[HDR-photo] Starting out...

[listmail at mab3d.com]

On Feb 12, 2007, at 11:51 AM, Otto Feiler wrote:

> Gentlemen,
>
> I am completely new to HDR imaging, but find it fascinating. I was  
> reading this thread, and was wondering if you could:
>
> stitch each group of bracketed image -> create HDR -> tonemap
>
> I hope this makes sense. In other words, do the stitching of the  
> -2EV shots, stitch the 0EV shots, stitch the +2EV shots, and then  
> do the create HDR, and then the tonemapping. Would that work as well?

Hi Otto,

This is how I do it most of the time. I don't entirely understand why  
Bernhard didn't include that method, but perhaps it was because with  
this method you might not get predictable results using auto- 
alignment or ghost removal algorithms when building the HDR. The key  
to success with this method is that, during stitching, you must be  
able to *exactly* recreate the transformations for each set of  
images. I tend to stitch one set (with the best visible details) then  
batch stitch the other sets using the same transformations, using  
panotools in my case. In some software (Autopano Pro, f.e.) this  
rigid repeatability cannot be achieved. Also, some of the blending  
algorithms that attempt to correct for parallax and moving objects  
(SmartBlend, Enblend 3) cannot be used because of the potential for  
differences between blends in each set.

So, yes this is a workable solution (and far more predictable than  
tonemapping *before* stitching!), but it requires care and precision  
when shooting the panorama frames, and moving objects will most  
probably show up as "ghosts." It works well for my static landscapes,  
in any case! IMHO, any tonemapping done *before stitching* negates  
the effectiveness of the best tonemapping operators, since the full  
image (and its final HDR pixels) are not being considered. Also, the  
tonemapping decisions you make as an artist are similarly myopic,  
since you are not seeing the full image either!

-Mark