News and Views from Dave Wilson

HDR Tip #1 – HDR From A Single Raw

by on Feb.28, 2011, under Photography

This week, HDR aficionados Rick Sammon and Trey Ratcliff are offering a week of HDR tips so I figured what better time to throw out a few tips of my own. I can’t promise to write one each day but I’ll try to get as close to this as possible. Without further ado, here’s tip number 1.

Caerlaverock Castle

Most of us who do HDR find ourselves in a situation once in a while where we would love to produce an HDR image but only have a single raw image to start with. Common causes for this are that we didn’t bracket when we originally shot the image we want to process now or that we want to take a picture that contains a lot of movement. In these situations, we can, however, still make use of Photomatix to produce an HDR-like image from that single original exposure and the results can often be very appealing. The image above, for example, was generated from a single exposure I shot while on a tripod-less trip back to Scotland a four years ago.

Many people, including several photographer friends whose HDR work I respect enormously, produce HDRs from single raw files using a workflow that goes something like this:

  1. Create 3 virtual copies of the original image in Lightroom.
  2. In the develop module, pull the first image exposure up by 2 stops so that it is brighter.
  3. Again in the develop module, pull the last image’s exposure down by 2 stops so that it is darker.
  4. Export all 3 images to JPEGs (hence creating the bracketed set you need to create an HDR).
  5. Open Photomatix and merge the 3 JPEGs together to give an HDR image which you can then tonemap and process as normal.

This sounds reasonable, doesn’t it? To create an HDR, you need a set of brackets and this gives you that set of images so all is well.

What these folks don’t realise, however, is that they are wasting about 20 minutes per picture and likely ending up with an image that is not as good quality as it should be.

Why not try this instead?

  1. Open Photomatix.
  2. Choose “File/Open…” and point it at the single RAW file you want to work on.
  3. Process as if you just merged a bunch of brackets.

“That can’t be right – it’s far too easy!” you are likely shouting by now. The truth of the matter is that this is likely to give you a better result than the long and complex “generate your own fake brackets” approach described above. The original raw file has at least 2 stops and most likely at least 3 stops more information in it than the JPEGs you create in the awkward workflow. The JPEGs you export there throw away all but 8 stops (8 bits) of the data so by using exposure modification to generate your pseudo-bracket,all you are doing is deciding whether you throw the information away from the highlights or shadows in the different JPEGs. You then open these in Photomatix and have it try to reconstruct the very same data that you were so careful to throw away in the export step.

Photomatix is perfectly capable of extracting all the information from your single raw file so why not have it do the work instead of trying to help (but actually making Photomatix’s job more difficult)?

People who have heard me talk about this before may note that I’m breaking one of my golden rules here which is never to have Photomatix process my raw files. HDRSoft themselves admit that Adobe Camera Raw (the raw file processor inside Photoshop and Lightroom) does a far better job of rendering a raw file than their processor so I always use Lightroom to export my brackets to Photomatix. If you want the best possible quality, you can do the same thing here as long as you remember one very important choice. When you export your single image from Lightroom to send to Photomatix (which, incidentally, you can do via the standard Photomatix plug-in by just highlighting a single image rather than a whole bracketed set), make sure you select “TIFF 16-bit” as the output format. A 16 bit TIFF file allows you to save every bit of dynamic range information from your original raw file and pass it over to Photomatix. If you select JPEG at this point, Photomatix will only receive 8 bits of information and can’t, therefore, pull out any extra dynamic range since you’ve already thrown all that data away.

Next time you are tempted to muck with your exposures in Lightroom and create a fake bracket, give this a try and see what you think. I’m confident you will see that in this case the easiest approach can also be the best!

:, , , , , , ,

14 Comments for this entry

  • Tom Baker

    Dave you are correct. I used to be in camp #1, but Photomatix is so good at it that splitting the raw into 3 different tiffs is a waste of time. In fact i agree the quality isn’t as good. Good advice.

  • Mike Olbinski

    Nice tip Dave…I like it. I’ve noticed if I do the virtual copy strategy, I end up with a bit more noise and ugliness. I like this simple approach.

  • Dave Wilson

    Tom, if you were splitting the raw into 3 different 16-bit TIFFs you were definitely wasting time since a single 16-bit TIFF will hold every bit of dynamic range data that was in the original raw. Remember, your raw file contains no more than 14 bits of information per channel (with the best sensors around these days) whereas a 16-bit TIFF can accomodate pixels with up to 16 bits of data per channel.

  • Mike Connell

    I’ve had mixed results with this. In my experience, it depends on the raw file as to whether Photomatix does a decent job with the conversion. It usually works OK with my Canon 30D, whereas I’m not happy with the results from my 5D or Olympus E-PL1 files. Adobe Camera RAW just does a much better job with demosaicing. What seems to work more consistently for me is to use Lightroom to break my raw into 3 TIFFs. This seems to work better than exporting a single TIFF to Photomatix for whatever reason. I usually just do -1,0,+1 EV. Using -2 and +2 EV may create excessive noise. I use Lightroom’s noise reduction on the -1 EV export just just a bit (this will greatly reduce noise in Photomatix.) I’ll export to a temp directory and then use Photomatix to open the files (don’t use Photomatix NR.) After generating a tonemapped TIFF, I delete the TIFF brackets I created with Lightroom. More time and steps, but I seem to get much cleaner results that way. Like most things in life, your mileage may vary!

