HDR preparation

[poker]

Hi there! I'll be camping in the desert (along route 66) this weekend and will pass by some small ghost town like structures. I want to shoot with the intention of processing HDR.

Any tips for me?

I'll be bringing a tripod, cable release, and a CP filter.

Thanks.

[EOSThree]

When I have used photomatix, I try for a natural look and find using three exposures +2, -2 and on exposure works well while increasing the DR nicely. With my experimentation with photomatix I've all but thrown away my GND filters.

[Frog]

Depending on the contrast within the composition, you may want more than 3.
The few I've done, I did not depend on camera's own bracketing range but metered off several areas in the frame, wrote the varying shutter speeds down and then with tripod, and shutter delay fired off as many as I thought I might need.
Better too many than not enough.

Lord I need to get out and shoot!

[OldClicker]

The more shots, the better (within reason, of course). Check that the graph of your lightest shot is 1/3 to 1/2 up from the left side and the opposite with your darkest. - TF

[armando_m]

I do not think you'll need the CP (circular polarizer ? ) for interior shots, if your camera does autobracketing you should be all set, perhaps just practice how to do the autobracketing

Sometimes I have found the +2, 0 , -2 does not cover the whole range, it is best to look at the histograms, make sure the shots are covering all the DR , with peaks in the histogram all the way from extreme left to extreme right and in between. without clipping, i mean without loosing shade detail on the brightest or highlight detail on the darkest.

The method described by frog should be even better to figure out the right exposures.

If you shoot outside and there are clouds in the sky you want to do the multiple exposures as fast as possible to avoid movement in the clouds.

[poker]

Wow....a bit more involved than I expected. I'll do my own bracketing when time permits and auto bracket when I have to be quick.

Thanks for the tips. I'm excited to share my results.

[poker]

I just wanted to cap off this thread and share my attempts:

[EOSThree]

Your attempts don't really look like they need an HDR treatment. I would guess that they really don't look much different than your middle exposure does. The tonal range here looks like it fits easily into the DR the sensor can capture.

When I prepare to capture an HDR the first thing I ask myself is whether I need HDR to capture the range I am looking for. So I look to see if the tonal range in the frame will fit into the usual range of the sensor. A lot of times the sensor is plenty to capture what is there, especially in the light you've captured in the above photographs. They both look like they were taken within an hour or so of noon. This is absolutely the worst time to take any photo, and because the shadows and highlights are all relatively well lit not at all good for HDR.

If the tonal range won't fit on the sensor, then I bracket +/-2. That gives me a tonal range of about 8 stops for the sensor give or take and then two stops above that and two stops below that for a total of 12 stops which is really close to what the human eye can capture.

Then I generate an HDR image in Photomatix and tone map it to look as realistic as possible. After the tonemapping, I re import the photo to LR and tweak it even more to remove the clownish look in the areas than need it.

[OldClicker]

I think these two are excellent candidates for landscape HDR - especially the rocky canyon with the blazing sun and deep shadows. Both still have shadows and highlights with no detail and probably could have used another exposure in both directions. - TF

[EOSThree]

I think these two are excellent candidates for landscape HDR - especially the rocky canyon with the blazing sun and deep shadows. Both still have shadows and highlights with no detail and probably could have used another exposure in both directions. - TF

I couldn't disagree with you more. Both of these photographs are pretty much exactly what you would see if you were looking at this scene. Yeah there are some dark shadows, but what would you really be adding to these scenes if you brightened them? You'd actually be taking away from the contrast that is there. You need black to white and everything in between. The problem is your camera can't cover all of this ground.

These are HDR attempts already. The HDR really didn't add anything to these scenes. This is pretty much what your camera would record.

Here's an example of extending your dynamic range.

This is the original, yep, I missed...by a bunch. I just plain wasn't paying attention.



After correcting the middle exposure to a decent mid tone, and adjusting both the upper and lower exposures the same amount as the mid tone. I merged and tone mapped to get this, extended DR with great contrast.

[EOSThree]

The more shots, the better (within reason, of course). Check that the graph of your lightest shot is 1/3 to 1/2 up from the left side and the opposite with your darkest. - TF

And please explain this one to me. How would shooting more shots make your HDR better? Your sensor can capture around 8 stops, your eyes around 12-14. How much more do you need? And taking them in small increments just makes it easier to introduce errors by camera shake, slight movement in your support, changes in atmospheric conditions, etc.

Vanilla Days HDR Tutorial recommends 3 exposures. +/-2 and on exposure.

Stuck in Customs uses 5 but would use three if his camera would automatically do it. His machine only shoots in 1EV brackets so he does 5, but the end result is still +/-2 and on exposure.

