HDR vs single raw image

[Digital Nigel]

Here's my edit; I hope you don't mind?

I used the high exposure you posted to provide the ceiling detail I thought was missing from the adjusted exposure:


Ceiling manually replaced with the brightest exposure version. Less green shift, much more detail.

I think that makes the ceiling too bright. Even though it's going through an HDR process, we don't want completely uniform, unrealistic lighting; the dimly lit areas should remain in relative shadow.

 

[Digital Nigel]

The HDR looks better in some key ways here, but there's no reason why the single RAW shouldn't look as good or better with different processing - it looks like the necessary data is all there.

Yes, I probably could make the single shot look more like the HDR stack. In fact, the single shot image is probably more accurate, but I like the look the Affinity 'Dramatic' preset achieved with the tone-mapped HDR stack.

If all your shots weren't all at ISO 6400, and all share the same SS and aperture, you didn't "exposure bracket" but, more likely, ISO bracketed - which doesn't really offer any advantage whatsoever over a single (non-clipped) RAW. Proper exposure bracketing for a low light scene like this would typically be on a tripod at base ISO with varying exposure for each shot. That doesn't appear to be the case here.

How would you know that from a single shot?

I shot handheld, but was exposure bracketing, as I said. The ISOs varied from 400 to 25600. These were the two shots at the extreme ends of the five-shot range:


Lowest exposure


Highest exposure

These are all shot with exactly the same exposure - the same SS, the same aperture.

No, as you can clearly see, that's not the case. The shutter speeds varied a lot.

Well, your originally posted shots all indicated 1/125”. In any case, they still only varied 2 stops (1/30" - 1/125") - which is probably what happens when the bracketing function hits a SS and/or max ISO wall. So you're right, not only ISO bracketed, but mostly. The single RAW of greater exposure (1/30”) would likely be better alone than combined with two noisier lesser exposures (that don’t bring any extra DR to the table).

The darkest one had way too little ISO brightening applied, the brightest one had way too much.

The point is that highlights are used from the darkest shot, and shadows from the brightest.

Yes, but only over a 2 stop exposure variation here (and the less exposed ones won’t improve the DR at all).

So long as a single RAW is shot at an ISO that doesn't clip any important highlights, you will have all the same detail to work with that is contained in all these variably brightened alternates. Ideally, the optimal RAW would have enough ISO applied in camera to bring the highlights just below clipping, but if even if you're down 2 or 3 stops it won't make much difference with most modern invariant sensors.

A single RAW at 1/15" and around ISO 800 probably would have probably yielded the best results here.

No, that wouldn't be optimum with a handheld shot. It would also be very under-exposed.

Why not? The 10-18 is stabilized, no?

Yes, both the lens and the body are stabilised, but I still prefer not to drop below 1/30.

1/15" at 10mm should have been easy and would get you twice the exposure of you're brightest shot (1/30"). At 10mm with stabilization you probably could have gone with a much slower SS and increased the exposure even more.

I wouldn't have considered going below 1/15 handheld, and that's only one stop more than I used. And I don't think that was necessary.

But this takes us back to a version of my original question: does a very high dynamic range scene like this need HDR processing of multiple shots, or does a single raw file have enough dynamic range to enable adequate HDR processing?

This is a new version of my single shot processing of the middle image (the one I'd have shot if only shooting a single image, rather than a bracketed burst). I've deliberately left the dome darker, as this is more realistic:


Processed from single raw image

And this is a revised merged HDR stack, with different settings to before:


HDR merged stack, with settings changed to reduce the tone mapping effect, to produce a more realistic image

A reminder: this is the OOC JPEG:


Note just how dark the dome was in reality

I think my conclusion is that you can play around with both methods to get a similar result, but the single image is slightly bigger (ie, it didn't need the cropping required after aligning slightly misaligned stacked images), and it's much less effort.

Incidentally, this is the domed church in question, as seen from our ship:


Candlemas Holy Orthodox Metropolitan Cathedral, Santorini, Greece

 

[Nick5]

Well done.
When were you in Fira?
We were there in 2019 and have the interior dome as my screen savor on my iPad.
I bracketed three images 0, -2, +2 with my Canon 16-35 f/4 L IS Hand Held. Having IS is Key as my 5D Mark III’s DSLR’s don’t include IBIS.

The “Yia Yia”at the door took one look at me and gave me a few finer points of the Cathedral as she knew I was Greek. Yia Yia’s know.

 

[Digital Nigel]

Well done.
When were you in Fira?

As you can see from the photo, 23 November.

We were there in 2019 and have the interior dome as my screen saver on my iPad.
I bracketed three images 0, -2, +2 with my Canon 16-35 f/4 L IS Hand Held. Having IS is Key as my 5D Mark III’s DSLR’s don’t include IBIS.

Yes, my A6500 has IBIS, and so do most of my lenses.

The “Yia Yia”at the door took one look at me and gave me a few finer points of the Cathedral as she knew I was Greek. Yia Yia’s know.

Strangely, there was no-one at the door or visible inside. I had the church to myself. There was a sign saying that non-flash photography was allowed outside services, which I respected.






