Enhance! Superresolution Tutorial in Adobe Photoshop

In this tutorial Ian Norman shows us how to enhance the resolution of a camera sensor with a technique called superresolution. With this technique, it's possible to mimic the sensor-shift high-resolution mode found on cameras like the Olympus OM-D E-M5 Mark II to squeeze more megapixels out of the sensor. In his example, he increases the resolution of a 24 megapixel photo to more than 90 megapixels. See the full written tutorial at http://photoncollective.com/enhance-p...

Gear Used:
Sony a7ii http://amzn.to/1vOttdO
Sony Zeiss Sonnar T* FE 35mm/2.8 ZA http://amzn.to/1EdwLhL
Sony RX100M3 http://amzn.to/1AT7CJY

Music Opener: "Recruitment Gong (Instrumental)" by BOPD (http://jasonsigal.cc) CC BY-SA 3.0 Unported

---

Closed Caption:

[Music]
everyone I mean Norman from photon
collective and today I want to talk
about technique called super-resolution
super-resolution is essentially a way to
increase the spatial resolution
capability of your camera which actually
means that we can almost quadruple the
amount of resolution that your camera
could create so for example if you have
a 10 megapixel camera this technique
will allow you to take a photograph
that's nearly 40 megapixels
this isn't like a panoramic stitching
technique or anything like that this is
like actually utilizing a trick to
maximize the amount of detail that we
can create with your sensor so i'll go
ahead and show you some examples here of
what i'm talking about i'm currently in
san jose costa rica right now just on a
trip and I use this technique while
doing some street photography just
around the neighborhood that we're
staying in and I was able to take the 24
megapixel images from my sony a7 mark
too and I increase the resolution of
that image up to about 90 4 megapixels
so I've got two different examples here
side-by-side one of them is an original
image straight out of the camera 24
megapixels
I mean this camera is already really
really detailed you can see a whole lot
of detail in this image so if i zoom
into it you know about a hundred percent
here we can sort of scroll around this
is taking about iso 100 within aperture
about f8 i stopped down to try to
maximize the amount of sharpness in the
image and i want to sort of zoom in on
an area like an example area here just
to sort of show you a comparison between
this image so we're going to zoom in at
like sort of the sub-pixel level here so
we can really see the detail you can
start to see you know some some aliasing
in these power lines and if i if i go
over here to the
the image on the left this is our super
resolution image this image has about 94
megapixels in it and if i zoom into that
same location to get there so i zoom in
the same location you can see that
there's a lot more detail the most
obvious thing right here is this
corrugated steel roof top you can see
each individual corrugation and you can
even see the fasteners that they're
using a hold on the corrugated roof top
over in our are 24 megapixel image you
can't really see any of that detail and
even detail in the the power lines if we
sort of look at this little like bundle
of wire here you can just see that
there's is its much sharper and that's
not to say that that the original image
isn't a high-resolution picture I mean
it's 24 megapixels let me show you
another example of where this technique
is is is useful so we've all seen the
sort of thriller crime-solving TV shows
and movies where they have some sort of
security surveillance footage and
they're like trying to find some key
detail that's gonna like solve the
mystery for them in the security camera
and be like let's zoom in on this guy's
eyeball and look in the reflection on
his cornea so that we can see you know
some image of the bad guy or something
like that right and I mean obviously
it's like super far-fetched but one of
the things that might actually be
possible with the super-resolution
technique and it has to be sort of like
exactly the right conditions to to make
a lot of sense to even try but one of
the one of the advantages is we might be
able to increase the legibility of
certain text any image so for example
there's this car parked on the road and
you can't really read what its license
plate says I mean if I looked at that
i'd probably say something like I don't
know maybe like 25 5 282 or something
like that but in a super resolution
image we you know we're able to almost
make out what it says we can definitely
see that the last three numbers are 201
and then the first numbers maybe it's
806 dash 201 or could be b DG dash 201
you know i'm not quite sure but if we're
trying to solve a crime or something
like that we needed that license plate
number
maybe we could at least narrow it down a
little bit by using the super-resolution
technique so the way that this technique
works is we actually combine the data
from multiple images and by doing that
we're able to sort of like increase our
statistical data set we can we can
increase the amount of pixels that were
gathering and the trick with it is to
actually shoot handheld so we're
shooting like a whole bunch of images i
usually shoot a sequence of about 20
photos and by shooting handheld the
