16 replies to this topic

[loftus]

Well I've been been reading all this hyperbole about RAW conversion, particularly LR3 Beta 2,and I wanted to see what all the fuss is about. So I thought I'd start playing with RAW conversions from old Aperture 2, new Aperture 3.02, Lightroom 3 Beta 2 and NX2.
I chose for my first test a lousy image that shows one of the main problems we encounter underwater with sunballs and did simple RAW conversions of NEFS with standard default presets in all these programs. I did nothing else, no attempt to apply noise filtering, sharpening, adjusting presets etc. These were shot with my D200 so likely to bring out the worst in all these programs.
I'm posting these images in order, the screen grabs are obviously not as easy to evaluate due to the changes with web conversion etc, but I'll give my two cents.
NX2 standard, old Aperture 2 , new Aperture 3.02 and LR3 Beta 2.
My observations are as follows.

NX2 was definitely the smoothest conversion but still with some cyan banding, and interestingly keeping information throughout the image even the brightest highlights.

Aperture 2, the worst Cyan banding, keeping info throughout the image - overall my least favorite conversion.

Aperture 3.02, the least cyan of all, more blue with some banding, but actually the easiest image to manage with color balance tools. Also appears to be the sharpest conversion when examining the underside of the manta and the least noise. I have to say the image is much more blue on my desktop monitor before web conversion.

Lightroom 3 Beta 2 more cyan banding than NX2, but much better than Aperture 2 and worse than Aperture 3.02 Outer transitions are smoother than Aperture 3, inner transitions are worse.
Conclusions: None of these programs are perfect but I was pleasantly surprised with Aperture 3.02 as it is my database / import storage program. I like the blues, the detail and noise characteristics. LR 3 is good, but not nearly as smooth in the transitions as NX2. Smmother than Aperture in some areas, worse in others. Default sharpness of LR and NX2 are less than Aperture. Noise, Aperture and NX2 the best, though none are bad.
Ultimately I still think I will use NX2 for some difficult images, particularly for it's smooth transitions. Neither LR or Aperture are ideal in this regard, and which I would choose if there is banding would probably be image dependent. One thing about Aperture is it had the least cyan, so color balancing is simpler having to correct more for blues only and not cyan as well. For most standard images I would go as far as to say I am very pleased with Aperture and I am comfortable using it as my baseline RAW convertor, maybe even preferable to NX2 and LR. Difficult images I will compare NX2 and Aperture and choose the one I prefer.
I will continue to look at my images and compare, this was just a quick down and dirty test o see if I should be dumping Aperture. Finally the huge hyperbolic gushing ovr LR3 Beta 2 hardly seems justified. Certainly better than before but hardly superior to the others out there.

Edited by loftus, 07 April 2010 - 05:35 AM.

[jeremypayne]

I don't see the point in comparing these applications at their default default settings ... I think the only fair comparison of converters involves getting the best you can out of each and comparing those results - which itself is quite subjective.

When you do that, you see that they all can produce nice images ... and you are left comparing feature sets.

[loftus]

I don't see the point in comparing these applications at their default default settings ... I think the only fair comparison of converters involves getting the best you can out of each and comparing those results - which itself is quite subjective.

When you do that, you see that they all can produce nice images ... and you are left comparing feature sets.

Unfortunately the only way to compare is to use defaults as there are obviously way too many variables once one starts adjusting the defaults. I would assume that each manufacturer is providing what they consider the best standard RAW conversion as their default.
The point I make is that the hyperbole with regard to LR3 Beta 2 being so superior is in my opinion unfounded. For Nikon, NX2 still gives the smoothest transitions by far in the type of image I have shown. Both Aperture 3 and LR3 are improvements over their previous iterations, but both show banding which may be difficult to overcome with adjustments, and I think neither is vastly superior to the other. So difficult images with banding, choose NX2, standard images without these kind of problems, pick your poison.
The other point in terms of 'Nikon blues' NX2 is still my favorite followed by Aperture 3 followed by LR3.

Edited by loftus, 07 April 2010 - 06:54 AM.

