Ken Norton wrote:
> Moose, there is a lot of red, white and other BBQ'd meat to chew on. I'll
> opine on a couple of items here.
>
>
>
>> "Provia has ... The highlights in this case would have busted a digital
>> camera image and would have been extremely difficult to have captured
>> without getting out-of-gamut colors.
I didn't quote you to provoke discussion on the specific issue, but in
answer to your later statement "How you read into my essay that film was
superior to digital in this discussion is beyond me."
> One area that I know nearly every digital camera isn't the best at is when
> the sensor reaches saturation. This is a physical trait of the technology.
> Essentially, sensors are A-D devices and in nearly all cases have a hard
> limit after which ugly things happen. This is usually not that big of a deal
> when the highlights happen to be of a non color, but when the highlight is of
> a color that specifically uses one or two of the three sensor colors, we get
> out of gamut problems and we get weird solarization-like artifacts. One
> particular aberration is when the image contains a bright red object--the red
> will turn white or yellow and be fringed by red. (This has been a really big
> problem with fireworks photos).
I absolutely agree. I'm sure you've seen my efforts at recovery of lost
highlight colors by using PS's Match Color tool and my occasional
comments about blown red channel on images posted here or
red/orange/yellow flowers. [Yes, I know. PWP undoubtedly has an even
better, purer function.]
On the other hand, once one fully adjusts to the fact that digital just
isn't like film and learns the capabilities, as well as the limitations,
of digital, comparable, if not, as Dawid asserts, better results in
these situations than film.
> .... LOL, deal. I'm still dreaming of a Fuji-sensor equipped
> full-frame digital-OM body.
Or a breakthrough in Foveon technology.
>
>
> Different, probably, how much, I don't know. A spot reading for highlights
> might be just right.
>
>
> I went back and checked my notes. I'm lousy at keeping field notes, but did
> make a concerted effort this time and wrote about 30 pages in a small
> notebook. Unfortunately, "exif" style notes aren't my forte. I'm too busy
> shooting to distract myself with geeky stuff.
My problem , as well. I start out with good intentions, then slip as I'm
more interested in subjects, light and so on.
> Anyway, it looks like I did use -2/3 exposure-compensation to protect the
> highlights as much as possible AND because I was up against the shutter-speed
> limit. This, however, is more hotographer-preference/experience/desire and
> less technical in nature.
>
My motto is "When in doubt, -2/3 EV." That's the default I have set on
my digicams. I'd rather underexpose in flat conditions than blow
highlights in contrasty ones. Flat ones are low DR and easy to correct,
anyway.
>
> As I said, a DSLR exposure would almost certainly be much shorter.
>
>> Possibly, as Chuck pointed out, even shorter than I calculated, as digital
>> has no reciprocity failure. Thus the issue of light changing during a dSLR
>> exposure might be quite a bit less than you experienced.
> True. I could have used a faster film too, which would have leveled the
> speed playing field and selected a print film to give me additional latitude.
As long as you have the OTF ability to handle changing light, why not go
with slow film and gain on the grain front? With CN film, the over
exposure latitude could be a big help with the highlights. It should
take you back to 0 EV and make the exposure even longer, so there is
more shadow detail available.
> But, in the digital world, just shoot RAW and worry about it in post.
Here's where I start to react as though you are bashing digital and my
use of it. I do "shoot RAW and worry about it in post", but not
unconsciously. I know you've taken a gazillion more digital images than
I have, but I'm willing to bet I've spent as much or more up close and
personal time with individual images in post. All those event shots
don't mean anything to me, nor do flash images.
In my time in PS, I've learned a great deal about what can and can't be
done with my RAW images. I very much enjoy taking shots that I know
can't be done in camera only, but where I am able to visualize the
post-post results before pushing the button.
> My artistic desires, though, may require exposures that run long. This may be
> for star trails, car head/taillight streaks, water reflections, etc. Shorter
> speeds are fine most of the time, but having the flexibility of "confetti"
> free images in time-exposures may be important for some of us some of the
> time. It's nice having options both ways.
>
Sure, although Dawid's testimony about his experiences make confetti
seem like it's not a problem.
>
> Color fidelity -
>
> The 5D is the one Canon body which all others should be compared to.
I'll take that as a compliment on my choice. :-) I spent a LOT of
time researching cameras before going with the 5D. I've never for an
instant regretted it.
