[acoufap]

Hi Bruce, I think two of my Posts in the DA ltd. zoom thread really fit well into this thread. So I'll fulfill your wish.

x-post no. 2

I think we are going a bit off topic ... and we should stop somewhere. But before I'd like to comment on Bruce's and Des' posts and write down my latest thoughts.

@BruceBanner
The idea of your technique I know from Pseudo HDR technique. The strength is that you only use one shot. So no problem with movement in the scenery and you can opitmize the use of all sensor data. It helps. But I think, unfortunately, that this doesn't suffice the scenery Des describes.

@Des
I'd say in your scenery the already very good sensor DR is exceeded by a significant amount. So much that you can't create the shot you'd like to geet. So single shot even using Bruce's / Pseudo HDR won't help. I also did moon shots in the past. One learning was ... it is done best in daytime / nighttime transition phase. Then you can get details of the moon and environment much easier cause the contrast isn't that huge. But what could we do in your special the scenery?

My thoughts apart from what I said about using GND filters lead to the conclusion that you might use multiple exposure to cover a very wide dynamic range in a special way. I'll describe my thoughts and results from simple experiments in more detail step by step. Camera used: Pentax KP.

Basic Settings
Since M program is best for nighttime photography we choose this program plus live view showing the histogram. Since we are longing for best image quality and need do some post processing we work with raw files.

The moon moves much faster than we might think. So multiple exposures have to be done quick. Let me recommend HDR mode for this. Later you'll understand why. To fix some basic parameters we set an appropriate aperture, maybe f/8 or f/11, and ISO to a value that's appropriate for the lighting of the scenery and targeted exposure time.

HDR mode gives us three exposures. One baseline "0" exposure and two using plus and minus exposure offset. To get the broadest EV spread, set +/-3EV bracketing in the corresponding HDR Capture menue of your camera. The three exposures are stored in one raw container file, pef or dng. I always choose dng.

The cameras special HDR mode isn't that important since we will do our own HDR processing or blending based on raw files. The camera generatet jpg file isn't important to us. So we can simply set HDR automatic.

What's the next step?

Determining HDR "baseline exposure"
Now we have to determine the baseline exposure time "0" for the HDR capture. It's the one when we subtract 3 ev-stops the moon shows the detail you wish to get. In this exposure nealy everything else will drown in dark tones. How can we determine this exposure time?

Remember? We use M program and liveview. Using the front dial we reduce exposure time exposure time step by step. This doesn't work with manual lenses. So please use an "automatic" lens. With every step (shorter exposure time chosen) live view gets darker. At some point you'll realize detail in the moon. Reduce exposure time further until you see the details you'd like to see in your picture. That should be near ETTR (exposure to the right) in the histogram. Now something unexpected ...

M program - something unexpected in HDR mode
While experimenting today, I discovered something really convenient. Usually M mode doesn't work with ev compensation settings. But, hey - in HDR mode it does! Amazing, really cool! So we now simply can go into compensation adjustment and dial in a compensation of +3ev. That's easy because we are supported by the ev compensaion scale to do it. The baseline exposure time now is set. Exposures will be 0, -3, +3.

Dynamic Range covered
Now we trigger the shot and get the raw file containing three raw images bracketed. They cover an extreme wide dynamic range using ETTR (exposure to the right) technique. Example: if we say the sensor is able to cover a DR of 13ev we theoretically get 13ev + 6 ev = 19ev! That's our playground in post processing. Cool, isn't it?

PDCU HDR split (also see DCU5 and HDR Photography Workflow - Articles and Tips | PentaxForums.com)
What to do next? In PDCU (Pentax Digital Camera Utility) we split the original raw container into three raw files. The filenames show postfixes _1, _2 and _3. I can't imagine to get better base data to work out my image vision. To get there we now can use specialized HDR software or layering and blending techniques. Another opportunity to experiment a lot ...

[acoufap]

@BruceBanner: Hi Bruce, I think two of my Posts in the DA ltd. zoom thread really fit well into this thread. So I'll fulfill your wish.

I'm not sure if the following was said already but it IMO should be said.

Question: When are we confronted with an HDR scenery?

Answer:

step 1) set custom image profile of the camera to "flat". Why? Because the histogram we want to check is a histogram of the chosen profile applied to the raw image data. The profile influences the histogram visibly. I only know one camera that delivers a raw histogram. That's the Leica Monochrome, maybe Phase One does too.

step 2) Check the liveview histogram. Try to adjust the exposure so that the whole brightness destribution values fit into the histogram area. If this is not possible you're within a HDR scenery and it may make sense to use an HDR technique. But you simply also could decide that some highlight and/or shadow values are not important. Then you have to adjust your exposure setting so that the important values are captured. So a clear decision based on a precise exposure is what you have to make. Without working accurate you may loose what you need in post processing. Not to mention we should work with raw files.

