[HDR-photo] xmp files

[Ferrell McCollough]

Hello,

I think xmp files for HDR images would be a benefit to the user.
Although PM includes in the file name the number of images and the starting file, for example Set01HDRI5from_DSC2535.hdr,  it doesn't help us if we have 5 bracketed images and only use 2 or 3 of the set.  An xmp file should list the file name w/ extension, image size, exposure values, and EV steps.

Some cameras, like Nikon have a max 1EV step when using AEB mode, so in order to use AEB and get a +2EV and a -2EV sequence we need to shoot 5 images and delete or not include the +/-1EV images in the merging. I've notified Nikon of the need for AEB of 5frames @ 2EV steps. Hopefully it's just a firmware update. Other camera's like Pentax, Fuji and Olympus only allow 3 frames at a max of 1EV step (except for the Fuji S3 Pro which allows 2EV steps)

Here's a typical Nikon AEB shoot sequence:
5 images @ 1EV step:

5385.nef    0EV
5386.nef    -2EV
5387.nef    -1EV
5388.nef    +1EV
5389.nef    +2EV

If I used 3 images of this 5 image set I'd like to know in the xmp file if it was the 0EV, +1EV, -1EV, or the 0EV, +2EV, -2EV for example. Using all 5 images isn't necessary.

Here is an example of xmp info in which the +/-2EV steps were used. It includes the file names, sizes, exp and EV step.

DSC5385.NEF    4320x2868    1/50sec - F/18    +/-0EV
DSC5386.NEF    4320x2868    1/200sec - F/18    -2EV
DSC5389.NEF    4320x2868    1/13sec - F/18    +2EV


Canon on the other hand is able to shoot 3 frames @ 2EV steps in AEB mode. For a high contrast scene in which +/-4EV is needed, the photographer can dial in -2EV then shoot AEB for a sequence of -4EV, -2EV, 0EV. Then dial in +2EV and shoot another AEB set for a sequence of +4EV, +2EV, 0EV. The end result is a very high dynamic range capture but there are two images that are 0EV. One must be deleted or not used in the merging. It would be nice to know, especially in ghost analysis, which 0EV was used.

So as you can see present camera technology is designed for small incremental bracketing or limited bracketing. This is good for choosing the best single shot of the scene but not for HDR merging. So as we shoot we capture more images than we need, some get deleted, and some become part of the merging. An xmp file will show us which ones were part of the merging.

Shooting in AEB mode is the superior method as it doesn't require touching the camera between images. The less you touch the camera the less the chance of camera movement. Also AEB allows for continuous shooting at high fps.  Shooting in manual, or using exposure compensation requires dialing-in each exposure and unless the camera is tethered there is a chance of movement even if on a pixel level. Shooting in manual also has a slight disadvantage of not including EVsteps in the Raw xmp file, just f-stop and shutter speed.

With today's limititions in DR of our monitors, HDR images are a stepping stone to the final tone mapped image. How much we need to open and view our HDR images will be limited, it's mostly archived information once the tone mapping is completed. If we decide later the tone mapped image is not to our liking we can begin with the HDR file again or begin with the raw files again. The tone mapping data would be very valuable either with the hdr or as a separate file like PM has now.

I believe we should always save our original RAW files, especially the 0EV for that single shot metered exposure of the scene. I think it would be nice but somewhat redundant if the HDR image contained info that the Raw file has, white balance, ISO, Color Mode, Focal length, lens used, etc. It doesn't change between exposures so it's not critical. File size doesn't change between images, but it's important to see if an hdr is derived from a reduced image set, for speed of processing. 

On another note, I think it would be great if the HDR image had information of the real world dynamic range captured expressed as EV range. I'd like to know if the scene had a range of 6EV, 9EV or even 12EV. It would be valuable in learning to associate a particular scene with a dynamic range. That would then give us some insight into bracketing range.  Right now the only way to learn the DR of a scene is to spot meter the scene and record the EV of the darkest and lightest part of the scene or as most of us do, just wing it. Of course the sun would have to be eliminated from the calculation, as photographers we rarely want to expose for the sun's disc but only the blue sky around it.

So I think an xmp file should have:

File names
File sizes
Shutter speeds & f-stops
EV's
Dynamic Range









-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.hdr-photography.com/pipermail/hdr-photo/attachments/20061210/d12d6658/attachment.html