> I thought that SmartShooter thing was a joke because of the lens aspect,
I first thought to dismiss it. But I found the methodology was pretty
compelling. Instead of technical comparisons, they standardized
everything, hence the lens choice, and let many folks who make their
living doing people pictures rate the prints in identical conditions.
> but I've seen enough images from other professionals to know that the camera
> has succeeded in being able to deliver good skintones and background colors
> simultaneously. Within the E-system, if you use in-camera JPEGs you have the
> same successful color fidelity, but not necessarily when shooting RAW and
> processing with ACR.
ACR is very useful and powerful, but it has limitations. Lots of people
don't seem to realize how it works by default. As far as I can tell, it
uses very little of the setting information from the camera, preferring
its own default settings. It's default results on a sunset image similar
to yours are quite different in WB and contrast than those form Canon's
own converter, DPP. Fortunately for me, I tend to like ACR's
interpretation better in most cases.
If I were using a camera where that was not true and I wanted to use
ACR, I'd be working on an ICC profile to use in PS for automatic color
balance correction.
> From everything I've seen, I'd suggest that the 5D and E-1 are actually about
> as close as any two cameras from different brands out there.
>
Interesting.
>
> On the other hand, nobody knows what the true colors of this image were.
>
>> You admit both that they couldn't be seen by the naked eye and that the film
>> used has differential color channel reciprocity failure. So the colors are
>> only an impression or approximation of colors we wouldn't see
>> in a "true-to-life", close to all black, web version or print.
> Exactly. But maybe we need to define "true colors". Is what the human eye
> sees in low-light accurate? Is our own vision flawed or is it the
> gold-standard? I wasn't sure what I was going to get when I was shooting
> the scene, but I knew that I was going to get something interesting. Neither
> Provia or Velvia were accurate. But that doesn't matter too much, because we
> make many choices as photographers to warp reality. Besides, I did do a
> little bit of "enhancement" to the scan to get the final contrast and color
> densities where I wanted. This, however, is no different than any image
> processing that we perform on any digital camera image. Scans rarely are
> exactly as the original slide or negative, and we have to modify curves,
> sharpness and saturation in post just as we always do with DSLR images.
>
No argument from me. I personally believe there are such huge
differences between the human visual system and photography that there
simply is no possible absolute.
Depending on the situation, I vary across a range between trying to
duplicate my memory of the subject and one of looking for the best
impression of what the subject "felt" like. I think a lot of good art
photography is "Impressionist" in that sense.
> My comments about the dynamic range of the sky in this photo wasn't very
> clear. In this case, I needed to compress the dynamic range in any way
> possible. During the scanning process I even reduced contrast of the
> mid-ranges just to make it where the saturated colors reproduced.
>
I think perhaps I don't understand very well because I long ago stopped
shooting slide film, and have probably forgotten how contrasty and
unforgiving it can be. As you mentioned above, CN film would be easier
and, at least in my experience/opinion. the contrast range in such a
scene wouldn't be a problem in properly exposed - to the right - digital.
> ...
>
>
>> My supposition is that you want accurate/good skin tones without post work
>> when shooting all those people at events, but are more interested in
>> effecting your vision of what the image should look like in nature.
> Precisely. Event shoots are no place for creative interpretation of colors.
>
>
> I'm not sure just what you mean by "practical dynamic range".
>
>
> Moose, what I am thinking here is that in in reality the bulk of our image
> really utilize five or so stops of dynamic range. Everything else really ends
> up in the toe and shoulder. When a camera has, say, 12 stops of dynamic range
> and you push all of that into a narrower range for printing/viewing that the
> image starts to look flat or just off. So once you get the midtones stretched
> properly, everything else gets compressed into the toe and shoulder.
Yup. We have to deal with what the viewer expects. In abstract looking
images, one may get away with compressing a large DR into the narrower
one our display technologies can handle. But with most subjects, where
the viewer has expectation based on their experience viewing similar
things in the "real" world, certain norms must be followed to please the
eye. Tht's also why HDR really has limited application in creating
natural looking images
One thing wide in camera/film DR does do is give one the choice of where
to place the middle tone ramp and how to treat the shoulder and toe.
This allows a level of natural looking adjustment of apparent scene
brightness that wasn't practical before.
You've kicked me off into working on some examples that may lead to an
essay, maybe even for ZX.
> ....
>
>
>> I wasn't and am not suggesting that you should have anything but what you
>> are using. You are doing excellent work and enjoying it. Why look any
>> further until that changes?
