Ugly Hedgehog - Photography Forum
Home Active Topics Newest Pictures Search Login Register
Main Photography Discussion
Low dynamic range technique for HDR
Mar 23, 2020 21:03:25   #
bleirer
 
I've been rewatching Marc Levoy's lectures on digital photography, and in lecture 2 he talks about a method for scenes with a high dynamic range. Instead of the usual exposure bracketing, one would expose many shots of the same scene at the exact same exposure such that no highlights are blown, then average them in Photoshop and then brings up the shadows.

The idea is that the random part of noise is reduced by a factor following the square root of the number of images, so 16 shots averaged would have 1/4 the noise. So the less noisy shadows could then be brightened in photoshop and being much less noisy have more detail to recover.

Has anyone tried this sort of thing? Does it seem like a sound idea?

Reply
Mar 23, 2020 21:37:45   #
Gene51 Loc: Yonkers, NY, now in LSD (LowerSlowerDelaware)
 
bleirer wrote:
I've been rewatching Marc Levoy's lectures on digital photography, and in lecture 2 he talks about a method for scenes with a high dynamic range. Instead of the usual exposure bracketing, one would expose many shots of the same scene at the exact same exposure such that no highlights are blown, then average them in Photoshop and then brings up the shadows.

The idea is that the random part of noise is reduced by a factor following the square root of the number of images, so 16 shots averaged would have 1/4 the noise. So the less noisy shadows could then be brightened in photoshop and being much less noisy have more detail to recover.

Has anyone tried this sort of thing? Does it seem like a sound idea?
I've been rewatching Marc Levoy's lectures on dig... (show quote)


Yes. It is commonly used for night sky photography.

Reply
Mar 23, 2020 21:57:29   #
bleirer
 
Gene51 wrote:
Yes. It is commonly used for night sky photography.


So it's as simple as it sounds?

Reply
 
 
Mar 23, 2020 22:01:13   #
TriX Loc: Raleigh, NC
 
Yep, averaging is commonly used to reduce noise with both imagery and electronic signals. The concept is that noise is random or uncorrelated (while the desired image or signal is correlated), so the noise averages out. You gain less and less improvement as you increase the number of averages, so there is a point of diminishing returns.

Reply
Mar 23, 2020 22:07:10   #
JD750 Loc: SoCal
 
bleirer wrote:
I've been rewatching Marc Levoy's lectures on digital photography, and in lecture 2 he talks about a method for scenes with a high dynamic range. Instead of the usual exposure bracketing, one would expose many shots of the same scene at the exact same exposure such that no highlights are blown, then average them in Photoshop and then brings up the shadows.

The idea is that the random part of noise is reduced by a factor following the square root of the number of images, so 16 shots averaged would have 1/4 the noise. So the less noisy shadows could then be brightened in photoshop and being much less noisy have more detail to recover.

Has anyone tried this sort of thing? Does it seem like a sound idea?
I've been rewatching Marc Levoy's lectures on dig... (show quote)


You seem to be talking about two things. First is there is "HDR photography", and that technique has been around for years. It is commonly done, often to excess with hideous results. Besides PS, there are several other programs available that are specifically for HDR.

Then there is averaging for noise reduction which is used in astrophotography. Again that is a technique that has been around for quite a while.

The catch with these techniques is the scene and camera have to be still across the frames. If the camera moves or the scene changes, between frames, sharp results are compromised.

Reply
Mar 23, 2020 22:10:26   #
bleirer
 
TriX wrote:
Yep, averaging is commonly used to reduce noise with both imagery and electronic signals. The concept is that noise is random or uncorrelated (while the desired image or signal is correlated), so the noise averages out. You gain less and less improvement as you increase the number of averages, so there is a point of diminishing returns.


Better, worse, or the same as a standard 5-7 shot hdr bracket?

Reply
Mar 23, 2020 22:31:39   #
bleirer
 
JD750 wrote:
You seem to be talking about two things. First is there is "HDR photography", and that technique has been around for years. It is commonly done, often to excess with hideous results. Besides PS, there are several other programs available that are specifically for HDR.

Then there is averaging for noise reduction which is used in astrophotography. Again that is a technique that has been around for quite a while.