[Steve Williams]

Hey partner, unfortunately it simply doesn’t work that way. I’m afraid your basic assumptions are fundamentally flawed. Your scenario is akin to throwing a bag of hammers on a pile of wood and waiting to see which one builds the best house. The software you’re attempting to compare are tools and they give us an amazing ability to create the images we want, but they require some knowledge and skill to use. I’m really surprised that you would make such a global recommendation based on your approach.

The excitement over the LR3 beta is around the much faster process engine and the luminous noise reduction slider among other things. To take a look at what’s possible you would have to move the slider control to see any result and the same is true for Aperture.

The baseline setup of the software can’t possibly be what the programs believe to be best. There are too many different cameras, different formats, etc. Most importantly “best” is a purely subjective term. My “best” image is the one that comes closest to looking the way I envisioned it. How in the heck would some programmer know that?

I’m all for understanding which program provides us the best opportunity to develop our images and I applaud your intention; we just need to come up with a way to do it that’s a real representation of the products capabilities.

Steve

[loftus]

Steve,
I obviously can only speak for one camera as I have ( Nikon D200) with a difficult image as shown. I am speaking only of one simple step, RAW conversion.
I am simply stating that there has been a lot of hyperbole with regard to the Aperture vs Lightroom discussion including pretty global statements about the RAW conversion, and statements such as 'The metadata issues are just a complete deal-breaker for any serious photographer'. Of course I have only tested one image, and I plan to test more. Bottom line; with that image LR is not my preferred RAW conversion engine. If folks are so convinced of the superiority of LR Raw conversion over Aperture and NX2 then they should post their results demonstrating this, otherwise it's just fanboy stuff. I use default because in my opinion it's the best place to start; of course every image can be adjusted tweaked etc, but they can for any convertor and the combinations and permutations become massive and results totally incomparable and quite frankly based on the skill of the tweaker. I can move sliders in any program, not just LR.
I am not discussing any other features of any of these programs, simply baseline RAW conversion of an arbitrarily chosen image. My conclusion so far is that there is nothing to make me choose Lightroom over Aperture or NX2 based on this one simple step. If one chooses Lightroom for other reasons that's quite valid, as do folks who chose Aperture, NX2, ACDC, Capture One etc. I just don't see any evidence to the claim that folks should switch to Lightroom because of a RAW convertor that is inherently superior to other presently available RAW convertors including Aperture 3.
Personally the initial RAW conversion is a very critical step that I was unhappy with in Aperture 2, but I am now very pleased with in 3. This was a concern that was edging me towards changing to LR, but I held off because I actually dislike the LR interface. Other issues such as the metadata etc are not an issue for me, but I don't have the needs that say Eric does.

Edited by loftus, 07 April 2010 - 09:16 AM.

[echeng]

I'd like to see this same test done with a modern camera's RAW file. To my eyes, this image fails in every RAW converter -- the cyan is simply too overwhelming. Newer cameras seem to be much better at capturing sunballs. The test is also flawed because Lightroom has camera-specific profiles that look different than Adobe's standard profile. I always use my camera-specific camera profiles when doing RAW conversions in Lightroom.

Regardless of which one has the better processing with default settings, if you're using a new camera, you may have to wait months for Aperture to work with it. That is definitely one reason to choose Lightroom over Aperture!

[loftus]

I'd like to see this same test done with a modern camera's RAW file. To my eyes, this image fails in every RAW converter -- the cyan is simply too overwhelming. Newer cameras seem to be much better at capturing sunballs. The test is also flawed because Lightroom has camera-specific profiles that look different than Adobe's standard profile. I always use my camera-specific camera profiles when doing RAW conversions in Lightroom.

Regardless of which one has the better processing with default settings, if you're using a new camera, you may have to wait months for Aperture to work with it. That is definitely one reason to choose Lightroom over Aperture!

I purposely chose a bad image to try to highlight differences. I'd expect an easier image to be even more difficult to judge. I remain convinced that NX2 is still by far the best convertor for Nikon for this type of image, even if there is a Cyan cast. I can correct a cyan cast in fairly straightforward fashion. Banding on the other hand, evident in both LR and Aperture is more difficult.
OK, when I start buying a new camera every month, I'll change to Lightroom.