[MrB1]

1. Bruce, sorry if I'm being dense but the only image I see in Post #12 is the no HDR photo; no other images and no visible links to images. ??

2. Quote, "Anyone that has used the 3 EV Bracketed value in any HDR mode will know from chimping the back of the screen that the shots will look very HDR like."
The first two of the JPEGs I linked to in Post #7 were captured using Auto HDR with + and - 3 EV set in the HDR options. ??

Philip

[BruceBanner]

1. Bruce, sorry if I'm being dense but the only image I see in Post #12 is the no HDR photo; no other images and no visible links to images. ??

2. Quote, "Anyone that has used the 3 EV Bracketed value in any HDR mode will know from chimping the back of the screen that the shots will look very HDR like."
The first two of the JPEGs I linked to in Post #7 were captured using Auto HDR with + and - 3 EV set in the HDR options. ??

Philip

Hmm... odd, others seem to have seen it ok. I have went back and refreshed the links and c/p them in, can you see them now? Here's a quick link to the post #12

HDR Mode - PentaxForums.com

[MrB1]

Hmm... odd, others seem to have seen it ok. I have went back and refreshed the links and c/p them in, can you see them now? Here's a quick link to the post #12

HDR Mode - PentaxForums.com

That's it - thanks Bruce.

Philip

[thepedant]

Bear in mind that it doesn't matter what HDR mode is set, the camera always takes the same three raw exposures. It is only in the development of a JPEG in camera that the mode becomes relevant and results in very different looking images. If you split the three images using DCU5 and combine them using some other software, then again it doesn't matter what HDR mode (auto, 1,2,3,adv) you had when you took the picture, the result will be the same because the three raw files will have had the same exposures e.g. 0,-3 and +3 (or whatever bracketing was set). The mode does make the thumbnails look different of course as these are also in-camera JPEGS.

[BruceBanner]

Bear in mind that it doesn't matter what HDR mode is set, the camera always takes the same three raw exposures. It is only in the development of a JPEG in camera that the mode becomes relevant and results in very different looking images. If you split the three images using DCU5 and combine them using some other software, then again it doesn't matter what HDR mode (auto, 1,2,3,adv) you had when you took the picture, the result will be the same because the three raw files will have had the same exposures e.g. 0,-3 and +3 (or whatever bracketing was set). The mode does make the thumbnails look different of course as these are also in-camera JPEGS.

Yeah, I get that. The issue I have from my examples in post #12 is that we have a singular file (lets call it '0') and we have examples of -1 0 and +1, -2 0 and +2, -3 0 and +3 and the merge that LR is applying to them is not really providing any benefit at all, there doesn't seem to be additional dynamic range to play with over a the singular 0 file. Why is this?

[Bob 256]

I was going to ask if RT could split the files up, can I ask how you managed that feat?.........................

You’re probably better off with your approach since you mentioned LR can't handle TIFs, but in
case you want to give HDR sub-image separation a go with RT, do the following after opening
RT on your machine.

Open the RT browser and import your HDR RAW file (around 125MB with a K-1 for 3
bracketed shots) by double-clicking on the file thumbnail. It should open in the Editor window
(note only one of the HDR sub-images will display).

1) In the upper right of the RT Editor window (see photo 1), you have options for
processing. Choose RAW (alt+r) and the drop-down should show a "Demosaicing"
group. Choose an option for demosaicing the Bayer layer (I use AMaZE but you can
choose your own - this is also where you choose pixel-shift demosaicing if it applies).

2) Directly below that is a listing of the "sub-images" which should show numbers
corresponding to those contained in your HDR file. This control allows you to select each
of the HDR sub-images individually (allow a short time for the display to process and
update after changing that selection-blue progress bar at the bottom left).

3) You’re now ready to select each sub-image individually and send it to the queue (see
photo 1 - lower left gear wheel icon). Once you have done this for all sub-images, go to
the queue (Photo 2) and make sure they’re all there. Choose where you want to export to
and the type of files you want to export. Files can be saved as 8 bit or 16 bit TIFs or
PNGs, or JPEGs representing the individual sub-images, ready for HDR processing in
the HDR program of your choice. (I use Photomatix but there are others available).

4) Click on the top left arrow and exporting begins. The thumbnails will disappear as they
are processed and exported (processing doesn’t apply here unless you’ve specified some
particular processing for each sub-image).