little motions in our hand sort of they
will that they move parts of the image
to a different part of the camera sensor
and that allows us to utilize the camera
sensor on the sub-pixel level red green
and blue values are recorded for every
single location to create an image that
has much higher resolution than the
actual pixel counts would make you
believe there's another type of
technology that's coming out that does
this built into the camera and it's made
for tripod shooting and it actually uses
the camera sensors image stabilization
system to move the sensor while taking a
photo so the sensor sit shift technology
in the omd em5 mark to let the camera
take a 40-plus megapixel image from a 16
megapixel sensor so that's essentially
what this technique will emulate is
being able to move the sensor as we're
taking the image and we do that by
taking a burst of images there there is
one downside to this technique and that
is that any motion that's in the image
ends up blurring so the cars on the road
that we're moving here and this guy
walking down the road is just a kind of
a big blurry mess you can see that
there's a car entering the intersection
in this other man
and and that's also kind of blurry mess
so it's not advantageous for all
shooting situations so this technique is
mostly applicable to static scenes i
really want to try this technique out on
landscape photography I think it's a
perfect application for landscape
photography where your setup is nice and
slow
you can sort of you know wait for the
light you have a lot of time to make
your exposures I really think that
landscape photography is a great
application for this for this technique
one of the other advantages of doing
super-resolution it isn't just the
increase in detail but is decreasing
noise if we go to the same portion of
the image on the original we can sort of
see just how much how much gray near the
original image is you know there's
there's a lot of random noise in there
and that this is this wasn't taken at
high iso or anything
this is a way to get noise that's at a
lower level that even I iso 100 and you
know that's that's actually a huge
advantage to pretty much any situation
situation if you want the cleanest image
possible so when it comes to setting up
your camera for this type of photography
there's a few things that you want to
think of and the first and foremost is
doing everything we can to maximize the
amount of resolution that your camera
can create usually I like to stop down
to about f8 most lenses are just about
as sharp as they can possibly get at FA
i use a just a 35-millimeter f 2.8 lens
the little compact lens on my sony a7
mark to stop down fa and that'll
probably help a lot with this technique
we also want to shoot a series of images
relatively quickly i like to set my
camera to continue a shot mode or or
burst fire mode you'll notice as you're
shooting that
your hand motion will still introduce
some movement between each photograph
and that's actually a good thing that's
one of the tricks behind this technique
is that we're moving the image to other
parts of the camera sensor I've got my
20 images here of that same street scene
that extra do earlier i'm so let's take
a look like a quick look at it one of
the things that I like to do is zoom in
and we can actually control the amount
that we're doing in here if I open up
the navigator and I like to zoom in
about like three two one just so i can
really really see some detail and all
usually scroll through each image and
just check to make sure they all look
sharp definitely be sure to exclude any
images that are out of focus from your
image sequence when when trying this
technique so now that I've checked all
of my images there's a few ways that we
can bring these into Photoshop as a
stack so we can start the
super-resolution method you can select
all the images and hold down shift and
click to select all of the images and
then you can just right-click and choose
edit in and then open as layers in
photoshop so if you want to open all of
your images from Photoshop into layers
there's a way that you can do that as
well and that is to choose file and then
scripts and then load files into stack
and from this dialog box in the lower
layers dialog box you can select browse
and then just browse for all of your
images and loading your images in
Photoshop will take a little bit of time
so you know just be patient as it loaded
loads each image
ok so now it looks like we have all of
our images loaded in the photoshop and
we can look and see all those images
loaded into the layers palette here so
the very first thing we want to do is
increase the resolution of our image to
our target resolution this technique
works best if you increase the
resolution to four times what it was
originally for about two hundred percent
on either edge so i'm going to go to
image and then image size
and then in the image size dialog box we
can we can adjust our height and width x
% I'm gonna go to two hundred percent on
the with now one of the important things
here is under the resample checkbox you
want to resample using the simplest
resampling method which is called
nearest neighbor right now we're ready
