Aperture and Nikon RAW (NEF) performance.

Discussion in 'Digital Photography' started by rjp, Apr 29, 2008.

  1. rjp macrumors regular

    Joined:
    Feb 25, 2008
    #1
    How's Aperture do with Nikon RAW (NEF) files? I've heard that nothing is quite as good as Nikon's own RAW conversion in CaptureNX since it uses all the original in-camera settings as a starting point for the initial rendering, but it seems many users don't think the SW is very well written otherwise.

    Anybody using APerture with Nikon RAW and liking it?

    If so, have you also tried CaptureNX?

    Thanks,
    Rich

    Nikon D40
     
  2. compuwar macrumors 601

    compuwar

    Joined:
    Oct 5, 2006
    Location:
    Northern/Central VA
    #2
    You can test both applications yourself and see if Aperture lives up to your expectations. When I tested, Aperture was the worst of the group of NX, Bibble, Apture and ACR.

    If it's an important shot, I convert it first in NX, then if it needs distortion correction it goes to Bibble, otherwise to PS for any adjustments (because I'm more comfortable with the Adobe interface- I haven't tried too much to learn the whole control point thing in NX) and then it's cataloged in Aperture.

    Longer workflow than I'd ultimately like, but I want my images to be as good as they can possibly be.

    None of the converters converted the same way, and if I recall correctly, they didn't even convert the same pixels- there was a 2-3 pixel difference from the same NEF.
     
  3. rjp thread starter macrumors regular

    Joined:
    Feb 25, 2008
    #3
    I sorry to hear that. I was hoping Aperture would be as good as NX for NEF conversion. Why is it that nobody can match the quality of NX here? Is it that no non-Nikon SW can read the in-camera settings embedded in the NEF, or are Nikon's rendering algorithms really that much better?
     
  4. ChrisA macrumors G4

    Joined:
    Jan 5, 2006
    Location:
    Redondo Beach, California
    #4
    It's a metter of opinion which software does the best conversion. Also those people who argue about it are looking at 100 englagements which are like making 30 inch wide prints. It's a totally unrelistic test. We call them "pixel peepers" in real life and of the raw converts can product the image you want but each takes a different route to get to that image.

    Then you shoot raw all those setting you make for "sharpness" and white balance has no effect. those are used to control the raw to jpeg conversion process and if you don't do jpeg converson they are moot. But Nikon's software can put these settigs to use inside CaptureNX and have NX duplicate what the camera would have done had it beed set to record in jpeg format. Bt what's the point? If that is what you want shoot jpeg.

    NX has a poor user interface and it is slow. Aperture is designed for processing large numbers of photos and is set up well for this.

    One way to deside what to use Adobe Camera Raw, or Aperture or Lightroom or NX is how many phots you shoot. Do you come home with 1,500 images or 50? and then what do you do with them. Do you make prints or are they mustly going on the web?
     
  5. Mr.Noisy macrumors 65816

    Mr.Noisy

    Joined:
    May 5, 2007
    Location:
    UK™
    #5
    Personally I shoot 100% RAW (NEF), ive tried Nikon Capture and prefer Photoshop, i convert Images to Tiff using CS3, any editing then is done in CS2 as this is loaded with Alien skin, Nik & andromeda filters ;)

    p.s. I also tried Aperture but feel i got more control using PS :)
     
  6. compuwar macrumors 601

    compuwar

    Joined:
    Oct 5, 2006
    Location:
    Northern/Central VA
    #6
    Not really, you can see which software does, so the actual conversion is easy to see- you can argue between "best" and good enough, but "best" is basically easy to see and objectively measurable in terms of sharpness, tonal range, and other IQ-based metrics. If we were talking about a relatively similar basis for conversion, then it'd be different but the differences are objectively measurable.

    It's only unrealistic if you're not concerned about making large prints or producing the best possible image you can. For most people, "good enough" is well, good enough. For others, especially if you're selling large prints for lots of money, or you're looking downstream at upcoming reproduction technologies, or selling significant crops at relatively large sizes, that pixel peeping is a totally valid test. One of the galleries I'm trying to get in will sell fine art prints that are 8'+ on a side- and you can see the differences in quality of output at sizes like that. When you're hanging beside other photographers, that tiny IQ difference makes your work look better.

