Thank you, Godfrey, for the detailed analysis.
The possible artifacts from the combination of different sensors is an
interesting aspect that I hadn't thought about. The math behind that
optimization procedure should be quite interesting (minimizing artifacts
occuring from blending information from different-focal-length sensors)
In some way, t's akin projection of the globe onto a 2D map using
different projection types and then merging the results together.
If I were to guess, from what you've described, I suspect the effect of
hand-holding in this case would be one of the most prevalent (besides
the atmospheric haze).
Should you have an opportunity shooting a similar type of scene
with the same camera on a tripod, - I'd be curious to hear about/see the
outcome.
By the way, - from your response, I am getting that the sensors are not
fired simultaneously but in a sequence. Is that right? I understand why
that could be beneficial from the electronics point of view (lower demand
on the processor and the power), but it is obviously undesirable from the
photographic point of view, - for the reason you've mentioned.
As for pixel-peeping, - it was my inner-researcher's curiousity about
how things work rather then being grumpy about noise/quality/...
And once the full-res image loaded on the screen it was hard to avoid
looking at the detail.
Cheers,
Igor
Godfrey DiGiorgi Thu, 24 May 2018 12:32:38 -0700 wrote:
Thank you, Igor!
There are several sources of "noise" or, more precisely, imaging artifacts
in this photograph. First is just atmospheric haze and the necessary
rendering manipulations to minimize it; yesterday wasn't the clearest of
days. Second is the fact that it's an auto-merge of six hand-held
exposures (no, I did not take the tripod with me), so a bit of image
artifacting at the blending seams happens. And third, remember that any
Light L16 single image is actually a merge of 10 out of 16 small sensor
camera exposures, with three equivalent focal lengths (28mm, 70mm, and
150mm) being combined to achieve each single exposure. There are due to
this some small blend artifacts which crop up in certain kinds of subject
areas, where the camera images are overlapping, and foliage is one of the
hardest subjects to image and merge like this due to wind and "not quite
exactly at the same instant" exposures made at the 10 cameras.
So while the total pixel count is very high, the imaging qualities and
actual picture resolution/acutance differs somewhat from a single
large-sensor camera photo, and is variable across the field of view. I've
done some experiments by shooting a blank wall with just a little texture
and then bashing the tone curve into truly pathological shapes that brings
out all the individual camera exposure areas and where they blend: It's a
fascinating study. The benefit of a camera like this is to have
essentially a 50Mpixel camera with an equivalent FoV range from 28mm to
150mm in a package that's just a hair bigger than a "Plus" sized iPhone
and weighs similarly. It's my favorite "pitch it in a tiny bag and go for
a bicycle/motorcycle ride" camera by far, aside from the iPhone 8 Plus
itself.
But of course, only photographers sit around at 1:1 resolution examining
all the pixels in an image. I do too, just for the fun of it, when I'm
learning and analyzing the performance of my cameras. But when I'm viewing
a photograph like this, what I really want to see is it printed nice and
big, and view it normally from 5-6 feet away, looking closer at details
occasionally. The artifacts of its capture and processing will likely be
invisible then. I think I want to have this one printed for me on smooth
canvas … it will print to 12" tall by 50" wide at 360 ppi and should look
stunning. (I could print it myself, but I so infrequently want to print
anything that large it seems absurd to buy a roll of 13" paper and battle
the printer just for this job… :-) That little print job will run me about
$180 for a stretched canvas wrap. Perhaps I'll have it done in August and
call it a birthday present to myself.
onwards!
G
—
"I don't mind if you don't like my manners.
I don't like them myself. They're pretty bad.
I grieve over them long winter evenings." – Philip Marlowe
On May 24, 2018, at 10:18 AM, Igor PDML-StR wrote:
That's a gorgeous view!
I just realized that I've never driven that way from San Jose.
I was just curious: When I look at the full-res image, fully zoomed-in, I see
something that looks like noise. At first, I noticed it in the
blueish-grayish portion (the mostly-urban valley and trees surrounding the
buildings). I attributed that to the haze in the air.
Then, I've noticed a similar noise in the green trees and bushes that are to
the right of the nearest dozen-or-so trees just behind the grassy area. Then
I've noticed those almost everywhere...
It seems to be less apparent in some areas, and more apparent in others.
Is this just the sensor noise or the optical limitation of the lens (e.g. of
the diffraction-limit of the lens aperture)?
Or, could be some type of artifact, e.g. of processing/stitching, etc.? (I
assume you were using a tripod and not testing your bike's suspension while
shooting, so the camera shake is probably ruled out. :-) )
Do you know what's the culprit?
Thank you,
Igor
Godfrey DiGiorgi Thu, 24 May 2018 06:54:40 -0700 wrote:
Took a ride up Mount Hamilton Road on the Moto Guzzi V7III Racer yesterday,
shaking out details in the suspension settings. I didn't quite make it all
the way to the top, due to lack of time, but about half way up I caught this
six frame panorama of Silicon Valley looking from the southeast:
https://flic.kr/p/24JaZqU
The real magic is in the full resolution image (87 Megapixels!), but it's
about a 20 Mbyte image file, compressed. If you want to take a look, here's
the link:
https://farm1.staticflickr.com/975/27452792127_01f7db597c_o.jpg
Enjoy!
G
--
PDML Pentax-Discuss Mail List
[email protected]
http://pdml.net/mailman/listinfo/pdml_pdml.net
to UNSUBSCRIBE from the PDML, please visit the link directly above and follow
the directions.