Edited by loftus, 07 April 2010 - 10:41 AM.

[jeremypayne]

Post the RAW file and I'll do my best with LR3 ... and you do your best with Aperture and NX2 and then we can compare ...

[loftus]

Guys, guys......forget it. I'm not trying to see who can get the best result by manipulating my image, I'm just trying to make a point that LR step one, with no manipulation, none, nada, is not intrinsically better than the others out there. so bring out your files, enter your camera standard presets, do the same for Aperture and NX2 and show me that LR does a better job than the others. But NO manipulations.

[jeremypayne]

Guys, guys......forget it. I'm not trying to see who can get the best result by manipulating my image, I'm just trying to make a point that LR step one, with no manipulation, none, nada, is not intrinsically better than the others out there. so bring out your files, enter your camera standard presets, do the same for Aperture and NX2 and show me that LR does a better job than the others. But NO manipulations.

All processing is "manipulation"; I'm not sure I understand what is so special about the manufacturer defaults ... at least with LR, they aren't supposed to "the best" ... just a fairly neutral starting point.

My final settings are no more or less a manipulation than the default default - which I personally haven't used since the 1.0 Beta ...

I have defaults per camera ... and per ISO ... that I created, not adobe ... and they are quite flat with linear TRC and varying degrees of sharpening and NR based on ISO and camera.

But heck ... if you don't want to see what LR can do ... you don't want to see what LR can do ...

[TheRealDrew]

But heck ... if you don't want to see what LR can do ... you don't want to see what LR can do ...

It sounds like Jeff is saying that straight out of the box initial conversion, and where it gets him in the ballpark to start from, LR didn't float his boat as much as he thought it would based on some of the discussions about LR3 and what people want. But you know how it, some people like Canon, some Nikon, some Macs, some PCs.


The ones that shoot Canon and use Macs are obviously the superior people who make proper decisions, but the others are entitled to think what they want.

[loftus]

All processing is "manipulation"; I'm not sure I understand what is so special about the manufacturer defaults ... at least with LR, they aren't supposed to "the best" ... just a fairly neutral starting point.

My final settings are no more or less a manipulation than the default default - which I personally haven't used since the 1.0 Beta ...

I have defaults per camera ... and per ISO ... that I created, not adobe ... and they are quite flat with linear TRC and varying degrees of sharpening and NR based on ISO and camera.

But heck ... if you don't want to see what LR can do ... you don't want to see what LR can do ...

Jeremy,
All RAW convertors can have customized settings. Hell all settings, even Photoshop manipulations are just clicks of a mouse. And if you want them non-destructive use layers and smart filters, even actions to standardize things and automate the process. That's simply not my point.......
If you want to show what Lightroom can do that's fine, just post with similarly preset / adjusted / manipulated NX2 and Aperture 3 images for comparison, so I can see that LR really is better. Sigh......

[loftus]

The ones that shoot Canon and use Macs are obviously the superior people who make proper decisions, but the others are entitled to think what they want.

Thats clearly why I can't understand this, I don't shoot Canon.

[jeremypayne]

so I can see that LR really is better. Sigh......

Go back and read my original post ... my expectation is that all three converters in qualified hands would produce comparable results and you would most likely end up picking your poison based on feature set.

Where the new LR excels relative to the old LR is with respect to noisy images. Images at base ISO on most DSLRs do not really show off what the new pipeline can deliver - the more noise, the more you see the impact of the improvement ... AND ... on top of that ... at defaults, the new pipeline often looks worse - but what you can ultimately achieve in your raw conversion is FAR superior.

Comparing LR 2.x to LR3.0 in the manner you did would likely convince some that the LR 2 engine was better - which is just not the case. That's why I think comparing at default settings that aren't truly comparable and merely arbitrary starting points is often misleading ...

[loftus]

Go back and read my original post ... my expectation is that all three converters in qualified hands would produce comparable results and you would most likely end up picking your poison based on feature set.

Where the new LR excels relative to the old LR is with respect to noisy images. Images at base ISO on most DSLRs do not really show off what the new pipeline can deliver - the more noise, the more you see the impact of the improvement ... AND ... on top of that ... at defaults, the new pipeline often looks worse - but what you can ultimately achieve in your raw conversion is FAR superior.