    If you look at the flying Eagle in my online gallery- the image will go *almost* to Super A3+ before it starts to fall apart when you put your nose up to the glass. It's a heavy crop, and Imagekind won't print it as large as I will- but even then, I'd hesitate to sell too large a print if you couldn't see the feather details in the tail on close inspection- even though you have to get inside the normal viewing distance to see that level of detail. Could I sell prints that don't meet my quality standards? Sure. Would I? Most likely not.

    If you lose a step of detail, you'll never gain it back. So, from my testing, I'd say that no- you can't really get the same results from all the converters- which is a shame and actually pretty puzzling since ACR should be using Nikon's own API- but it doesn't even convert the same pixels! Sure, it's very minute detail, but that'll show up as apparent sharpness in an image, it'll show up in extreme enlargements too. How important that is depends a lot on what you're trying to achieve, which is why I said "see if Aperture lives up to your expectations."

    The point there is for the times you don't want to wade through a long shoot re-setting WB, but you do want to provide the client with files that are editable multiple times without a loss of quality. Personally, I think you're better off shooting a white or gray card, but if the light's changing a lot and you're following action that can be difficult.

    Yes, but you can batch convert in NX without really spending a lot of time in its interface and still have the best-quality NEF conversion possible.

    All of them will batch process, so for me that's not the major decision point. I own PS CS3, Bibble Pro, Aperture and CaptureNX. I tend to use all of them for something, but mostly none of them every time. If you're only looking to do processing in a single tool, then you'll have to choose which one works best for your workflow- but that's not the only option on the table.

    Why shoot at 12-bit if all your output devices are 8-bit and the difference for a 12-bit printer isn't really that visible? It's futureproofing more than anything because sooner or later we'll get better output devices and the files will print even better than they did originally. Is that worth it for everyone? Not really- but to dismiss it totally is rather silly.

    All the converters won't produce the same image result, in fact- in my testing, none of the produced the same results. The differences are subtle, but then back in the film days, the differences between some developers and film types were subtle- but some of us still made our choices based upon those subtleties.
     
  7. rjp thread starter macrumors regular

    Joined:
    Feb 25, 2008
    #7
    My point here is that the initial rendering uses the in-camera settings. as you said, it's showing you what the camera would have done if you set it to jpeg. This provides a nice starting point for further adjustments. It is my opinion that my Nikon camera makes some pretty good choices most of the time actually, just tends to clip highlights a bit. Shooting NEF gives me a couple of extra bits at the high end to correct for this.
     
  8. compuwar macrumors 601

    compuwar

    Joined:
    Oct 5, 2006
    Location:
    Northern/Central VA
    #8
    You can dial in a bit of negative exposure compensation if you find the metering to be biased that way constantly.
     
  9. rjp thread starter macrumors regular

    Joined:
    Feb 25, 2008
    #9
    It's not really a matter of consistency, it's just that the camera doesn't always know how important the highlights are in a given shot. I guess it's just a limit of what their matrix metering can do: it makes it's best guess. How can the camera know if the tree out the window is more important than the person on the chair inside the room for example? This is just a modest D40 I'm using, btw. Seems many people bash the matrix metering on this model (and also the D80) saying it's defective. I think it's pretty good actually, just have to be careful and reshoot occasionally. Sometimes I'll dial in -1/3 or -2/3 just as a precaution as you mentioned. This avoids most of the overexposure issues, but causes lack of resolution in the shadows. NEF, of course, solves both problems by providing the extra dynamic range. I've heard some of the better Nikons don't have this metering problem, though this is hard to imagine since their sensors have pretty much the same dynamic range. Must be some pretty smart electronics if so.
     
  10. compuwar macrumors 601

    compuwar

    Joined:
    Oct 5, 2006
    Location:
    Northern/Central VA
    #10
    Basically, Nikon pre-loads an algorithm from database of shots that are similar to most shots taken, and the zones where the subjects normally are in those scenes are given the priority. So, the size of the database and the pictures in the database are mapped into the matrix areas and weighted- the bigger and better the database is, the more likely the camera gets the subject "right." Obviously, if you're shooting differently than the "norm" for that shot, then it'll be wrong.

    The D40's algorithm is from a db is 30,000 shots, they're mapped by brightness, contrast and color. Most of the consumer and "prosumer" Nikon bodies have the same algorithms from the same 30,000 image database. I've seen reports that the pro bodies have a significantly larger database of 300,000 of these image patterns- but I'm not sure if it's just a typo or what.

    Nikon includes distance information for the "3D matrix metering," but I don't know how much it weighs (no pun intended) that information in its choices.

