I have tried HDR on five images first processed from RAW using RawTherapee in order to convert first to jpg.
I have recently come across to enfuse linux program that does nicely the job and most important it does not apply any colour profiles like Luminance HDR does to the final image.
There are several options to try with enfuse and the final command line options I’ve tried are the following (based on this url Enfule reference manual
enfuse -v --exposure-weight=0 --saturation-weight=0 --contrast-weight=1 --hard-mask --output=out/test_hardMaskContrastBased.jpg *.jpg
but I loose the highlights generally. Is there a better way via the command line to achieve better highlight results generally with enfuse?
Here is my resulting image:
And in this zipped file are the 5 jpgs I used for producing it:
NIKON D4_09-Apr-23-141.zip (19.0 MB)
Not knowing enfuse from heart , but I think setting exposure weight to 0 is essentially saying you don’t care about how good exposed a pixel is, so yeah you will loose highlights i guess.
Using the defaults but loading it into RT, Darktable or gimp to tweak the final colours and contrast will get you a nicer result quicker i guess.
Hard-mask helped me sometimes but gives issues on noisy images (and i always tried enfuse on linear tiffs, no clue how it works on final jpgs like you seem to be doing ).
Enfuse basically looks at your input pictures , and decides pixel by pixel how much of each picture to use .
Exposure weight means how far the pixel is from ‘ideal exposure’, saturation weight means how colourful the pixel is.
If you dont want/need colour in your highlights , set the saturation weight lower like you are doing .
But if you dont want to have over exposed pixels , give some weight to exposure , so that the over exposed pixels contribute less to the final image .
If that means the contrast isn’t to your liking , open it in another editor and apply some (local) contrast / clarity to the highlights.
I can offer no helpful advice about your issue with enfuse. However, one piece of general editing advice is that you should avoid working with jpg images and should in fact be using tiff images after converting your RAW files. JPGs are compressed files that do not remember the value of individual pixels where as uncompressed tiff files will remember the value of an individual pixel. I usually edit my RAW files and export as 16 bit tiff files for further editing in programs like GIMP or for archival storage.
Hello @Terry and @jorismak
Thank you for the advice. I tried both of your suggestions. First I turn those RAWs to .tifs and converted via enfuse my image with using weight on exposure to 0.8 into again .tif. I am not going to upload this image as it is almost similar to that I have already uploaded (a bit more lightened in all aspects) as it came to bbe 125Mb in size.
Well those 16bits per channel did not made any difference (specially on a laptop’s monitor) and certainly it was not what I wanted for the highlights which are blown out as in the original jpg.
The underexposed jpg I include in the zipped file seems to hold all the info in the highlights what I do not see in the final blended/rendered in HDR image.
Yet I have not tried with that tiff to pass it again to RawTherapee and process it further on those highlights to see with what result I can come up with…
They actually make a huge difference in the ability to edit the image further. An 8 bit image only has 256 shades of red green and blue pixels value while a 16 bit has potentially many thousands of shades of rgb. BTW, I personally have not had great success using enfuse for HDR merging myself. I would be interested in hearing from those who have mastered the process with enfuse. Currently I use Adobe’s Lightroom for HDR merges and Lightroom exports a 32 bit image for further editing. I would love to find a FOOS solution for HDR merge, especially one that could align hand held images.
Here is a screen shot from lightroom. Even it is struggling with the very bright highlights. This may be because of the intermediate jpg files. I usually would merge the raw files in Lightroom. I will try enfuse next. I apologize for using a commercial program on a FOOS forum, but I am just trying to use Lightroom as a bench mark for your images.
Here is the Lightroom generated DNG edited in Darktable. It has retained more highlight detail in my opinion. DT was definitely a more powerful edit program for the DNG than Lightroom.
I generally use HDR merge for this: HDRMerge
@Terry Yes the highlights are quite what I was looking for on your images. And I did try to bring those back from the .tif image using RawTherapee without success. What it was burned it did not come back. As I understand many have tried enfuse before trying to blend HDR images with partial success.
@paperdigits Mica I have used HDRMerge but it did non succeed in other cases on the same session I did. Speaking again about highlights for this program it usually whenever it cannot compensate those it puts a purple or so colour that ruins the merged image. I’ve tried several tweaks as I remember on HDRMerge but I have never managed to eliminate those bizarre colours at the end result. Plus it took ages processing each and only one image for that reason.
