Single Exposure HDR?

[clanthar]

A GND will probably work fine as long as the highlight area all falls within the ND area. Its not that it works for most of the time. It will work for a select few times. As I said more than one way to skin a cat and Photomatix isnt the only way to achieve HDR images. Seas where you have uninterrupted sky will be good for your GND. But the concept of using multiple exposures grew out of broader range of lighting situations. Now without jumping on me and what Im saying, how about just looking at the images posted by sparky and tell me how you would handle them with your GND. Under an extremely limited lighting situation where there is no dynamic range a good exposure can be had with a single shot. But it cant be done, for example, with any of those images posted by Sparky. If it can, please enlighten us.
Clanthars bird shot doesnt have a great dynamic range. Its pretty flat actually. So simply brightening it up globally produces the final image. Not rocket science there.

You've received multiple acknowledgements in multiple posts from multiple people that a GND filter has limited application and of course won't work on a photo like Sparky's. Why are you asking for that again? How many times do you need to hear it? And your argument in that regard has likewise been set aside. Just because a GND filter won't always work doesn't invalidate it. Your multiple exposure method won't always work either.

Now you please explain how your multiple exposure method can capture a flock of flying birds. Should I keep asking you until you answer?

And: "Clanthars bird shot doesnt have a great dynamic range. Its pretty flat actually. So simply brightening it up globally produces the final image." You're completely wrong here. Backlighting isn't flat and globally brightening the shot would not work and that's not what I did.

If you don't want to explain how multiple exposure can work on flying birds from a moving boat then how about a moving roller coaster at the park.

Here's another example -- straight out of the camera JPEG: Exposed to hold the highlight detail in the sky.

Given the backlight sky in the background and the sun still hitting the opposite shore that's a high dynamic range.

Here's the processed photo from the single shot RAW file:

I didn't globally lighten it. It's an HDR photo from a single RAW file.

Look at the back end of the boat. Notice the water churning from the engines. The boat was moving.

Now please explain how you would HDR this with say five multiple exposures.

Joe

 

[480sparky]

I fail to see total black (0:0:0) AND total white (255:255:255) in either the bird or the boat shots.

This means.......... the dynamic range of the image was captured in a single frame by the camera and no HDR process is even needed.

Now, once again...... I will challenge anyone to do the same with the image I posted.

 

[clanthar]

I fail to see total black (0:0:0) AND total white (255:255:255) in either the bird or the boat shots.

This means.......... the dynamic range of the image was captured in a single frame by the camera and no HDR process is even needed.

Now, once again...... I will challenge anyone to do the same with the image I posted.

Well Sparky when you can't see you can always measure:

A look at the straight out of the camera JPEG provides clear evidence that this was a high dynamic range condition. The histogram makes that clear regardless of what your eyes can't see. As for the processed image, of course it holds the highlight -- that's the point. I made sure the exposure held the highlight -- and it does have black. A normal and/or standard photo that can be transferred to print is an 8 bit RGB file. No 8 bit capture technique could have handled the dynamic range of the above photo.

It would be impossible to recover the processed photo I posted of this boat from the camera JPEG. Therefore using a 16 bit RAW capture is a legitimate HDR technique.

This technique can't span as extreme a dynamic range as combining multiple exposures can. I would know that because I do a lot of those. That's OK. I never said it could.

Likewise spacefuzz never claimed that a GND filter could either.

A photo like the one you posted wouldn't be appropriate for this technique or for a GND filter. Here, I'll say that again for you, A photo like the one you posted wouldn't be appropriate for this technique or for a GND filter.

On the other hand this photo couldn't be handled with multiple exposures because the main subject is moving.

Combining multiple exposures is a terrific way to accomplish HDR. When it works it's great. All spacefuzz and I have done is point out that it's not the only way to deal with high dynamic range subjects and that it likewise has it's limitations. Spacefuzz likes seascapes. I spend a lot of time in a boat. Do you realize how dumb a tripod is in a moving boat? I nonetheless encounter high dynamic range conditions. Since the multiple exposure method is limited to static subjects/cameras only, I have to go a different route when I'm on the water. I achieve a fair degree of success doing that as I've demonstrated.

Now I'm going to say it ONE MORE TIME: You showed us a circumstance that can't be helped with either a GND filter or my method of using a 16 bit RAW capture. That's fine. We know that and we accept that. That in no way however means that we don't have legitimate methods for successfully dealing with high dynamic range conditions. We do. And as I said, I use the multiple exposure method when I can. In fact I teach it in my classes. And sure enough I always get some smart-ass student who asks, "how do you do that for sports?" I have an answer for that question.

And I want to know what you're planning to do with that bunch of antsy kids in the foreground of your photo -- hot glue them down?

Joe

 

[480sparky]

Funny. Every histogram I've seen has the numbers 0 and 255 on 'em.

 

[clanthar]

Funny. Every histogram I've seen has the numbers 0 and 255 on 'em.

