I've tried three different browsers but I don't see the image change
when I pass my mouse over it. I do see what I assume is the original
image when I click on the image. But, since there are supposed to be
three images and only two histograms it's not clear what's what. Is the
smooth histogram from the original or a 16 bit edited version?
The change between original and whichever image I'm seeing is quite
clear to me. Whatever has been done has introduced artifacts across the
entire image that look to me just like digital noise. The smooth skin
tones of the original have been lost.
I think the major failing of the image, and maybe why you don't seem to
be satisfied with your editing, is the blown doorway surrounding dad and
baby's heads. In any photograph our eyes are naturally drawn to the
brightest area and the blown doorway is just too large, too bright and
too much competition for the dad and the kids.
I'm not going to try it because it's a lot of work but if it were my
image my first attempt at recovery would be to fill all the blown white
areas in the doorway with the blue-gray or perhaps the lighter turqouse
color from the edges of the door frames while still trying to hold onto
whatever tonal variations do exist. Difficult and still might not work
but worth a try if it's an important image.
It looks to me like you've made a signficant change to contrast (LCE?)
but I don't think that's needed here. Since I haven't seen the other
image maybe it's much better. But, on what I see, what is needed is
some brightening of the people and a little more selective brightening
of the eye sockets and other shadowed areas of the faces. But I'd stop
well short of exposure changes that appear to mottle the skin. Skin
color and texture are much more important than brightness and contrast here.
Probably the first and maybe the last time I'll ever give the Moose
advice on editing a photo but this one is people stuff and that's my
main game.
Finally, your comment about converting 8 to 16 bit before processing
causes me so say, huh? I'm not sure I could do the math correctly even
if I was more awake but I'm skeptical that this would lead to any
different result after the 16 bits were finally converted back to 8 bits
for printing. But I'm not sure. Certainly having 16 bits to start with
is the desirable case but it's not clear to me at all that converting an
8 bit image to 16 bits will change anything.
Chuck Norcutt
Moose wrote:
> Chuck Norcutt wrote:
>
>>I meant to mention that the round-off and loss of some brightness values
>>when editing a JPEG may or may not be a bad thing since modest losses
>>will not be visible in a print. Although the dynamic range of a JPEG is
>>very limited in comparison to what the camera can record, the dynamic
>>range of a print is much less than a JPEG.
>>
>
> I think you might have avoided some confusion here if you had said
> 8-bit, rather than JPEG. The issue applies to any 8-bit image, whatever
> the source.
>
>>When you look at the histogram of the image after doing significant
>>brightness/color adjustments you may see that it looks pretty "spikey".
>>Instead of a smooth distribution you'll see "holes" where certain
>>values no longer appear and spikes where round-off errors have caused
>>values to accumulate at that spot. If these adjustments are severe
>>enough it will show on the image as "posterization". It looks like a
>>poster painted with too few colors. But what shows up on the histogram
>>and the screen may not show on the print. The print doesn't have the
>>ability to show all the color and brightness detail in a JPEG and what's
>>missing might not have been visible on a print even if it was perfect.
>>
>
> For those who haven't seen it, here is an example. I've taken an image
> and processed it, saving an action as I did so, then run the same action
> on it again, but at a different bit depth. So both versions have been
> identically processed
> <http://www.moosemystic.net/Gallery/tech/8vs16Edit/July406.htm>. Down at
> the bottom, you can see how different the final, 8-bit, histograms are.
> As Chuck points out, it can take quite a bit of this to be visible. I
> think I see very subtle difference as I roll the mouse over the image,
> but it could be wishful thinking and in any case, isn't significant -
> and sure won't show up in a print.
>
> I'm not all that happy with either result, not my best work, although
> the lighting is tricky indeed. Impromptu shots can be a technical trial,
> but those expressions are great! I guess I was distracted by the demo
> nature of the work, but it does show what was intended. I do think I'll
> redo it before sending on to the parents and making prints for them - if
> I can just remember that other kid's name. :-)
>
> I certainly have had images where I forgot to convert to 8-bit at the
> start, that did start to visibly deteriorate as I worked with them. My
> basic idea is that I should convert every 8-bit image to 16, then back
> to 8. Why take a chance of wasting work?
>
> Moose
>
> ==============================================
> List usage info: http://www.zuikoholic.com
> List nannies: olympusadmin@xxxxxxxxxx
> ==============================================
>
>
==============================================
List usage info: http://www.zuikoholic.com
List nannies: olympusadmin@xxxxxxxxxx
==============================================
|