There are probably other faster ways to do this, but I have access to these programs so the work-
flow is a good one for me and the results are good. Hope this helps Bruce.

[BruceBanner]

You’re probably better off with your approach since you mentioned LR can't handle TIFs, but in
case you want to give HDR sub-image separation a go with RT, do the following after opening
RT on your machine.

Open the RT browser and import your HDR RAW file (around 125MB with a K-1 for 3
bracketed shots) by double-clicking on the file thumbnail. It should open in the Editor window
(note only one of the HDR sub-images will display).

1) In the upper right of the RT Editor window (see photo 1), you have options for
processing. Choose RAW (alt+r) and the drop-down should show a "Demosaicing"
group. Choose an option for demosaicing the Bayer layer (I use AMaZE but you can
choose your own - this is also where you choose pixel-shift demosaicing if it applies).

2) Directly below that is a listing of the "sub-images" which should show numbers
corresponding to those contained in your HDR file. This control allows you to select each
of the HDR sub-images individually (allow a short time for the display to process and
update after changing that selection-blue progress bar at the bottom left).

3) You’re now ready to select each sub-image individually and send it to the queue (see
photo 1 - lower left gear wheel icon). Once you have done this for all sub-images, go to
the queue (Photo 2) and make sure they’re all there. Choose where you want to export to
and the type of files you want to export. Files can be saved as 8 bit or 16 bit TIFs or
PNGs, or JPEGs representing the individual sub-images, ready for HDR processing in
the HDR program of your choice. (I use Photomatix but there are others available).

4) Click on the top left arrow and exporting begins. The thumbnails will disappear as they
are processed and exported (processing doesn’t apply here unless you’ve specified some
particular processing for each sub-image).

There are probably other faster ways to do this, but I have access to these programs so the work-
flow is a good one for me and the results are good. Hope this helps Bruce.

Thanks for that, good to know.

---------------------------

I have weather today that is indifferent to the dull overcast conditions than before, so again I thought I would do some further testing of my approach.

First up we have the single RAW file with Auto Settings applied in LR to give a pleasing overall exposure. Nothing is too blown, shadows are containing details;



EXIF for this file is;

30mm
f8
ISO 200
1/400
Av Mode
-0.3 EV

Auto Settings in LR gave these slider values;



Next up I took a shot with the same EXIF, this time in HDR Auto mode with a 2EV Bracket (-2, 0 and +2). This gave a 120mb DNG file, taken firstly to DCU5 for the split, then I imported the 3 RAW DNG files into LR and did a LR>Photomerge>HDR. I used Auto Align and Auto Settings to produce this file;



The Auto Settings given to this HDR Merge were;



I then changed the Auto Settings values of the HDR Merge DNG file to being the exact same as the singular file and got this result;




Not much difference huh. Furthermore, adjusting sliders of Highlights and Shadows on the HDR vs Singular file produced no further range or scope. At this time I am seeing little merit to this approach. Remember, when we do an HDR we also lose quality due to the stack/ghosting and movement, we also lose Pixelshift as a benefit.

Perhaps I am not pushing things hard enough...

I recall a few years back trying the ETTR principles. I produced this result;



With ETTR I chose the clouds as my exposure, they are really well exposed with plenty of details. You can see however that it was a strong blue sky day and where there are shadows they are extremely black. Perhaps in my example above with the shot outside my house looking across the street I am not reaping the benefits of HDR enough, if I perhaps exposed more for the bright parts (stronger blue sky, really safe and detailed bright parts) we would start to see the benefits of an HDR merge?

Further testing required...

[stevebrot]

There's always more than one way to skin a cat, as they say, and for producing a HDR or tonemapped image, one could shoot a set of exposure bracketed shots, with enough range in the EV to be sure to get good exposure in all areas of the subject.

This is basically what I do minus tonemapping. I have used HDRMerge* and the HDR feature of Lightroom and several permutations of the two tools with good success.


Steve

* Yes, I know...ancient tool, but it works.

[stevebrot]

At this time I am not sure why. Is it because the dynamic range of light from the day of the shot (dull/overcast) was not enough? Was it the merge in LR, it didn't reap the benefits properly?

HDR software basically asserts itself when there is clipping and lets things be when there is none. Your efforts at ETTR accomplished what might be done with a three shot HDR. Where one might see more interesting differences is with a broader bracket. I typically do five shots +/- 2 EV and find that I get better shadow tones.


