[rsoph]

I have been experimenting with the HDR feature on the K5 with some respectable results. The K5 HDR feature has 5 modes (Auto, Standard, Strong 1 - 3). Does anyone know what the different modes do? When I venture outside auto mode, I don't really know what to expect and my experiments have been rather inconclusive. Can someone explain what the manual doesn't?

Attached Images

 

[Docrwm]

Excellent question, I too have had the same experience so far. No real idea what to expect across the 5 so far. Looking forward to reading about it here. Thanks for asking the question.

[blackcloudbrew]

I've been doing HDR now for four years and while I consider myself accomplished at it, I do it all in post processing from the RAW files. While both the K7 and K5 can do HDR from the jpegs in the camera, I've never played with it. I may not be unique in this either. I suggest that you set your camera on a tripod, find a scene with a lot of variation of light and shadow and do some test shots with the different modes. See which you like.

[joe.penn]

My understanding is the exposure and merging algorithm is adjusted accordingly based on the selected HDR mode:

HDR Strong: -3.5 | 0 | +3.5
HDR Normal: -1 | 0 | +1

etc, etc, etc...

That probably did not help much in answering your question though...

[rsoph]

I have been trying just that, but unfortunately the one I like varies from shot to shot and I guess I was hoping someone could explain what the camera does in each scenario so I might venture off the Auto mode and perhaps predict which mode might be best in a given situation. Perhaps we can start with someone explaining or even hypothesizing on why they name the modes "strong 1 - 3". What makes an HDR image more or less "strong"?

[rsoph]

The Manual suggests that HDR always uses -3, 0, +3 as the exposures from which it works, but perhaps I am just reading that wrong.

[joe.penn]

Here, read this (it's for the K7 but could also explain a little more):

The Pentax K-7: The era of in-camera High Dynamic Range Imaging has arrived! from Adorama Learning Center

As for the -/+3, that could be the case and then different dodge/burns are applied to the copies before merged. Maybe a few others will chime in on this - doesn't seem to be any info really on the exact internal process out there...

[rsoph]

The article you site on Adorama includes a comment with a reference to the patent covering the process. I am afraid both my lawyer speak as well as my technical understanding of the processing of digital imagery prevent me from understanding what it says, it does appear to give someone the opportunity to translate it for the masses (please)!

United States Patent Application: 0080252768

[vanyagor]

The article you site on Adorama includes a comment with a reference to the patent covering the process. I am afraid both my lawyer speak as well as my technical understanding of the processing of digital imagery prevent me from understanding what it says, it does appear to give someone the opportunity to translate it for the masses (please)!

The patents are purposefully worded in such a way that it is impossible to understand anything. The way it works, engineers write the patent then forward it to the lawyers, whose task is to bring it in accordance with the standard which means make it totally impossible to understand

[maxfield_photo]

I was excited when I heard the K7 would feature in-camera HDR, but then I found out it only works in JPEG mode. That's like taking two steps back to (hopefully) take three steps forwards. In a lot of HDR shots I see on flicker, the original scene probably only had 5 stops of DR to begin with, and the photographer has transmogrified the scene into a 20 stop monstrosity complete with halos and colors that don't exist anywhere in nature. The HDR shots from the K7 by comparison look pretty natural, but still, you have to give up all that control and latitude for correction. My thoughts: just shoot in RAW and create responsible HDR shots in post.

[joe.penn]

After a crap ton of thought (this topic is intriguing me) and playing around with a few different photos, I think the in camera HDR can be somewhat duplicated using the Lightness value (HSL) of the RGB channels - for example:

HDR Normal:
-3 | 0 | +3
Correctly Exposed Snap:
L | R=0, G=0, B=0 (where 0 is neutral/normal value)

HDR Strong:
-3 | 0 | +3
Correctly Exposed Snap:
L | R=+20, G=+20, B=+20 (where +20 is the adjusted channel lightness value)

When HDR merging the above, it seems to mimic what the "in camera process" does - maybe not 100% but it is close. I am sure there is more to it like contrast adjustments and such; this would be a good project for someone to tackle, may even lead to stumbling upon a new HDR routine...

[rsoph]

I was excited when I heard the K7 would feature in-camera HDR, but then I found out it only works in JPEG mode.

Its not clear what mode the initial images are shot in (RAW vs JPEG), the Adorama article touches on that. Do any of the out of camera programs allow you to save the HDR result as RAW? Seems to me that wouldn't be possible in the purest sense of the term.

Now that I think about it, it is really unnecessary for the camera to pick a file format for the intermediate images. They are never files. All it really needs is the memory image of the sensor from each exposure and it would use its own proprietary algorithm to manipulate the bits into the final result..

[alspe]

About HDR photos in K-5,

have you tried multiexposure against HDR mode?

And another, in PS/Aperture (program) do lighted/darker parts of picture or do you use shadows/highlights if try to expand dynamic range?

[demp10]

A RAW image, especially from the K-5 with 14-bit data, is essentially an HDR image. It holds a theoretical 14 f-stop dynamic range.

Given that the displayed (or printed) images are 8-bit, the RAW file represents a -3|0|+3 set of data in the case of K-5 or -2|0|+2 in the case of K-7 and other 12-bit models. With proper exposure in the middle, the entire 14 f-stop range can be captured in a single exposure and with post processing (pulling down the highlights and boosting the shadows) an HDR-like image can be created.

A more advanced mapping can be done by manipulating the histogram curves to get the desired levels.

Obviously the noise levels in the dark areas will be higher than traditional HDR with separate exposures, but a single exposure eliminates other HDR issues like registration errors of images, light changes (shadows, white balance, etc.) and movements.