Expessing myself without a camera

D800 Long Exposure in need of Therapee

About a month ago an issue was raised and discussed in Emmanuel Coupe’s blog about the noise that appears in long exposure photographs taken with the D800. We aren’t talking about the usual red and blue pixels which the camera handles remarkably well, we are talking about those white dots that litter the entire image and ARE visible on print which is what matters. A few samples were sent to me and they all showed the same behavior so it can’t just be a bad batch. Users of the Nikon D7000 which has the same pixel density sensor as the D800 also told me that they were experiencing the same issues and were forced to always use Long Exposure Noise Reduction (LENR) from the camera menu though this I can’t confirm since I have never taken a LE photograph with a D7000.

I am not going to be definitive about it since there is no firmware update at the time that this post is written, but I am guessing that Nikon update, which is bound to appear soon due to the locking problems reported, isn’t going to address this issue considering that LENR is an adequate fix and it is there for a reason. Many users are OK with this, spending 5 or more minutes doing a long exposure and waiting 5 more for the camera to work on the noise. I am not. I don’t like it one bit, it limits my shooting options a lot especially in times of day when the light is changing fast.

I have two raw developing software available, Nikon Capture NX2 and Adobe Camera Raw 6.x, out of which I use Nikon’s one, you just can’t beat the excellent rendering of the NEF file from Capture NX2. These two could not deal with this issue in post process with the files I tested without any loss of detail using their noise reduction feature and that was a letdown. If it can be done in camera why cannot be done in post?

There was a program though that was able to deal with it astonishingly well, a free program mind you, which put its two competitors to shame: RawTherapee. What you see bellow are two details of a 4min exposure cropped at 100% showing a smooth tonal area which makes the white noise stand out much more in order to be visible by most. The first image is the Raw data as they appear in all three programs including all the white noise with some hot pixels to add in the mix. After using the “hot/dead pixel filter” located in the “RAW” tab you can clearly see that almost all problems disappear and I couldn’t detect any quality loss which is of course good news.

ISO 100, F/11@240sec

Raw Tab

Hot/dead pixel filter

End result after applying the filter

There is a drawback in this whole process though. One has to learn a whole new program in order to develop a raw file. Unfortunately doing your raw editing in your program of choice and then inserting a tiff into RawTherapee won’t work for obvious reasons and RawTherapee will ignore any third party metadata which is kind of a disappointment. The good news is that RawTherapee is an excellent, even though not one of the most intuitive programs out there. It requires a bit of a learning curve but nothing that can’t be done. Plus if an open source program managed to do it I can’t see any reason why the “big guns” can’t implement this feature in their own software. Or have they and I just couldn’t find it? Let me know. Personally I am still bumped with having to learn something new and forced by Nikon to abandon their very own wonderful program.

Best case scenario is for Nikon to fix this. Until then it is either LENR or RawTherapee.

Konstantinos

P.S. I am still in the process of learning the basics of RawTherapee if you have any more experience with it or any helpful tips and tricks let me know.

10 Comments

  • Ruben Fernandez on Nov 24, 2012 Reply

    Hi all, an idea that maybe works. Develop a copy in raw therapee whithout noise reduction, another with it. After in ps do the diference for obtaining a mask of the white points that it will be utilized in the real image edited in nx2 , acr, etc… after. What can i do with this mask? By example, open in ps a copy of the image developed in the raw editor of your choice, make a copy in another layer, shift the layer at bottom a few pixels in one direction, just for misaling the white dots in both layers. Then, the quid, use the mask of white dots in the top layer for blind the areas where the white dots are and allowing the clean areas of the botton layer become visible. This effectively clone the dots with pixels of the surrounding areas. Maybe is necessary blur a little the mask for quality clone. Only is an idea…

  • Scott Reither on Jul 02, 2012 Reply

    Any further developments with this? I just got my D800E and am having BIG long exposure issues. Check out my post and pix here: http://scottreither.com/blogwp/2012/07/01/nikon-d800-e-long-exposure-issues-problems/ Like you, LENR ON is not an option for the same reasons – though I’m not sure how much I’m liking the idea of adding yet another step in the post-processing. Are you still using this process with Raw Therepee? Do you enter the NEF file there 1st, then into Nikons software?