to click ok and this will take a little
bit of time just because we're taking an
image that was originally 24 megapixels
and we're up sampling it 296 megapixels
and it's also doing this for all 20 of
our images so it's gotta work away at 20
photographs all at once so now we have
our 96 megapixel image but it's not
truly a 96 megapixel image of resume in
like to that area on that roof that I
was showing as an example earlier we can
see that there's still you know no
corrugation detail on the roof there's
still a lot of aliasing happening on
these power lines and you know if you
zoom in at the pixel level you can see
what's going on
so each pixel each like like pixel that
we see in color is actually made up of
four pixels and that you know that's
what we did with our resampling so what
we're going to do is we're going to
realign all of the layers so that any of
this camera motion that sort of like
happened in between each photograph
isn't it you know isn't visible so so
all the details from layer to layer will
align with each other in order to align
all of these images together we can do
it automatically if we just select all
the layers selecting the top layer and
then i'll scroll down and hold down
shift and select the bottom layer so I
have all of the layers selected and then
I can go to edit and then click the auto
align layers button
we're going to use the automatic
projection method we want to make sure
that vignette removal and geometric
distortion are unchecked and then we can
go ahead and click ok and this is
probably the other than loading our
images into Photoshop this is probably
the slowest part of this process we rely
on for super resolution is that
photoshop lines these images at the sub
pixel level so now that we have four
pixels for every pixel that we had in
the original image Photoshop can
actually align the image you know it has
essentially like for more locations
where it to put something by doing that
we're able to more carefully refine
where the detail should actually exist
in the image and details that are
normally lost you at the sub pixel level
can end up being revealed if we use
enough images the whole trick behind
super resolution is that we're actually
using 20-plus images to create a single
image and by using that much data we can
sort of you know use statistics
essentially in order to figure out
what's actually going on at the sub
pixel level
ok so now it looks like Photoshop has
finished aligning all of our images so
one of the checks that we want to make
before we composite all these images
together is to make sure that they're
actually properly aligned the easiest
way to do that is to turn on and off our
layers so usually turn it off and on a
couple times i'm using the little eye
icon to the left of the layer in the
layers palette so just go through and
check each image one at a time just to
make sure that it's properly aligned
with the image below it and it looks
like photoshop did a pretty good job
here
ok so now that I've checked all of my
layers we can turn back the visibility
of each layer so they're all visible by
clicking on clicking and holding on the
eye icon to the left of the layer in the
layers palette and they just sort of
like dragging up and that'll turn on all
the layers we you know to make sure that
they're all turned on so that they're
visible so now that everything's aligned
we're almost done
the very last thing that we need to do
is average all of the layers together so
that each layer is contributing to the
resolution of the image we can do this
really easily by just adjusting the you
capacity of each layer and the the
amount of opacity for each layer is sort
of determined by the number of layers
that you shot i like to use 20 layers in
this technique because a hundred percent
is easily divisible by 20 so each layer
will contribute 5% to the final image so
the the the best way to set your
capacity to properly average all of the
layers together is you from top to
bottom
you set the layer to the lair number
divided by the total number of layers so
I have the first layer here at the very
top and that layer will be 1 out of 20
and you know 11 / 20 is five percent so
set the opacity to five percent and then
four layer 2 will want to set the
opacity to two out of 20 or ten percent
so I'll set that one to ten percent and
we just keep on going
you know in this case its multiples of 5
which makes it really easy so make the
third layer 15-percent the fourth layer
twenty percent and we'll just continue
that all the way down until we have each
layer set to like its respective opacity
and by the time you get to the bottom
layer the bottom layer ought to be set
to a hundred percent and so as we do
this you can see that the areas in our
image that we're moving are starting to
blur away
alright so now we have all our layers
averaged together and this is like kind
of a good time we can go ahead and like
zoom in and what will zoom in and at
that example location that we're looking
at originally so now when we look at
this route that we were looking at
earlier we can actually see the detail
in the corrugated metal on the top of
the roof you can see this fine pattern
of lines that wasn't visible previously