The catch with these techniques is the scene and camera have to be still across the frames. If the camera moves or the scene changes, between frames, sharp results are compromised.
You seem to be talking about two things. First is... (show quote)


That's where this seemed interesting. He discussed regular hdr, with bracketed exposures, but then discussed this alternative way to handle scenes with high dynamic ranges. The highlights are exposed below the point where they would clip, but normally the shadows would be lost due in part to noise. This way the shadoiw details would be recoverable because the noise would be controlled.

Reply
 
 
Mar 23, 2020 22:56:00   #
TriX Loc: Raleigh, NC
 
bleirer wrote:
Better, worse, or the same as a standard 5-7 shot hdr bracket?


Good question, and I don’t know how to calculate/estimate that. Maybe set up a controlled experiment with a high dynamic range scene? My guess, and it’s only that, is that you could increase the DR more (if that’s the goal) with HDR, but it’s only a guess. I would guess that you might reduce the noise by 1-1.5 stops (increasing the DR by that amount) with a 6 shot average, and you could probably increase the DR twice that amount with HDR. What do you think?

Reply
Mar 23, 2020 23:19:38   #
bleirer
 
TriX wrote:
Good question, and I don’t know how to calculate/estimate that. Maybe set up a controlled experiment with a high dynamic range scene? My guess, and it’s only that, is that you could increase the DR more (if that’s the goal) with HDR, but it’s only a guess. I would guess that you might reduce the noise by 1-1.5 stops (increasing the DR by that amount) with a 6 shot average, and you could probably increase the DR twice that amount with HDR. What do you think?


Hdr is so easy and automated in Lightroom, but stacking in Photoshop isn't too much harder either. Might be worth a try to do both with the same scene. Probably been done by somebody already.

Reply
Mar 24, 2020 07:52:22   #
bikerguy
 
I have used this technique both hand held and on a tripod. It works to reduce noise, especially at high ISO. However, it also allows hand held simulation of very slow shutter speeds such as used to blur water. It is very easy to accomplish with Photoshop.

Reply
Mar 24, 2020 11:03:45   #
mflowe Loc: Port Deposit, MD
 
bleirer wrote:
I've been rewatching Marc Levoy's lectures on digital photography, and in lecture 2 he talks about a method for scenes with a high dynamic range. Instead of the usual exposure bracketing, one would expose many shots of the same scene at the exact same exposure such that no highlights are blown, then average them in Photoshop and then brings up the shadows.

The idea is that the random part of noise is reduced by a factor following the square root of the number of images, so 16 shots averaged would have 1/4 the noise. So the less noisy shadows could then be brightened in photoshop and being much less noisy have more detail to recover.

Has anyone tried this sort of thing? Does it seem like a sound idea?
I've been rewatching Marc Levoy's lectures on dig... (show quote)


If you're going to expose so no highlights are blown, often you're going to have clipped shadows. How do you recover them with this technique? IMWTK

Reply
 
 
Mar 24, 2020 13:55:15   #
SuperflyTNT Loc: Manassas VA
 
mflowe wrote:
If you're going to expose so no highlights are blown, often you're going to have clipped shadows. How do you recover them with this technique? IMWTK


Shooting raw there’s plenty of detail in those shadows that can be brought out.

Reply
Mar 24, 2020 15:09:01   #
mflowe Loc: Port Deposit, MD
 
SuperflyTNT wrote:
Shooting raw there’s plenty of detail in those shadows that can be brought out.


I agree shooting raw is a great help and my D800 has oodles of dynamic range, but if something's clipped, it's usually pretty hard to recover any info. I'm talking about totally clipped, not just underexposed. I know it is easier to reclaim the shadows then highlights. That's just my experience

Reply
Mar 24, 2020 15:36:40   #
bleirer
 
mflowe wrote:
If you're going to expose so no highlights are blown, often you're going to have clipped shadows. How do you recover them with this technique? IMWTK


I think shadows that are truly clipped would be lost, but 14 bit raw files have a lot of latitude below the middle, but noise limits the details. I think it's similar to how some cameras and phones now have night mode, where they take 4-6 shots, each with its own noise in different random pixels, but averaging them makes it seem like the camera can shoot in the dark.

Reply
If you want to reply, then register here. Registration is free and your account is created instantly, so you can post right away.
Main Photography Discussion
UglyHedgehog.com - Forum
Copyright 2011-2024 Ugly Hedgehog, Inc.