Comparing LR 2.x to LR3.0 in the manner you did would likely convince some that the LR 2 engine was better - which is just not the case. That's why I think comparing at default settings that aren't truly comparable and merely arbitrary starting points is often misleading ...

I made no mention of LR2, I do not have this program for comparison. I only presented the 3 most current RAW processors from Apple, Adobe, and Nikon. I did present Aperture 2 however as the worst of the bunch for comparison.
Honestly noise is not at all my primary concern in these images, noise differences were quite subtle even though I did mention them. I agree that all these convertors are good and can produce comparable results, I was simply challenging the notion that LR3 is somehow intrinsically as good as NX2 and better than Aperture, which I have heard claims to the effect, but not seen any evidence. Frankly if one has to have a whole lot of different presets as you do, it defeats the purpose for most of us. Similarly if it takes a significant amount of playing with sliders to get the ideal image in RAW, before one even starts the actual post-processing, it partly defeats the purpose of a simple smooth workflow. Seems a little odd that all of a sudden more noise in the initial RAW output is considered a bonus just because you have a slider that can then remove it. And again who is to say that if one manipulates the presets and sliders in any RAW processor, that similar improvements to those claimed for LR, cannot be achieved. So I'm not putting down LR, I'm glad LR users are happy with their new product, just don't make claims of superiority without demonstrating that similar results cannot be achieved with other RAW processors. In presenting my findings, I had to start somewhere, and defaults seems to be the logical place to start. I was in fact hoping that others would post similar tests, maybe demonstrating I was wrong, rather than making excuses for LR superiority claims by shooting down my methodology
Finally, and still my main concern in the type of image I posted, was the type of banding one sometimes has to deal with in images like this. Yes, new cameras are better, but they are not perfect, and under certain conditions relating to depth and visibility, this type of blooming can still be a problem from what I've seen. I can't say I fully understand the problem, but they have to do with dynamic range, and gamut issues I think. A cyan cast that is smooth is manageable in post with color balancing adjustments, the type of banding demonstrated in my LR and Aperture examples, and not in NX2, is much more difficult to manage in post, particularly when preparing for printing. Obviously if one can start out without the problem, and not have to correct it, that's a bonus. RAW processors are simply a recipe, with regard to banding NX2 still seems to have a better recipe.

Edited by loftus, 08 April 2010 - 07:01 AM.

[Tom_Kline]

I made no mention of LR2, I do not have this program for comparison. I only presented the 3 most current RAW processors from Apple, Adobe, and Nikon.

LR3 is presently a beta app, LR2 is in fact the current version.

After looking at the first post I would select 'none of the above' since what hit me first was the lack of a good Dmax in all the examples.
I think LR3 will make getting a good Dmax easier, at least I was able to with the D2H image I posted in another thread.

As Eric mentioned there are camera specific profiles, as well there are several for some camera models that would directly affect pix before any slider tweaking commenced.

I think an important consideration is how one uses their photography as well as ones work flow. I use my library as a research tool (for both science and photography per se) so frequently access the 'database', e.g., several times in the last few days for different purposes. LR may be better for this. LR may work better on laptops, a very important consideration for working on the road. Once Aperture is debugged it may be better for multimedia (re. Tony Wu). etc. etc......

Adobe is a big graphics arts player so I fully expect they will eventually address the need to deal with video for presentation (appears to be strictly a library function thus far in LR3, maybe b3?, but more likely a later LR) - maybe when more magazines are formatted for Ipads and the like, which will likely have video content. I would not be surprised if they came out with a 'videoDNG' to help solve the video standardization issue.

Tom

[jeremypayne]

I made no mention of LR2

I wasn't suggesting that you should have.

My point was this:

• Lightroom 3 is equal to or better than Lightroom 2
• If you were compare LR2 to LR3 in the same manner as your example, many would pick LR2
• Therefore, the method has issues, IMO

I was in fact hoping that others would post similar tests, maybe demonstrating I was wrong, rather than making excuses for LR superiority claims by shooting down my methodology

Jeff ... you aren't reading my posts ...