  • Dave Wilson

    This is what I don’t understand, Mike. TIFF-16 is a bucket that is more than big enough to contain all the information in your raw file so I can’t understand why you would ever get better results by creating a pseudo-bracket from the raw. Have you done comparisons between the two methods since I would love to see the results of my method vs. yours.

  • Servalpe

    My solution is an hybrid. I do the virtual copies but export then as 16 bit TIFF files and later to PS to generate an HDR file from 3 TIFFs. Finally I tonemapped this at Photomatix. I know that the information is redundant cos RAW file should be around maximum 14 bits. In any case, I always do my HDR files with Photoshop cos it generate less noise, it is better with alignment and I prefer his de-ghosting tool !

  • Dave Wilson

    I’m not with you on this one, Sergio. If a single 16 bit TIFF can contain all the information and since you are starting with a single image meaning that ghosting and alignment won’t be a problem, what do you gain by generating 3 images first?

    My main argument here (which I would love someone to argue against) is that you can’t create more information by creating a pseudo-bracket than you already have in a single file so why do it?

  • Mike Connell

    Dave, I’m with you on the information being there in the TIFF. I’d say the little bit of noise reduction I do prior to tone mapping definitely helps. The resulting tonemapped image is absolutely cleaner than letting Photomatix denoise it or, of course, using no noise reduction at all. Apart from that, maybe there is something about the way Photomatix’s algorithms work off of one raw/tiff vs 3 files. With a single file, would have to do its own extrapolation of under/over exposed data points. Stretching too far will create noise. Feeding Photomatix separate files, it is told ahead of time this is -1, this is 0 EV, etc. All the pieces to the puzzle are in the raw/tiff bucket, but they are assembled by different means. This is pure conjecture.

    Unfortunately, I haven’t preserved any images that I ran directly through Photomatix’s raw engine vs the Lightroom image breakout for an apples to apples comparison. I should probably try it sometime! Here is one of my single raw images from the breakout approach:

    As I mentioned before, I have seen better or worse results depending on the camera. It also depends on the exposure. I’d say if your approach works for a given file, that is by far the easiest way to go. If you see unsatisfactory results, try the harder way. 🙂

  • Sergio Valverde

    I am agree with you that I don’t get more information. I was talking about HDR from several real exposures to get the advantages of PS with alignement, cos as you say, they is not ghosting, but PS can create an HDR file. And what happens with the noise ? Photomatix is worst with this. In this case, as you said if you have only one exposure with the 16 bits file you have all the info. But the question is if Photomatix process better a 16 bit TIFF file or an 32 bits HDR file ? In both cases, the info is redundant as the RAW don’t give all the into to fill the bits, but what happend with the algorithms to create this kind of files and the conversion.

  • Dan Roads

    Hey Dave, I thought Photomatix converted RAW straight to JPEG anyway? :/ Maybe I missed something though.

    • Dave Wilson


      I’ve heard rumours that Photomatix may use JPEG as an intermediate internal format but I find this hard to believe since the minute you convert a single raw to a JPEG, you throw away all but 8 bits of dynamic range. After doing that, it would be impossible to pull any extra detail our of your single exposure so I strongly suspect that they are converting the raw to a higher bit-depth format internally before merging to the 32 bit HDR format.

      Perhaps it’s time to send a note to the HDRSoft support line to get an answer to the question. I’ll update the post when I find out.

  • Pedro Mota


    depending on the type of photography, what you have written may not be entirely accurate.

    If you produce multiple Tiff files from a single raw file, you will change the way the software interprets the HDR image.

    How? You’ll modify the relative importance of each part of the image.

    I did a little test from a single CR2 file. This file is a photograph of a lightning bolt (extreme situation).

    I created 3 HDR files in Photomatix.
    • One directly from the CR2 file.
    • Another from a 16 bit Tiff converted from Adobe Camera Raw (0 ev).
    • And another from five 16-bit Tiff also converted from Adobe Camera Raw (in this case -4, -2, 0, +2, +4 ev).

    Then I tone mapped all three files (not with Photomatix, here I’m a outsider…).

    The result was that I was able to get more information from the 5 Tiff HDR file, that from the other two cases. Particularly from the bright areas of the image, as I intended.
    Although the end result is not the most natural, the truth is that the information was there.

    In the link below, you will find the 3 images tone mapped.

    If you want, I can provide the original CR2 file.

    Please comment.

    Best Regards

  • Dave Wilson


    Your results are pretty convincing even though I don’t understand why the “do it yourself bracket” approach should give a cleaner final result. I really must send a note to HDRSoft and see what they say about this.

    Thanks for posting this information!

1 Trackback or Pingback for this entry

Leave a Reply

CommentLuv badge

Looking for something?

Use the form below to search the site:

Still not finding what you're looking for? Drop a comment on a post or contact us so we can take care of it!