New HDR algorithm in Darktable anytime soon?

What do you mean by raw-based? I´m using Hugin for panos only up to this point. For import I usually export the images from darktable as 16bit Tiff not dng or pef. Does Hugin allow raw-images at all?

I suspect that’s one of a few known bugs that are being worked on. Extreme dynamic scene loses most details · Issue #113 · jcelaya/hdrmerge · GitHub

Try building from Commits · Entropy512/hdrmerge · GitHub - that has significant improvements for many use cases, but there is still work to be done. I’ve been busy with non-photography things/being a horrible procrastinator for a while, you reminded me of yet another thing I’ve been slacking on…

LuminanceHDR is another option for merging.

Get a remote release (wired or wireless works, I’ve got a wired one for Sony cameras made by JJC) - I have not had any issues with movement between exposures when using it. Even 60+ image stacks work well. (Sometimes I bracket, but for waterfalls I will usually do an average stack instead to synthesize what is effectively a long ultra-low-ISO exposure.)

The images are less about needing HDR , but rather testing various software’s ability to align the hand held images and deal with the pesky tourists. This is one time that Lightroom wins out. It is the only function I really use Lightroom for now. Microsoft Image Composite Editor is my go to panorama stitching program, but Lightroom and many other programs including Hugin also do an excellent job for panorama. I am just glad I have a perpetual license for Lightroom, but Adobe will work hard to make that less useful in the future. Lightroom V6 will not be downloadable from 2024 and also new cameras and lens are not compatible. Compatibility issue can be circumvented using Adobes DNG convertor.

I would have to agree with this sentiment. It may be nice to have full functionality such as HDR, but is the time cost justifiable when there are good programs already?

I’d agree as long as “good” means 3 in the range from 1 to 5. The way it does here in schools. :slight_smile:

The integration of HDRMerge didn’t work for me on Windows. Yet, it’s the only thing that works like I put the source images into it and it spits out the merged one. All the other stuff is either way too complicated, or I could never make it work. And I had tried every free alternative I could find and run on Windows.

But why do you need the integration with an external program?

The more since such integration can be fragile, in that any change in the external program can break the integration (e.g. all the export tools for web services like flickr). I prefer by far to just export the images “by hand” and import them in the external tool.

Additional advantage: I know I have access to all the functionality of that external tool. OK, it is a bit slower.

Because its convenient.

I didn’t come up with the integration argument. :slight_smile:

Also, let’s drop DT entirely, because GIMP already does most of it all pretty good. :smiling_imp:

1 Like

But is doesn’t. Sadly, non-destructive editing is not even here yet (I know the team is closing the gap, but still).

I have been and continue to be a huge fan and promoter of GIMP, but darktable has taken away 99.9% of my need for GIMP. BTW, there are numerous free or paid for panorama stitch or HDR merge programs out there. They don’t need to incorporated into DT to be useful to our workflow. If you own a Canon they provide software that includes Canon photostitch, which is a pretty good option at no addition charge.

This thread has convinced me that while HDR in darktable would be nice, it is just too big an ask for the volunteer developers when good options already exist.


I have probably underestimated the time and work needed to redo the module and add alignment as well as exposure fusion. It appears to me, that those functions were well worth the effort, since most landscape photographers use it a lot. But most people here seem not to. Up to this point I haven’t even got as far to make hugin work for hdr. I love the pano function, but for hdr I fail every time I try. I will try the digikam approach. Maybe this works better for me. I really enjoy working with DT otherwise, but however I’m surprised, that this particular module hasn’t been touched since so many years and most users here seem to deem it expandable all together.

I would add that this statement is like saying lets replace a truck with a motorbike. It is not that one is necessarily a better mode of transport, but rather they are different forms of transport designed for different jobs. Gimp and Photoshops main strength is the ability to use layers and layer masks and primarily was developed in an era when JPG images were the most common output from cameras. Darktable’s main strength is in RAW file editing and because of its great masking abilities in many cases negates the need to then go onto to further processing with gimp to use layers and masks for localised adjustments etc. BTW, it is darktable’s masking ability which has won me over from Lightroom and the many excellent free RAW converters such as RawTherapee. This single feature is the real deal breaker for me. DT developers have excelled with this feature in my opinion.