Photomatix recommends whatever you need to capture the DR of the scene, but most of their examples show 3 exposures.

[daq7]

I believe that more shots probably averages out random noise better, but I cannot prove it. Most of my HDR work now involves taking ONE raw shot and manually blending exposure layers from it. I still often use a photomatix layer as one of the blended layers, but I haven't been doing a lot of shots with HDR software. I agree that for the vast majority of shots -2,-, and +2 are good enough.

[OldClicker]

And please explain this one to me. How would shooting more shots make your HDR better? Your sensor can capture around 8 stops, your eyes around 12-14. How much more do you need? And taking them in small increments just makes it easier to introduce errors by camera shake, slight movement in your support, changes in atmospheric conditions, etc.

Correct, there is a compromise and if movement is a problem, you may end up only being able to use one and/or blending in one (or more) of the original non-HDR images. However the software (and manually, if that is how you do it) works better when there is a gradual change between the gradations it uses. If there were only two, it would have to compensate for a big jump every time it switched from using one to the other. It works better with many little jumps. If you take several (I believe some bodies will do 9 automatically), you may not be able to use them, but if you don’t have them… - TF

[daq7]

Tone mapping can bring out local contrasts and color enhancements that are not at all similar to what you would see if you just looked at the scene even when there is NOT a lot of dynamic range. Of course these types of experiments often do not look particularly natural. I suspect that the first rock scene here has a fair amount more texture than would be the case in the untreated original. I could be wrong or course.

[EOSThree]

I believe that more shots probably averages out random noise better, but I cannot prove it. Most of my HDR work now involves taking ONE raw shot and manually blending exposure layers from it. I still often use a photomatix layer as one of the blended layers, but I haven't been doing a lot of shots with HDR software. I agree that for the vast majority of shots -2,-, and +2 are good enough.

+/-2 is a bigger step, and Adobe does recommend using 1 step increments to keep the image smooth. I think Photomatix is vastly superior though for HDR merging, and Topaz can do quite a job with a single exposure.

In my experience, and I do post a fair amount of HDR photos here, +/-2 works very well with photomatix. Most of the latest landscapes I've posted here are all HDR tonemapped photographs. I've been off my game lately because of a big move, but once this all settles down I hope to get out more again.

[OldClicker]

I couldn't disagree with you more. Both of these photographs are pretty much exactly what you would see if you were looking at this scene.

Which is the point of HDR.

Yeah there are some dark shadows, but what would you really be adding to these scenes if you brightened them?

Texture, and it probably already has.

You'd actually be taking away from the contrast that is there. You need black to white and everything in between. The problem is your camera can't cover all of this ground.

Which is why I would use HDR..

These are HDR attempts already. The HDR really didn't add anything to these scenes. This is pretty much what your camera would record.

Maybe Poker can show us the originals and we can see if more shadow and/or highlight detail was added.

Here's an example of extending your dynamic range.

A good save, but it doesn't change my desire to pull as much detail as I can from the rockky canyon scene above.

TF

[EOSThree]

Correct, there is a compromise and if movement is a problem, you may end up only being able to use one and/or blending in one (or more) of the original non-HDR images. However the software (and manually, if that is how you do it) works better when there is a gradual change between the gradations it uses. If there were only two, it would have to compensate for a big jump every time it switched from using one to the other. It works better with many little jumps. If you take several (I believe some bodies will do 9 automatically), you may not be able to use them, but if you don’t have them… - TF

OK, let says the regular range of your camera is represented by numbers. You take a scene such as the canyon above. We agree your camera can capture about 8 stops of light, and substituting a number for each stop lets say we have a scene with a 10 to 17 range captured.

So the Canyon is represented by 10, 11, 12, 13, 14, 15, 16, and 17, that's 8 "steps" of light. Now I add +1 and -1 to the scene. That gives me 9, 10, etc, etc, 17, 18 now, for a total DR of 10 stops. Now I add, as you suggest, another +1 and -1 to the scene: now I have 8, 9....18, 19. So I've shot 5 shots to get a range of 8 to 19 or 12 steps or stops if you will.

How is it any different if I tag a -2 and +2 on both ends? The -2 will cover 8, 9, 10, 11, 12, 13, 14, and 15. And the +2 will cover 12, 13, 14, 15, 16, 17, 18, and 19. The sensor doesn't suddenly skip a tonal range or zone because you took a +2 instead of a +1. It's still covering the range of a +1, that tone is still within the DR of the sensor, and therefore still captured by that sensor.

The change between graduations will be, or should be the same, even if you took a shot at +/-8 which should give you smooth graduations from -8 through to +8 with a properly exposed 8 in the middle for a total of 16 stops. It's not like the sensor only captures the EC you've dialed in, it just shifts those 8 stops toward the direction you dialed the EC in.