    The D40, D50 and D80 seem to share a 420 pixel metering system, while the D70 and D40x seem to have the same 1005 pixel system of the "big boys." The D3 and perhaps D300 may have a database of almost twice the prior one, and their matrix metering system is different than the previous ones (this seems to be one place where Nikon continues to change from body-to-body.)

    White balance information is also derived from a database as well as the sensors.
     
  11. OreoCookie macrumors 68030

    Joined:
    Apr 14, 2001
    Location:
    Sendai, Japan
    #11
    I think the negative bias is systematic to avoid bleeding highlights. In the reviews of Canon gear, I've read that they do the same thing.
     
  12. compuwar macrumors 601

    compuwar

    Joined:
    Oct 5, 2006
    Location:
    Northern/Central VA
    #12
    Canon seems to do well with highlight detail, where Nikon tends to do well with shadow detail for sensors which effectively have the same exposure latitude. I'm not sure if it's because they bias their exposure in firmware though.

    In digital, you want to come right up against the highlight end of the spectrum to get the most tonal range from your gear, so this is actually the "best" way to do things if you're making cameras, but like with positive film, you can't recover a blown highlight, so it's a tough balancing act.
     
  13. OreoCookie macrumors 68030

    Joined:
    Apr 14, 2001
    Location:
    Sendai, Japan
    #13
    Right. This was automatically compensated for in the labs when you shot on negative film.
     
  14. rjp thread starter macrumors regular

    Joined:
    Feb 25, 2008
    #14
    This is true only if you require highlight resolution in the particular shot you are framing. If the detail you are looking for is in the mid to shadow range you'd be wasting bits trying to capture the various shades of bright white out a tiny window in the background for example. You may want to let the window clip and enjoy 4 times the resolution on the subject of interest.
     
  15. OreoCookie macrumors 68030

    Joined:
    Apr 14, 2001
    Location:
    Sendai, Japan
    #15
    No, he's right. The problem is that brightness `bleeds' to other pixels. This means even if the incident light is not `bright white', if neighboring pixels see light of high intensity, then the currents wander to other pixels in the vicinity.

    That's why at least Nikons and Canons (probably all) camera manufacturers have a negative bias for metering. You knowingly sacrifice detail in the shadows, the mids are unaffected.
     
  16. compuwar macrumors 601

    compuwar

    Joined:
    Oct 5, 2006
    Location:
    Northern/Central VA
    #16
    It's a bit-shift issue, the further to the right you expose, the more levels of tonal range are available to you, and the sensors are linear in terms of values, so you should actually get more shadow detail as well if you expose to the right, since you're moving the exposure up a stop on the range, making more bits available for capture. It does mean you have to do more PP to get things back to the "look" you want in terms of tonality if you want things to be "natural" looking.

    Sure, you may want to blow some highlights, but in terms of the detail you're capturing, the further up to the highlight end of the spectrum you get the information you want to capture (not what you want to blow out,) the more fine-grained your tonal range is going to be. Go over that limit though and the information is gone, you're at all ones.

    http://www.luminous-landscape.com/tutorials/expose-right.shtml
     
  17. ChrisA macrumors G4

    Joined:
    Jan 5, 2006
    Location:
    Redondo Beach, California
    #17
    One could be technically better, but "good enough" is when the difference is not detectable by the eye at normal viewing distance on the intended output media.

    One example: Last night my sister came by having returnd from a trip to Hawaii. She was convinged the iPhone built-in camera was "great" because of the quality of the photos she saw. But she was looking at the pics that were burned to a slide show on a DVD on a TV set. So a cell phone camera is "good enough" if the photos are going to be shown at 480 lines of resolution. All the blurrynes and noise just goes away when they are that heavly downsampled. This is an extream case but all images are always downsampled somewhat. Only in the case of large (over 8x10) prints are they not

    And then many of the differences are subjective. For example the trade off between having fine detail and morie patterns. And the difference between "sharp" and "over sharpened", between "vibrant" and "natural" color. There is no "correct" it's artistic judgement.

    Past versions of Aperture were not as good as Adobe by anyone's judgement but now with the new 2.0 raw engine most of the difference is in the relm of subjective areas where people can disagree. Not only that all of the raw converters are very, adjustable and have many, many sliders that can tweak the image. So all of them can likely do what you want

    You don't have to use only one workflow. You can put 90% of your work through Aperure because it is "good enough" for web and DVD based slide shows and then use some other software for making larger prints.
     

Share This Page