reduction, colour profiling, contrast and exposure boosts (it is explicit about this), which combine to produce a very appealing image yet still with a medium file size.
Raw Photo Processor‘s default rendition is, in fact, not shown here as it is the only software not to provide exposure compensation out of the box. Following demosaicing the image is more than a stop darker than shown here and I have used its auto exposure compensation feature to the same level as the other conversions (an increase of 1.03 eV). With no noise reduction or sharpening the file size is, as noted, the smallest.
What this analysis of default renditions in our RAW Converters Comparison really reveals is differing philosophies about what should be initially supplied to the user. Photo Ninja, DXO Optics Pro and Capture One Pro supply something broadly equivalent to an out of camera jpeg (shot on default settings), albeit at much higher quality. They are images that could be used, at a push, without further editing. If I had to pick a winner here it would be Photo Ninja.
Lightroom, Aperture and Irident Developer choose to apply a little colour noise reduction and compensate for it with a little capture sharpening to give a starting point for further editing. They all do a good job of this but if I had to pick a winner it would be Irident Developer. Aftershot Pro appears to be attempting the same thing but the decision to apply some quite heavy noise removal and then not attempt to compensate for it is a strange one and it doesn’t make for an appealing default rendition.
Raw Photo Processor goes one step further, or perhaps that should be ‘less’, in providing a neutral editing base by doing almost nothing to the image. A solid starting point if you want to go to Photoshop and use specialist tools and/or plugins for further editing.
Having looked at the out of the box renditions in our RAW converters comparison we can now move on to look at their ability to render high frequency details.
27 Comments
Wow, what an excellent article. This is the best RAW comparison that I have ever read. Extremely detailed and well explained. Thanks for your hard work!
Very detailed and through presentation. I to have used Lightroom since it was in the beta stages and it is now the raw processor of choice. However I have always been of the opinion that it is not the Gold standard of raw processing software, not the same way that Photoshop CS/ CC can claim to be as an image editor. For me there at least a half dozen other choices that can match or surpass Lightroom in this respect.
What keep me with Lightroom is the ease of use and the wide range of tools available.
Very interesting and comprehensive review. A couple of small points, some Capture one sliders such as noise reduction are calibrated in a relative manner rather than an absolute manner – thus even though the default setting is always the same it can represent quite different underlying settings.
When adjusting dark images such as your example I find it best, when using process 2012, to follow Adobe’s advice and make a mid tone adjustment using the exposure slider before making any darks and blacks adjustment.
Excellent comparison, lots of work. I started using LR early on, switched to A3 and recently bought C1P. I agree with your conclusion and I am very pleased with C1P. Most files (E-M1 and E-M5) require very little adjustment when using the auto adjust function. Now I only have to work out the DAM for which I still use A3.
I have found ACDSee Pro 7 to be the equal of CaptureOne in terms of raw development. I believe one would be hard pressed to tell the difference at all. Plus, I’ve found that I like the ACDSee user interface much better. If you like the CaptureOne output but don’t much care for the “experience”, ACDSee Pro is worth looking into.
ACDSee is not available for OSX, but I used it several years ago with a P&S
Well there is an ACDSee Pro 3 for Mac, though I’ve heard it is not the mature product that Pro 7 for Windows is. Being a Windows guy, I’m not too conversant on the Mac OS versions. I don’t know if it is available for OSX or not.
I agree. I used ACDSee Pro for years. I have also tried various iterations of DxO and CaptureOne. For the past year, I have been using Lightroom almost exclusively. Ultimately, I think ACDSee Pro, or Ultimate as they now dub it, probably offers the best balance of features and usability. I was disappointed with DxO for all but its lens correction features which are stellar. I found CaptureOne to be good but remarkably ungainly to use. Lightroom seemed a good alternative but frankly none of them offered the intuitive DAM functionality that ACDSee Pro did. I’m transitioning back to that product now.
Hello,
What a great comparison report!
Thank you for all the efforts and the pictures for the comparison.
Due to your result I checked CP1 as a long term LR and DXO user.
Since own several comparisons with LR and DXO and training with the new program CP1, LR is not used anymore and DXO very seldom.
I am completely satisfied with the possibilities and the results of CP1
CP1 is a great RAW converter but it needs more learning time than DXO an LR to discover the complete possibilities of CP1.