    • Scott Reither on Jul 02, 2012 Reply

      I updated my post and it is now here:
      http://scottreither.com/blogwp/2012/07/01/nikon-d800-e-long-exposure-issues-problems-2/
      I have the white-spot problem something terrible. I guess I’m going to have to follow your lead and try out this software – it looks like it’s working for you. I will link back to this page on my post. I’d love to hear an update how things have progressed for you…

      • Konstantinos Vasilakis on Jul 02, 2012 Reply

        There are no further developments to the problem. I doubt that Nikon is bothered with this issue since it can be solved with LENR. So for the moment things stand as they are. It is either LENR or RT. Nothing new to report. Unfortunately the DxO software update that included support for the D800 doesn’t have the RT feature so RT is still the only program that can solve this problem.

        When I use RT I do all the processing in RT. I do not go through Capture NX2, and then export the tiff from RT to PS. There is no point exporting a tiff to NX2.

        Thanks for the link back, I am glad this post was a bit of help.

        Regards

  • Simon on Jun 20, 2012 Reply

    Hi Konstantinos,
    Thanks for sharing this with us. For me (D800 on order) this is a massive deal breaker and I’m even thinking about cancelling my order. I will never use in-camera LE NR because I’ve had never have to use it with other Nikon cameras, period.

    RawTherapee sounds like a nice solution and I really do like this open source program but I am not keen on having two separate workflows. Since I use gradients in ACR a lot, i’d still have to use ACR as a main Raw programm (unless RawTherapee also supports gradients now?!)
    Can anyone tell me, at which exposure lengths the spots start to appear?

    • Konstantinos Vasilakis on Jun 20, 2012 Reply

      Hey Simon

      When I had my hands on the D800 I had no idea about this problem so I didn’t do any incremental testing to see when it appears. I have seen it on shots of 2 to 4 minutes which means there is definitely going to be there after four minutes but I cannot be certain of whether it will or how much there will be in exposures under two minutes. If I remember correctly a sample of 30 seconds was sent to me and had similar behavior but I am writing this with a big maybe cause I am away at the moment and I don’t have those files with me to check.

      I am not sure that RT has gradients, the most appropriate people to ask are the guys at their forum. i am sure they can guide you if they can. There are workarounds to the gradients that one can use in the ACR, one being work the file multiple times in RT. create two? maybe more tiff files and then do the masking in photoshop. It is more time consuming and maybe will involve getting you a bit out of your comfort zone but the camera is worth that extra hustle.

      Regards

      • Simon on Jun 21, 2012 Reply

        Thanks for your reply, Konstantinos.
        I found a 30s example on ken rockwell’s D800 review page:
        http://www.kenrockwell.com/nikon/d800/sample-images/800_0113.JPG

        many white spots on this one… (and I’m not talking about the stars :-)

        Since I do many 30sec+ exposures I think might wait if they can figure it out in the firmware before I buy one.
        Regards,
        Simon

  • DrSlony on Jun 06, 2012 Reply

    Hi!

    Glad you found our program useful and happy to hear it works well for you. I wouldn’t find constant LENR delays acceptable either. I can agree to shooting one dark frame for the whole shoot, but not one for every photo.

    I’m a bit puzzled by one statement:
    “Unfortunately doing your raw editing in your program of choice and then inserting a tiff into RawTherapee won’t work for obvious reasons and RawTherapee will ignore any third party metadata which is kind of a disappointment.”
    It will work, though you lose the benefit of having RT work at the raw level since it will have to work with your already demosaiced TIFF… but then again you could reverse the tables and first use RT to demosaic your raws, and then feed the TIFFs into NX2 for whatever it is NX2 does that RT lacks. Cumbersome, yes, but it will work.

    As for the metadata, RT can preserve all of it (Preferences > Image Processing > Metadata > Copy IPTC/XMP unchanged to output file). It just won’t use NX2′s sidecar file to develop your raw, for obvious reasons.

    Regards!

    • Konstantinos Vasilakis on Jun 06, 2012 Reply

      What I meant by the quoted statement is that when you import a NEF file into RawTherapee the hot/dead pixel function works like a charm. But when you import the same file as a TIFF it will not do anything. The white dots remain as is. I have tried it on different files the results where the same. Maybe I am missing something? Perhaps.

      Yes one can go backwards and apply the filter then export to tiff and work on the tiff with the program of choice but you loose the benefits of working on RAW. That goes for someone who doesn’t want to use RawTherapee.

      For me like I said it is either RawTherapee or LENR. The more I learn new things about your program the more I like it.

      As for metadata yes RT can preserve its own but can’t read the xmp say from ACR. From what I hear there is a working project on being able to do that, and if true it will be excellent.

      Regards

Leave Reply


nine × 3 =