and that's sort of the advantage of
using all of your layers each layer is
contributing something to the resolution
of the image
ok so now that we've averaged all of our
layers together we're ready to add some
just final touches to the image we're
going to give it a little bit more
sharpness and the first thing we want to
do is flatten the layers together
so go ahead and select the top layer
here in the layers palette hold down
shift and it's like the bottom layer to
select all the layers and then
right-click and choose flatten image
ok so now that all the layers have been
flattened into a single layer we're
ready to add the last final touches to
the image which is just adding some
sharpening so we're going to use the
smart sharpen filter which is under
filter and then sharpen and then smart
sharpen since the super resolution image
right after averaging tends to be
relatively soft you can actually use a
pretty aggressive amount of sharpening
noise levels are reduced in the super
resolution image so you don't have to
really worry a whole lot about
introducing too much noise of the image
i like to use about three hundred
percent you know maybe maybe between 200
and 300 percent on the sharpening I
think I think 300 actually like pretty
safe so three percent radius of 2 pixels
and I don't usually use any noise
reduction on my images I don't mind a
little bit of grain in there we already
did a really good job at reducing the
grain just by doing the stack so it
doesn't really matter a whole lot
alright so let's go ahead and click ok
alright so now it looks like a
sharpening is done so we're basically
done you might want to do a little bit
of cropping on the edges of the image
you can see where overlap was made i
usually just crop out that part of the
image
yeah we'll lose a little bit of the
original image size you know in this
case but we also gained a whole lot more
by doing super-resolution technique on
it is crop that out and you know so now
you can sort of see the final results
you know we've got a lot of detail on
that roof which was a little problematic
at first you know and like even viewing
this you know just a hundred percent it
still looks really sharp like sharpening
helps we have the artifacts in there
that we sort of expected like this car
that was moving is sort of blurred
and steps and you know this this guy
that was walking down the street
he's also kinda blurry so you know it's
definitely not a perfect technique for
every situation but in in terms of
increasing actual detail it actually
doesn't really really great job so if I
check the image size after dropping it's
currently 8017 pixels wide by 11,000 922
pixels so it's about 95.5 megapixels
which is pretty impressive you know we
started with the 24 megapixel image and
we were able to get an image that was
almost a hundred megapixels out of it so
that's just about it that's the way we
can use adobe photoshop to emulate the
super-resolution technology that's found
in the olympus omd em5 bark too i think
this multi-shot super-resolution
technology is going to find its way into
a lot of different cameras the fact that
we're already seeing olympus produce a
camera that costly thousand dollars i
think it's a real indicator that it's
something that people will want in their
cameras i also expected to become a lot
easier
obviously new cameras will do it in
camera but i also think it will become
much quicker so we won't have the
problems that we normally have with
motion blur in the image like this guy
walking across the street that we can
use really fast shutter speeds for this
tech for this technique and still get
really high resolution and gets i'm
using a 24 megapixel sony a7 mark to it
is a very new camera
I i would consider a very high
resolution sensor it's you know it
produces very very sharp images but if i
wanted to make you know a really huge
print of something you know if I
absolutely needed the you know pixel
keeping level of detail or if I wanted
the high-resolution capability of a
camera like the five dsr I'd probably
consider just trying out this method
first because it doesn't cost you
anything
I pretty much every modern DSLR has a
has a burst mode
it and you know all it all it means is
that you just have to take a few more
photographs at the location that you're
shooting and you can combine these
really high resolution images is it
practical for everything definitely not
i think that its most suitable to static
scenes because it doesn't handle motion
very well but the super-resolution
technology built into cameras like the
m5 mark to also don't handle motion very
well so I honestly think that this is
kind of like a good hold over until the
increase the capability of those cameras
to do super resolution techniques that
are much faster so the next time
somebody brags about the amount of
megapixels that their camera has show
them this video and show them how you
can actually increase the resolution of
your camera beyond what its sensor count
with otherwise indicate once again I
mean Norman i hope you enjoyed this
tutorial if you want to see more
photography videos please subscribe and
check out all of our tutorials reviews
and inspiration on Foton collective .
com