[EOSThree]

Which is the point of HDR.

Sorry I mis spoke here, I meant this is what your camera would record here. I would still bet that this scene is really close to 8 stops and looks remarkable similar to the middle exposure.

Texture, and it probably already has.

Texture is completely different from DR. If it's more detail you are looking for HDR does provide that as a by product. I am speaking of Dynamic Range. It's in the title: HDR

Which is why I would use HDR..

You need this contrast in a scene. To take away all of the contrast yields a flat boring image. It's a balancing act, you need those shadows, that's exactly the reason shooting at high noon is not a good idea, and why shooting at the edge of light makes for much better photography.

Maybe Poker can show us the originals and we can see if more shadow and/or highlight detail was added.

See above statement, from what I can see in the tonality of this scene it is within the capability of most sensors. It may have some true blacks, but true blacks are necessary.

[COLOR="red"]A good save, but it doesn't change my desire to pull as much detail as I can from the rockky canyon scene above.

Once you start pulling stuff out that your eyes know shouldn't be there, you start getting that clownish HDR look.

[poker]

I don't have time read what you guys are discussing until I get home.

I will share the original and maybe you guys can make an HDR image from them and teach me something new.

[Medley]

EOSThree, allow me to answer some of your questions. Why would I ever want to use more than three exposures? In a word- detail. IN an unretouched Raw image, a full 50% of the total information in the image will be found in the brightest stop. 25% will be in the second-brightest stop, and the half-life will continue down into the deepest shadows.

So if I were to take your average 6-8 stop dynamic range image, and reshoot it so the each one of those stop were in the brightest range (thereby letting anything brighter than those stops be blown out), then the resulting image would have 3-4 times as much detail,definition, sharpness -use whatever term suits you- as the original image. In fact, the problem then becomes the fact that you have TOO MUCH definition, and you need to discard some. How you go about that largely determines the success or failure of the HDR image (in the instance where extending the dynamic range is NOT the main goal).

Below are two HDR shots, with their respective middle exposures beneath them. Both were shot using this theory. Each is a composite of no less than seven images (actually, I think one of them had nine)





Now, can you honestly tell me you don't see a difference in the definition of the two images? And yes, three images will still give SOME of that same benefit. Depending on what you keep or throw away, the resulting HDR's could look very similar. SO why bother with the extra exposures?

For the same reason I didn't shoot the scene with my 4mp point-and-shoot: I wanted the best possible image.

So no, I don't think it was a waste of time for Poker to shoot these scenes in HDR, and I don't think the middle exposure will look the same. In fact, I'll be disappointed if it does. Personally, I see a world of advantage in shooting ANY landscape scene in HDR. The one thing you can't do in pp is add detail that the camera didn't capture.

- Joe U.

[Frog]

EDIT: Medley posted while I was writing and explains it more professionally than I can but I'll leave my .02 in anyway.

I believe that how hdr is used will vary from photographer to photographer as will the number of shots necessary to get the desired dynamic range. Sometimes 3 isn't enough and sometimes you'd only need 2.
What I find interesting is that so many rely on the cameras auto bracketing.
I've found times when + or - 2 is not enough. I like to write down the different shutter speeds necessary to expose different areas of a scene and then manually select those shutter speeds as I click off however many exposures I think I might need. I don't worry much about camera movement using a tripod and shutter delay and/or remote along with Photomatix's alignment, it just stops being a problem.
Try doing it freehand and I doubt you'll ever get anything really sharp.
Things that move within the frame is a different story and perhaps using the auto bracketing would be a good idea if that is a problem as you'd be able to get the exposures faster so the movement would not be as great. I'm thinking of clouds especially.
And every time I write something here, it makes me realize how little time I've had to shoot lately and I'm just going to have to tell some people that I won't be around for a day or two.

[EOSThree]

EOSThree, allow me to answer some of your questions. Why would I ever want to use more than three exposures? In a word- detail. IN an unretouched Raw image, a full 50% of the total information in the image will be found in the brightest stop. 25% will be in the second-brightest stop, and the half-life will continue down into the deepest shadows.

So if I were to take your average 6-8 stop dynamic range image, and reshoot it so the each one of those stop were in the brightest range (thereby letting anything brighter than those stops be blown out), then the resulting image would have 3-4 times as much detail,definition, sharpness -use whatever term suits you- as the original image. In fact, the problem then becomes the fact that you have TOO MUCH definition, and you need to discard some. How you go about that largely determines the success or failure of the HDR image (in the instance where extending the dynamic range is NOT the main goal).

