[HDR-photo] HDR as the new digital negative

[listmail at mab3d.com]

On Feb 8, 2007, at 2:40 PM, BillHughey wrote:

> Still the question from the other thread today is whether  
> operations like Chromatic Aberration  reduction are associative  
> over the process of  creating the HDR image from the RAW images.   
> Should it be applied before or after combining into HDR, does it  
> matter?   I think the question was also whether it should be  
> applied over the range of +/- ev settings?

Just converting the RAW data into any viewable image format  
(demosaicing, color interpolation, brightness control) can be done in  
many different ways, so you are already affecting the pixels (input  
for a later HDR) by choosing ACR, Aperture, RAWshooter, Bibble, etc.  
- or even Photomatix with its built in RAW decoder. That much is  
unavoidable, and variable. So any "user controllable" factors like  
exposure compensation, WB, CA correction, vignetting, and even  
removing "purple fringes" from bad highlight interpolation should be  
consistent across all images if they are going into a single HDR. In  
most cases, correcting or removing these in the HDR file is either  
impossible or impractical.

Add something like noise reduction on top of it (which can, in some  
cases, yield vastly different detail results from exposure to  
exposure) and you can really get very different HDRs based on these  
compounding decisions. Even different HDR construction methods will  
affect the output file (compare the HDRs produced from Photomatix and  
Photoshop with the same set of processed TIFFs!). Of course, this is  
all just part of a personal "special sauce" (as RH puts it) you can  
come up with to make compelling HDRs, but it is still many steps away  
from the "raw ingredients." This is why I see the camera RAW data as  
the closest thing to a "negative" I have access to today, lumps and  
all. Of course, I would prefer to have an HDR negative, but that's a  
different picture from a future camera.

> As far as the PITA of laptop assisted AEB, expect to see smaller  
> devices to appear in the next few years that can control as well as  
> download from a camera.  So it will be more like two camera sized  
> devices dongeled.

I know some folks who already have "portable" HDR capture and control  
devices which are just Sharp Zaurus' running gPhoto ( http:// 
gphoto.sourceforge.net/ ), but for any portable device (Linux PDA  
like the Zaurus, Windows Mobile handhelds, etc.), having a USB "host"  
controller in the device is the linchpin to control most cameras. But  
again, my preference for the speed and ease (and to reach the non- 
techie folks) is to have wide-range bracketing control added to  
consumer cameras themselves - going beyond simple 4ev AEB. I expect  
to see actual HDR capture in a few years once "convergence" starts in  
earnest, but that's only once the "megapixel wars" have simmered down  
and HDR processing gets more mature. My bet is on Sony being first,  
of all camera makers - since their vid-cams are already angling in  
this direction, and they have the money and reasons (marketshare  
gains, distinction) to do it.

Fun stuff, but today you cannot go much further back in your capture  
pipeline than the RAW file from your chosen camera manufacturer.  
Everything after getting the RAW off the card involves variable  
voodoo from lots of different folks, including the HDR creation. I  
suppose DNG is a better container for the RAW data, but that also  
relies on DNG not becoming abandonware. When the camera saves JPEG- 
HDR (with an uncompressed EXR option, of course), and the camera  
maker has "optimized" everything in the HDR creation for their sensor  
and processor, then we'll have a new long-term "negative" to rely on  
(and to debate over).

-Mark