I took this picture from just in front of the cathedral entrance,

 

[Glen Barrington]

Do you need HDR for natural looking photos?

I was still fairly new to HDR so I might be able to get closer results in a second round of testing.

But ever since, I have operated on the assumption that a 3-5 DR exposure spread won't hurt if you have the time, but won't automatically be better than a single good exposure.

 

[Digital Nigel]

Do you need HDR for natural looking photos?

I was still fairly new to HDR so I might be able to get closer results in a second round of testing.

But ever since, I have operated on the assumption that a 3-5 DR exposure spread won't hurt if you have the time, but won't automatically be better than a single good exposure.

Agreed.

In fact, the danger with HDR is that you overdo it, and end up with a low contrast, flat image. So you need to be careful to maintain reasonable contrasts between the brightly lit and shaded areas, rather than making them look equally bright. You want to lift the shadows moderately, not eliminate them altogether.

 

[CAcreeks]

And this is a revised merged HDR stack, with different settings to before:


HDR merged stack, with settings changed to reduce tone mapping, a more realistic image

This is vastly improved. Unlike your first HDR, I do not see halos around the chandelier lights. I thought the halos might be from camera shake, but decided not, because no other movement flaws were visible. Ranking:

1. the above
2. your first HDR attempt with chandelier problems
3. your one-shot Raw development
4. x
5. x
6. x
7. x
8. x
9. out-of-camera JPEG

 

[CAcreeks]

Here's my edit; I hope you don't mind?

I used the high exposure you posted to provide the ceiling detail I thought was missing from the adjusted exposure:


Ceiling manually replaced with the brightest exposure version. Less green shift, much more detail.

Why no EXIF?

In the absence of EXIF, could you please tell us how you did it?

 

[Digital Nigel]

And this is a revised merged HDR stack, with different settings to before:


HDR merged stack, with settings changed to reduce tone mapping, a more realistic image

This is vastly improved. Unlike your first HDR, I do not see halos around the chandelier lights. I thought the halos might be from camera shake, but decided not, because no other movement flaws were visible. Ranking:

1. the above
2. your first HDR attempt with chandelier problems
3. your one-shot Raw development
4. x
5. x
6. x
7. x
8. x
9. out-of-camera JPEG

The dilemma I have is that the HDR version looks good because it's illuminated the dark areas of the unlit cathedral to be almost as bright as the sunlit area opposite the ooen front door. You get bright colours and clear details. As such, it's pretty, but is a rather unnatural, artificial image. The single frame image is more realistic, if less attractive. The shadows have been lifted a lot, but still look shaded, not sunlit.

 

[All Bushs Fault]

Do you need HDR for natural looking photos?

I was still fairly new to HDR so I might be able to get closer results in a second round of testing.

But ever since, I have operated on the assumption that a 3-5 DR exposure spread won't hurt if you have the time, but won't automatically be better than a single good exposure.

Agreed.

In fact, the danger with HDR is that you overdo it, and end up with a low contrast, flat image. So you need to be careful to maintain reasonable contrasts between the brightly lit and shaded areas, rather than making them look equally bright. You want to lift the shadows moderately, not eliminate them altogether.

In your OP, you said the church was dark. Neither processed image depicts a dark church.

 

[Digital Nigel]

Do you need HDR for natural looking photos?

I was still fairly new to HDR so I might be able to get closer results in a second round of testing.

But ever since, I have operated on the assumption that a 3-5 DR exposure spread won't hurt if you have the time, but won't automatically be better than a single good exposure.

Agreed.

In fact, the danger with HDR is that you overdo it, and end up with a low contrast, flat image. So you need to be careful to maintain reasonable contrasts between the brightly lit and shaded areas, rather than making them look equally bright. You want to lift the shadows moderately, not eliminate them altogether.

In your OP, you said the church was dark. Neither processed image depicts a dark church.

The single frame image does show shadows, obviously lifted quite a bit.




Single frame, processed from raw, with shadows lifted, but not eliminated.

 

[Michael Fritzen]

The dilemma I have is that the HDR version looks good because it's illuminated the dark areas of the unlit cathedral to be almost as bright as the sunlit area opposite the ooen front door. You get bright colours and clear details. As such, it's pretty, but is a rather unnatural, artificial image. The single frame image is more realistic, if less attractive. The shadows have been lifted a lot, but still look shaded, not sunlit.

Hi,

interesting excercise you made here.

I think that the difficulties between what we "see" on location and taking a photograph of the scene - and hence part of the dilemma for processing the data, be it from a one-shot-RAW or a several shots HDR composition - arrises from human visual perception, in such a case especially in darker ambients when colour perception drops and turns into an almost grey scale perception. The camera's sensor doesn't suffer from this. But making use of the colour data available in the photo's data diverges from perception "in loco" and we tend to consider the look unnatural - which someone who hasn't been there would not.

So at some point it boils down to the decision (of the photographer) if he/she wants to show off (as an impressionist) the "true" lighting situation with (possible/likely) difficulties to discern details, colours and the like, or to capture and show off as much as possible of the details , the colours and the amazing handcraft work of the constructors.