Right, the arrows also point at the graph. Every ruler I've seen has inches marked on it too. We use rulers to measure things, that's why the inches are marked there. I put four arrows on the illustration -- two point at the numbers and two point at the graphed data. You are having trouble seeing.

Joe

 

[mistermonday]

I am not going to get into the arguments on what is HDR, value of GNDs or choose sides. However, would like to make a couple of points. It is possible to use HDR s/w as a RAW processor to produce some really good results from a single image, regardless of what you want to call the process. Given the right HDR s/w, the results can be much better than anything you can get from ACR or Lightroom.
My 2nd point is that you can process HDR from images of moving birds, boat propellers, and jumping kids. Photomatix 4.1 has a new feature which allows you to make a selection around objects in a merged set of bracketed images, then you select 1 of the source images and Photomatix will insert that area from the the chosen image and it will belnd it into the merged image. This is not the same as the deghosting during the merge - it's a new feature and it works well. Another technique to address the motion problem is to take a normal exposure of a specific subject, allow the subject to move out of the frame, take the remaining shots in the bracketed series, and merge the subject in using the HDR s/w or subsequently in PS. In the end, the results are what count more than how you got produced them.
Regards, Murray

 

[480sparky]

Funny. Every histogram I've seen has the numbers 0 and 255 on 'em.

Right, the arrows also point at the graph. Every ruler I've seen has inches marked on it too. We use rulers to measure things, that's why the inches are marked there. I put four arrows on the illustration -- two point at the numbers and two point at the graphed data. You are having trouble seeing.

Joe

I'm having no trouble at all seeing two numbers. I can load a shot of a gray card, and the numbers 0 and 255 will be on the histrogram.

What I cannot see is both 0:0:0 pixels AND 255:255:255 pixels in the image.



Of course, I don't have the original, full-size raw image like you do.

And I will lay out the challenge YET AGAIN: Try this with MY image.

 

[clanthar]

Funny. Every histogram I've seen has the numbers 0 and 255 on 'em.

Right, the arrows also point at the graph. Every ruler I've seen has inches marked on it too. We use rulers to measure things, that's why the inches are marked there. I put four arrows on the illustration -- two point at the numbers and two point at the graphed data. You are having trouble seeing.

Joe

I'm having no trouble at all seeing two numbers. I can load a shot of a gray card, and the numbers 0 and 255 will be on the histrogram.

What I cannot see is both 0:0:0 pixels AND 255:255:255 pixels in the image.



Of course, I don't have the original, full-size raw image like you do.

And I will lay out the challenge YET AGAIN: Try this with MY image.

I don't know what you're trying to see. I showed you the histogram of the camera JPEG which demonstrates that both 0:0:0 pixels AND 255:255:255 pixels are in that image.

Here's the processed file and histogram:

And this proves that both 0:0:0 pixels AND 255:255:255 pixels are in that image.

==================================================================

I have already told you that this won't work with your image; I told you twice in the last post that this won't work with your image. But hey, I'm accommodating: This won't work with your image.

But that does not mean this isn't a valid HDR technique. There are multiple techniques for dealing with high dynamic range subjects. Combining multiple exposures is one and this is another and using a GND filter is yet another.

The above image had a high dynamic range, to claim otherwise is a mistake. Given that the subject was moving, a 16 bit RAW capture was a good way to deal with it and it falls into the category of HDR photography.

Joe

 

[480sparky]

Yes, you have black in the image. But not much. Details as dark as 5:5:5 can be brought out in post. Once you get below that, things get iffy, and no amount of processing will reveal any detail if the data is 0:0:0 or 255:255:255. I would venture to say less than 2% of the image is below 2:2:2.

The histogram for your original shot shows it is DARK, but the bulk of the data is well above the threshold that can be recovered. Data in the 10:10:10 range and higher is a cakewalk, which the majority of your image is.

In short, you have captured the dynamic range of the scene in a single image, and HDR is not needed. This is exactly what I have stated all along. When the range is captured like this, it's just a matter of knowing how to properly PP the image data.

It has been claimed by some that no matter the range of the scene, it can be captured in a single frame if exposed properly, and recovered in post. I have yet to see anyone actually take an image that is 30% pure black and 30% pure white and make an acceptable image out of it, however.

 

[Bynx]

Its a waste of time trying to explain things Sparky. A broad statement....Single Exposure HDR. Sure a single image is capable of capturing all the range as long as that range is limited. Under unique situations its possible for a GND to bring down the highlights to match closely the settings for the rest of the scene. I dont really understand what the point is here. Its been asked, answered, ignored and twisted around in circles. Each side of the argument says the other doesnt understand. I say, just leave them alone and let them figure it out for themselves. Meanwhile, we can continue to do what we do and get the benefits of doing it. Tone mapping a single raw file may be the best way to approximate an HDR image of movement for now. But as someone said, Photomatix has a fix for that too. No one contends there are other possible ways to create a good looking image.