I think he was being sarcastic…


“let’s drop DT” was meant here as “let’s drop it from the discussion”, at least that is how I read it :).

And that basically agrees with the “do one thing good” ethos :).

But I understand the pain from the OP. If HDRMerge doesn’t work for you , you can try other tools but those often work on bitmap data, not raw data.

If you now what you’re doing , you can export flat bitmaps , do the merge and bring it back into Darktable or rawtherapee and continue working on it. But that doesn’t make it easy to do.

The Lua plugin tries to automate this, so kudos for that (keep the heavy work in other tools, but try to enhance and ease the workflow) .

But the example bracket given here, IS a tricky one. There is a lot of movement and it’s handheld and shifts a bit.

I tried with Affinity Photo and it chokes on the PEF files if you enable the ghosting reduction. If I feed it flat bitmaps it seems to work fine , but the result is not something I would be happy with. But maybe that’s just the source files in this case.

So the OP searches something besides HDRmerge , because this shot doesn’t really work in HDRmerge. But it’s a tricky bracket that I wouldn’t expect to work blindly in every tool, free or commercial.

And for me , I can’t remember the last time i thought about HDR bracketing. I guess this is why interest isn’t at an all time high. Lots of cameras need way less bracketing (or even better: people finally learn that they don’t need to bracket as much).
If outside and at low iso, I happily apply -2ev or even more to my m4/3 camera. And when indoor I dont encounter extreme DR that much (but I’m not an interior photog :)).

I can’t say I ever had a moment where my Sony a7m2 took a shot where I thought 'wow, glad I had my Sony with me '. To be honest , it’s an oldy in fullframe land and wasn’t great at DR / shadow noise even when it was released. But I don’t notice that m4/3 has less DR for instance. It for sure doesn’t cause me to want to bracket suddenly on m4/3. So I don’t think you should on APS-C.

I said I understand it more with older Canon DSLRs. To be honest , I don’t know where a Pentax sits in that regard. Very little experience and reading about how their cameras / sensors perform.


I agree that modern sensors reduce the attraction of HDR and I very rarely use it. The new ISO invariant sensors from manufacturers like Fuji probably make it less attractive. One of the real improvements with late model cameras is the dynamic range. The Pentax samples I posted here are not because this shot really needed HDR, but the images really test out the software’s capabilities. And for once Adobe wins on this one with Lightroom. But then lightroom spits out a DNG which darktable loves working with. So that is my HDR workflow if needed.

Another HDR solution that I use is to process the bracketed RAW files in darktable, export 16 bit tiff files and stack as layers in GIMP and use masks for blending the exposures. This works really well for landscapes where you want to blend two exposures using a gradient mask. The only caveat is that GIMP lacks the alignment options found in Photoshop so you must have used a tripod to keep alignment.

Yes, it would be nice if all this could be done in darktable, but I am not going to ask the developers to devote 1000 hours to make this possible when good solutions already exist.

Said HDRMerge does just that. Can’t handle misalignment well though precisely because of that.

Lightroom spits out a 32 bit floating DNG file from HDR merge. Maybe not raw but certainly a very manipulative file that can benefit from darktables non-destructive editing powers.

It does.

Nope. And what’s a “bitmap” representation? You think raws are “vector representation”?

Well, the HDRmerge merge manual tells a different thing, Just go there, it clearly says “As a result, you obtain a new raw image” or later "Given the nature of the color organization in a raw image (only a Bayer matrix is supported by HDRMerge) rotations cannot be compensated and translations must be performed in steps of 2 pixels. "

The output is a raw file with 1-piece-of-data per pixel which requires demosaicing. That’s a raw file for me. Also - at least for me - that’s a big plus for the dng specs, we can do such things safely.

About “fusing raw data”, doesn’t rt’s pixelshift code work on raw data? (That’s at least how i remember it when @heckflosse told me about it …) I know that fusing data in raw space is much more difficult than on demosaiced rgb data but physically impossible?