Steve

[BruceBanner]

HDR software basically asserts itself when there is clipping and lets things be when there is none. Your efforts at ETTR accomplished what might be done with a three shot HDR. Where one might see more interesting differences is with a broader bracket. I typically do five shots +/- 2 EV and find that I get better shadow tones.


Steve

I did some further testing. I haven't uploaded them anywhere to share yet, but I repeated the above scenarios and instead of going for a 'fair dinkum middle of the range kinda standard shot affair' like I did before I used a more ETTR approach and protected highlights further which thus yielded a fairly darkened looking under exposed baseline shot. I then did the HDR thing with a 3 EV bias.

But once again, when in LR and doing the merge and Auto Settings to the baseline ETTR shot and Merged HDR I was ending up with similar looking images, again the HDR not seemingly showing any advantages. For giggles I took a normal pixelshifted shot and used Auto Settings and then sharpened the image further (and therefore sharpened the other non pixelshifted shots as well) and compared, and the pixelshifted shot trumps them all, better details to be expected but better shadows with less noise.

So at this point in time I am starting to see absolutely no merit to the HDR option in the K-1 or KP compared to a single shot or better yet a pixelshifted shot, and I'm wondering if this is really due to the limiting 3 shots vs (what I have always done and been quite pleased with) a 5 bracketed shot. Could it be that a single well taken RAW file with a K-1 just has enough range in and of itself to satisfy most users for producing a well balanced shot, so well in fact that a 3 bracketed shot with even 3 EV bias is offering little to no benefit? That's the conclusion I am starting to come to...

[tduell]

So at this point in time I am starting to see absolutely no merit to the HDR option in the K-1 or KP compared to a single shot or better yet a pixelshifted shot, and I'm wondering if this is really due to the limiting 3 shots vs (what I have always done and been quite pleased with) a 5 bracketed shot. Could it be that a single well taken RAW file with a K-1 just has enough range in and of itself to satisfy most users for producing a well balanced shot, so well in fact that a 3 bracketed shot with even 3 EV bias is offering little to no benefit? That's the conclusion I am starting to come to...

I'm inclined to agree, that the 3 shot HDR is almost always going to be inferior to a 5 shot bracket. I don't think the 3 shot HDR provides enough subtlety, particularly if used with +/- 3EV.
I would also think the 5 shot bracket should give better shadow detail than a pixel shift shot, most of the time. There may be some subjects/conditions where this might not be so.
I tried a few test shots today but the conditions are so poor that it was a waste of time.

Cheers,
Terry

[thepedant]

So at this point in time I am starting to see absolutely no merit to the HDR option in the K-1 or KP compared to a single shot or better yet a pixelshifted shot, and I'm wondering if this is really due to the limiting 3 shots vs (what I have always done and been quite pleased with) a 5 bracketed shot. Could it be that a single well taken RAW file with a K-1 just has enough range in and of itself to satisfy most users for producing a well balanced shot, so well in fact that a 3 bracketed shot with even 3 EV bias is offering little to no benefit? That's the conclusion I am starting to come to...

It does help with a massive dynamic range like the shot below. The first image is a single shot. The second image is a 3 EV bracket with the shots separated with DCU5 and then merged using enfuse.

[acoufap]

I did some further testing. I haven't uploaded them anywhere to share yet, but I repeated the above scenarios and instead of going for a 'fair dinkum middle of the range kinda standard shot affair' like I did before I used a more ETTR approach and protected highlights further which thus yielded a fairly darkened looking under exposed baseline shot. I then did the HDR thing with a 3 EV bias.

But once again, when in LR and doing the merge and Auto Settings to the baseline ETTR shot and Merged HDR I was ending up with similar looking images, again the HDR not seemingly showing any advantages. For giggles I took a normal pixelshifted shot and used Auto Settings and then sharpened the image further (and therefore sharpened the other non pixelshifted shots as well) and compared, and the pixelshifted shot trumps them all, better details to be expected but better shadows with less noise.

So at this point in time I am starting to see absolutely no merit to the HDR option in the K-1 or KP compared to a single shot or better yet a pixelshifted shot, and I'm wondering if this is really due to the limiting 3 shots vs (what I have always done and been quite pleased with) a 5 bracketed shot. Could it be that a single well taken RAW file with a K-1 just has enough range in and of itself to satisfy most users for producing a well balanced shot, so well in fact that a 3 bracketed shot with even 3 EV bias is offering little to no benefit? That's the conclusion I am starting to come to...

Hi Bruce, did you follow my hint in post #17?

If the brightness distribution of your image fits into the histograms range or is only cut minimal you won't win much if anything using HDR mode. This will be the case in most scenarios because K-1 covers a huge dynamic range.