 

[mistermonday]

Its a waste of time trying to explain things Sparky. A broad statement....Single Exposure HDR. Sure a single image is capable of capturing all the range as long as that range is limited. Under unique situations its possible for a GND to bring down the highlights to match closely the settings for the rest of the scene. I dont really understand what the point is here. Its been asked, answered, ignored and twisted around in circles. Each side of the argument says the other doesnt understand. I say, just leave them alone and let them figure it out for themselves. Meanwhile, we can continue to do what we do and get the benefits of doing it. Tone mapping a single raw file may be the best way to approximate an HDR image of movement for now. But as someone said, Photomatix has a fix for that too. No one contends there are other possible ways to create a good looking image.

+1, & time to retire this thread and move on.
Regards, Murray

 

[clanthar]

The histogram for your original shot shows it is DARK, but the bulk of the data is well above the threshold that can be recovered. Data in the 10:10:10 range and higher is a cakewalk, which the majority of your image is.

Patently wrong. No recovery even remotely close to my processed RAW file is possible from a 8 bit capture of that towboat scene.

In short, you have captured the dynamic range of the scene in a single image, and HDR is not needed. This is exactly what I have stated all along. When the range is captured like this, it's just a matter of knowing how to properly PP the image data.

Well duh! That's what I've been saying isn't it. I captured most of the dynamic range in a high bit-depth RAW file. I have never said that this method captures as much data as combined multiple exposures. To claim that the lighting in the towboat shot isn't high dynamic range is flat out wrong. Is it the highest dynamic range lighting that anyone's ever seen? No, but it's well beyond the limits of normal capture with either film or a digital camera that produces an 8 bit final product. That makes it high dynamic range lighting. My final image would not be possible using either film or by editing an 8 bit camera JPEG.

Oh and once you've stacked a bunch of multiple exposures it's just a matter of knowing how to properly PP the image data. I'd call it a cakewalk.

As for Photomatix being able to handle moving subjects -- that ability is extremely limited. Imagine you have to photograph a parade late in the day. The marching band is marching toward you and three majorettes are tossing batons. They're backlit by full sun -- high dynamic range. Your method simply won't work.

You're stuck on the need to define HDR as a method that requires stacked multiple exposures. There's no reason to be so limiting. HDR should be defined by the image type and not the specific methodology.

This is right off hdrsoft's (Photomatix) website where they explain how to use a single RAW capture to create an HDR image.

==============================================

You can still use Photomatix when you have shot only one exposure in RAW mode. The big advantage, of course, is that you just need one image, so there is no need to use a tripod or to remember to auto-bracket, and it will also work if the subject is moving.
However, the range of "workable" exposures you can get from a RAW file is limited. If you are shooting a high contrast scene, you are unlikely to match the results you would have with taking the scene under different exposures that can cover the whole dynamic range.
There are three techniques for using Photomatix Pro with one single exposure taken in RAW mode:
· Technique 1: Open your RAW file in Photomatix Pro to tone map it directly.
· Technique 2: Convert your RAW file into a 16 bits/channel image in your favorite RAW converter, open it in Photomatix Pro, and tone map it.
· Technique 3: Create two or three exposures in your RAW converter and combine them in Photomatix Pro (or Photomatix Essentials) as if they were "real" bracketed shots

=============================================

Gee! They accept it as legit with the exact same qualifications that I detailed.

Joe

 

[480sparky]

The histogram for your original shot shows it is DARK, but the bulk of the data is well above the threshold that can be recovered. Data in the 10:10:10 range and higher is a cakewalk, which the majority of your image is.

Patently wrong. No recovery even remotely close to my processed RAW file is possible from a 8 bit capture of that towboat scene.

What are you saying is 'wrong'? That my interpretation of the histo is incorrect and 50% of the original image is totally black? If so, then show us your technique that brought out those details. I'm sure there's a lot of big$ software companies that would love to know how you did it.

You're stuck on the need to define HDR as a method that requires stacked multiple exposures.

Perhaps you could point out where I stated that.

 

[clanthar]

The histogram for your original shot shows it is DARK, but the bulk of the data is well above the threshold that can be recovered. Data in the 10:10:10 range and higher is a cakewalk, which the majority of your image is.

Patently wrong. No recovery even remotely close to my processed RAW file is possible from a 8 bit capture of that towboat scene.

What are you saying is 'wrong'? That my interpretation of the histo is incorrect and 50% of the original image is totally black? If so, then show us your technique that brought out those details. I'm sure there's a lot of big$ software companies that would love to know how you did it.

You're interpretation of the dynamic range of the original scene is wrong. You're trying to say that the original scene didn't have a high dynamic range and therefore nothing more than simple editing of the camera JPEG was necessary. That's wrong.

You're stuck on the need to define HDR as a method that requires stacked multiple exposures.

Sorry I inferred that. So then you do agree with me. Great.

Joe

Perhaps you could point out where I stated that.[/QUOTE]

 

[spacefuzz]

You're stuck on the need to define HDR as a method that requires stacked multiple exposures.

Perhaps you could point out where I stated that.

Post #23 page 2