Below are two HDR shots, with their respective middle exposures beneath them. Both were shot using this theory. Each is a composite of no less than seven images (actually, I think one of them had nine)

Now, can you honestly tell me you don't see a difference in the definition of the two images? And yes, three images will still give SOME of that same benefit. Depending on what you keep or throw away, the resulting HDR's could look very similar. SO why bother with the extra exposures?

For the same reason I didn't shoot the scene with my 4mp point-and-shoot: I wanted the best possible image.

So no, I don't think it was a waste of time for Poker to shoot these scenes in HDR, and I don't think the middle exposure will look the same. In fact, I'll be disappointed if it does. Personally, I see a world of advantage in shooting ANY landscape scene in HDR. The one thing you can't do in pp is add detail that the camera didn't capture.

- Joe U.

I know HDR adds detail, but I am talking completely in dynamic range. So to extend my dynamic range 3 should be plenty. If I am trying to increase detail in an image then maybe more exposures are needed. The photos you present as examples are excellent. They are well done. I guess they are HDR, but really they are not. They are normal dynamic range each will fit withing the tonal range of most sensors, but as you said the multiple layers bring out a lot of detail. I'm still not 100% convinced you couldn't achieve the same photo with 3 images, or using Topaz.

[OldClicker]

OK, let says the regular range of your camera is represented by numbers. You take a scene such as the canyon above. We agree your camera can capture about 8 stops of light, and substituting a number for each stop lets say we have a scene with a 10 to 17 range captured.

So the Canyon is represented by 10, 11, 12, 13, 14, 15, 16, and 17, that's 8 "steps" of light. Now I add +1 and -1 to the scene. That gives me 9, 10, etc, etc, 17, 18 now, for a total DR of 10 stops. Now I add, as you suggest, another +1 and -1 to the scene: now I have 8, 9....18, 19. So I've shot 5 shots to get a range of 8 to 19 or 12 steps or stops if you will.

How is it any different if I tag a -2 and +2 on both ends? The -2 will cover 8, 9, 10, 11, 12, 13, 14, and 15. And the +2 will cover 12, 13, 14, 15, 16, 17, 18, and 19. The sensor doesn't suddenly skip a tonal range or zone because you took a +2 instead of a +1. It's still covering the range of a +1, that tone is still within the DR of the sensor, and therefore still captured by that sensor.

The change between graduations will be, or should be the same, even if you took a shot at +/-8 which should give you smooth graduations from -8 through to +8 with a properly exposed 8 in the middle for a total of 16 stops. It's not like the sensor only captures the EC you've dialed in, it just shifts those 8 stops toward the direction you dialed the EC in.

Look at one pixel that is 16 in your +0 image. You or your software is trying to decide which of your images it should use for that pixel. With the 3 shots at +2, you have the choice of 14, 16, or 18. With 5 shots, you have the choice of 14, 15, 16, 17 or 18. At the point going across a gradient where you change images, the jump will be larger with the 3-image HDR. - TF

[OldClicker]

Poker - I apologize for high jacking your thread and thank you for giving us the opportunity to discuss this topic.

So that Poker does not need to get further involved if he doesn't want to, here are two other examples that apply to the discussion.

http://www.usa.canon.com/dlc/control...articleID=1786

http://www.boston.com/community/phot...k_brian_1.html (see Horseshoe Bend)

I think these both show how the HDR brings out the detail and texture and turns a flat image into something special.

TF

[OldClicker]

Sorry I mis spoke here, I meant this is what your camera would record here. I would still bet that this scene is really close to 8 stops and looks remarkable similar to the middle exposure.

We cannot tell from the result. See my post below with some examples showing the images used to make the final HDR.

Texture is completely different from DR. If it's more detail you are looking for HDR does provide that as a by product. I am speaking of Dynamic Range. It's in the title: HDR

Maybe detail would be a better word than texture (it's the detail that gives it texture). I don't see higher dynamic range as being the point by itself. If I have an area blown out by +2 and apply HDR so that it is only blown out by +1, it's rather pointless, isn't it?

You need this contrast in a scene. To take away all of the contrast yields a flat boring image. It's a balancing act, you need those shadows, that's exactly the reason shooting at high noon is not a good idea, and why shooting at the edge of light makes for much better photography.

To me, the opposite of "flat boring" is texture. One of the reasons for shooting in the evening light is that the dynamic range is compressed so that nothing (or less) is blown out or hidden in shadow. The harsh light and hard shadows are the problem with mid-day sun.

See above statement, from what I can see in the tonality of this scene it is within the capability of most sensors. It may have some true blacks, but true blacks are necessary.


Once you start pulling stuff out that your eyes know shouldn't be there, you start getting that clownish HDR look.

As I said, let's discuss the two examples I linked below so that we have the before and after.

TF