But did you try without any parameters and just use the default ? And maybe specify just the ‘normal’ exposed image and the underexposed image. If that works , add the overexposed versions one by one to see if they actually do something.
I decided to try using GIMP with layer masks and brushes to put this image together as I feel HDR may not be needed with a shot taken like this using a tripod for alignment.
@jorismak Yes that might be an option. Specially using only a narrower range than all of the 5 photos I took in bracketing mode. The --hard-mask is something that I might always use since it gives better results that the default --soft-mask one.
@Terry A very nice result with GIMP indeed! I do use that technique myself when there is no good results on merging HDR photos otherwise. As I remember correctly I used a couple of videos to extract info from YouTube and kept notes blending those techniques. Very nice result the one you came up with Terry!
I just iterate the point that enfuse is more about merging exposures and not getting a perfect ‘hdr’ image out of it in one go . Saturation , contrast, and most importantly local contrast is something to do after. Also the reason why most use something like tiff to keep an uncompressed input / output chain until the very last, final image.
Enfuse (to me at least ) is preserving the ‘signal’. Processing is after.
And I never seen enfuse produce clipped highlights of there is an input image without clipping.
Maybe i need to try myself , because it has been a while .
Tried a bit with your jpgs. And to be fair, it’s not as straight forward as I remember from my shots some years ago.
Just using the defaults indeed clips the highlights at the lamp. Adding --saturation-weight=1 (or some other value, in any way add saturation to the mix) helps in telling Enfuse that you rather not keep pixels that have no colour in there.
I ended with
--exposure-weight=1.0 --saturation-weight=0.5. I tried
--hard-mask, but it takes ages and in this case I actually like the result with the default soft mask more.
Furthermore, I also added
--blend-colorspace=ciecam, and I like what it does, but it is very, very subtle.
edit: I tried with just ‘*4996.jpg’ and ‘*4995.jpg’. After that I tried using all the JPGs, and the resulting image is brighter, but also the highlights with the lamp are more clipped (or at least show bright you don’t see all the details).
The result in Rawtherapee with a bit of log encoding to tweak it seems to work fine for me, but I’m not the one working on this shot :).
I do wonder, though. If this is indeed shot with a D4, and those jpegs from the first shot are edited with Rawtherapee… Why don’t just use the RAW file from the ‘4996’ shot? I see -2 EV with ISO 200 on a D4. I think you can easily boost that shot by 2 or 3 EV and don’t have to fiddle around with merging shots. If I’m editing the tif output from Enfuse in Rawtherapee’s log-encoding, you might as well load the RAW in there to do the same :).
As a quick test, I loaded all the JPGs into Affinity Photo 2 to do an HDR merge, and I do get a nicer result with less tinkering. So Enfuse (at least with those processed JPGs as input) is a bit finicky to work with. Or someone must master the tool more :).
Latest release is 8.5 years old? Wow. There are commits up to last January, but 2015 is the newest released version (at least, in installer form for Windows). Then again I can get lost in Github so I may be overlooking a newer download.
I guess if it works, I guess it works.
enfuse *.jpg–>RT–>Gimp for resizing
(It’s not an HDR: https://patdavid.net/2013/07/automatic-exposure-blending-with-enfuse/
and this with only 4995 and 4996 and resizing
There are newer builds for Linux, sorry.
For the 7 (?) years prior to my current computer, I ran Linux at home. Naturally…
I totally agree with this statement for any HDR merging software. Because I own Lightroom and it does a nice job putting hand held bracketed exposures together I use that for merging and then Darktable to process the 32 bit Tiff file. However, my perpetual LR license won’t include updates for new cameras and file formats so eventually that option will be less useful to me. However, doing raw to DNG conversions may resolve the problem of using Lightroom going forward. I would like a FOSS solution for Windows and I just couldn’t work out how to install enfuse for Windows.
Absolutely not FOSS , but you could buy Affinity Photo one time And use it for merging . You can then process it, but also just Dave it unprocessed as 32bit floating point tiff, or an EXR file or something. Open it in rawtherapee / Darktable and process further .
Although hdrmerge is then basically doing the same i guess.