>>
>
> As much as I've enjoyed the E-1 and appreciate some aspects of it, I'm well
> aware that it isn't the end-all, be-all in digital cameras. For its humble
> specifications though, it seems to hold its own quite respectfully. But I've
> kept my investment very low for several reasons. When it does come down to
> buying another new camera, I'm as likely to switch system as I am staying
> put. My reasoning for buying a different system are very similar to yours.
> Just because I can milk every ounce out of a camera doesn't mean that I want
> to all the time. It's actually very difficult to get what I think are
> top-quality images with the E-1.
>
>
>
>> ... As usual, I'm mystified by why one would go to a lot of trouble and
>> expense to use a film with demonstrably inaccurate color. You want funny
>> colors, you can always do it in post. I'm quite serious. I'm sure an ICC
>> profile to convert E-1 colors, for example, into Velvia colors should be
>> possible. With that, Velvia on demand is almost instant.
> Close, but not quite. I truly believe that when the 8-color sensor arrays
> come out we'll clearly understand the difference.
I'm not sure that more colors is what's needed, as opposed to something
on the order of the Foveon sensor to capture all three at each site. The
Bayer demosaicing seems to me to be at the bottom of a lot of subtle
color problems.
> Bit-bending isn't the answer to everything. Not all sensors see the same,
> just as not all films see the same. Point your cameras at a tasseled corn
> field. The tassels will appear reddish-brown with Velvia, yellow with
> Provia, and tan with Astia. Digital? My E-1 and A1 capture the tassels
> differently too. Meanwhile, the green leaves and blue skies may look all the
> same.
>
I'd like to make a distinction here between the tools in editors and ICC
profiles The profiles make point comparisons of 228 different
color/brightnesses. For example, it doesn't just look at midtone cyan,
it looks at cyan at 12 different brightnesses. It's a whole different
thing than simple balance to neutral gray midtone. If you look at the
examples I've posted before, you'll see how dramatically using a profile
may change tonal distribution/curve, contrast and saturation, not just
color. <http://www.moosemystic.net/Gallery/tech/Scan/VuesProf/>
Processed using ICC profiles, the tassels would look the same color in
all cases. I realize thai isn't always what you want, as you've said
earlier:
"I am in agreement that good ICC film profiles are very very nice to have, but
I personally have an issue with the leveling of different film stocks into a
singular look. I select films for their unique personalities and wouldn't want
them to be all the same. Otherwise, I'd just grab the cheapest rolls of film I
could find and produce my
masterpieces on it. :) But, to each his/her own. No harm, no foul."
> I expect Sony to have the first six-color array and this will be the next
> camera war over the following five years to see who can stuff the most colors
> on the sensor. Pixel-count battles have run their course and that
> means we'll see the same thing with colors that we saw with printers when we
> expanded from four colors (CMYK) to five, six, seven, eight...eleven. I
> believe the reason we haven't seen this technology yet (except for oddball
> two-green arrays from Sony) is that to effectively do it, we really need 24+
> megapixel to work with. When this does happen, we'll finally see our primary
> colors actually respond correctly--and maybe be able to finally capture
> purple.
>
Same caveats as above, although with enough sites, and a reduced final
output size, it might work well, depending on other issues. The other
issues I'm aware of are about color filters and sensitivity.
At least to date, color filters with anything but very long, soft
cutoffs have been expensive, and certainly nothing that could be put on
a jillion tiny sensors. Look at the transmission curves on even rather
expensive photo filters. Put a lot of differently centered color
filters on a sensor and you are going to have oodles of overlapping
response. I'm sure processing algorithms will be able to do a lot of
narrowing through correction by comparison, but how well the results
will be is anybody's guess but those in the labs.
The other problem is sensitivity. Assume great, sharp cutoff filters.
Not the light received by any of say six color sensors is no more than
half of what it was for three colors. Suddenly, your sensor system has
lost a stop of sensitivity and noise performance. Not a disaster for
many purposes, but just one more factor to juggle in the design process.
>> ...
> A similar shot, taken last year at the Badlands National Park with the E-1 is
> here:
>
> http://zone-10.com/cmsm/index.php?option=com_content&task=view&id=191&Itemid=1
>
Something wrong with the geography. I remember the shot and the nice
commentary I found at this link. However, I don't think Badlands Park is
on Lake Superior. :-)
Moose
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|