But afterwards it’s fast and quite easy to get perfect results.
Greetings
Thomas
Thanks for the RAW file processor comparison! That’s quite an effort and really appreciated!! Cheers, Jeff
I think you have completely dropped the ball with Aperture by claiming it has “no localised adjustments”. You can brush in and out practically EVERY single adjustment there is on the adjustment panel of Aperture and for a long time now.
Otherwise it is a nicely done comparison but you have to realise, that you would have to do this kind of work with every single camera out there to give a complete picture. Some are great with Canon/Nikon and very poor with Sony, Fuji etc. and vice versa.
I hope the next time you do a comparison that you’ll take a look at SILKYPIX at silkypix.us.
Hi Nik,
maybe, just maybe – in case you’ll do something similar in future – you could include DT http://www.darktable.org/ . It would be nice to see where it stands among these ‘Goliashes’.
Or maybe you can check separately and referred this article – e.g. “my rating among the rest would be ~5th place” – thank you! – and maybe you can thank me 😉
I thought for sure you’d include RawTherapee in the shootout. Do you have any experience with it and how do you think it would slot into your rankings?
very helpful article. thank you !
I was wondering what raw developer you use for the Merrill’s. I have a DP2S and love it except for one thing and that is the color noise. Even ISO 100 I get red and green noise in shadow and also can be in a bright sky. I process them in SPP which I actually like unlike most people however it does not have any way to edit color noise. I would love to invest in a Merrill if I knew how to cut down the red and green noise. Thanks.
That was helpful. Thank you!
What a fantastic comparison! This must have taken a long time to do. Thank you for this!
Many thanks. Any chance of an update?
Nice, detailed test. It’s interesting that while I agree with your general ranking in conclusion, when I look at many of the individual comparisons I quite disagree with your judgements. Which I think just goes to show how subjective it all is, especially once you start editing. e.g. several times you preferred renditions which I thought were overly noisy.
Had I known about this post earlier, I might not have done my own (much more limited) comparison (http://blog.wadetregaskis.com/raw-converter-comparison/). Though I’m glad I did, because now between the two of us we’ve provided more data. And it’s nice that our data leads to consistent conclusions.
when i was buying my 1st dslr i used Lightroom as a raw converter. Then I discovered DXO was there.
Well dxo is just superamazing. I can highly recommend it.
It can do a lot automaticly.
In the beginning I thought: automatic? can this be any good?
the answer is you bet!!
Thanks to the enormous data-base they have build during dxo-mark benchmarks testing dxo-optics pro knows every little bad thing about your camera and lens and can correct it automaticly(but you can do manual too).
I did testing and comparing lightroom and DXO.
DXO really is amazing and it beats lightroom everytime.
All my photos look so much more amazing when i use DXO.
and I don’t talk about a little better then lightroom. I am talking about a LOT better.
I have not tried capture one yet but i am very impressed with dxo. even when you just use the dxo standard automatic settings it already beat lightroom hands down.
I really don’t understand the whole uzz about lightroom.
Lightroom is not really good to be honest. but sometimes i get the impression the entire world uses lightroom.
Such a wealth of information here. Think you! I am testing the Picturecode Photo Ninja as aiming for batch converting RAW into TIF for 3D scanning purposes.
Excellent review. Actually, we’ve been on Capture One for years now, but for the first time ever we’re considering abandoning it due to Phase One playing software politics by deliberately disabling support for the Pentax 645Z, which in our view is entirely unacceptable for paid software. This makes users direct victims of boardroom politics, and benefits no one. True, some canny workarounds exist, but sadly these are not acceptable in a professional environment.
GREAT JOB!
thank you!
I am interesting if Serif labs company which makes Affinity Photo
if they build non-destructive photo editing sw and for windows
– maybe first step is here –
Affinity Designer for Windows is here 🙂
http://photo-typ.blogspot.de/2016/07/affinity-designer-for-windows-is-here.html
I’ve done a raw converter comparison here: http://sjp.id.au/digital-darkroom/raw-converter-comparison/
It includes the newly released ON1 Photo RAW, and the results are interesting. For me Photo Ninja is probably the most unique in terms of rendering quality, but it doesn’t have the bells and whistles that Lightroom does.
Thank you for sharing the results of your likely considerable time and effort.
Thanks for a brilliant review, but any chance of an update with so many changes having taken place since you wrote it?