The Future of Digital Photography

[mike]

8K, apparently...

and a quick google from there brings me to the brilliant Red website, from their 'epic dragon' camera schpeil:

6K resolution translates to over 19 MP, packing the same detail you expect from your DSLR into a cinema camera. When you can capture up to 100 frames per second at full resolution, you get 100 chances per second to capture the perfect picture. Every still is raw and Adobe Photoshop compatible, which means your workflow doesn’t need to change—even if technology does. The RED DRAGON sensor blurs the line between motion and still cameras, giving you the best of both worlds.

when the even-better-named 'weapon dragon' appears, my SLRs will be obsolete.  http://www.red.com/products/weapon-dragon

how much are they?  is it more than the eight pounds 32p I've got on my desk?

[Charlotte]

However much it costs, you're going to need to invest a shitload more in the back end tech to manage files that big.  Can you imagine how enormous a NAS you're going to need if you shoot three or four times a week with one of those things?

[Jaded]

Plus I struggle to pick the best shot from a burst of 8. Imagine having to do that from 800.

[LEE]

Plus I struggle to pick the best shot from a burst of 8. Imagine having to do that from 800.

It always comes down to pointing the camera at the right thing, with the right lens, the right aperture, the right shutter speed and in the right light.  A million bad photos is no better than 1 bad photo.

Sports photographers are bound to move to such technology, I understand that, it will allow them to capture amazing moments simply by videoing an athlete for the duration of the event and saving the 100 or so frames either side of any exciting bits.

Almost nothing I do with my camera would benefit from more than, say, 5fps.  Typically I just need 3 bracketed shots in fairly rapid succession.

4K video will be nice on my DSLR I guess.  I'm happy with 1080p at a cinematic-ish 25fps & 1/50th though.  I'm not chasing ultra "clinical" video, it's not a look I want (unless I go filming Polar Bears with David Attenborough).

[frankly frankie]

Pre-capture has been around for at least 10 years (in some Olympus compacts, for example), but oddly has never really taken off as a selling point, and has taken some time to penetrate to higher-end gear.  4K-capable cameras can certainly take that to the next level.
Now, the latest Panasonics offer a sort of very fast focus-bracketing mode, such that you choose your plane of focus in post - or end up with an image with foreground and infinity in focus, and the middle distance soft.
But all of this takes industrial levels of time and resources, to make the most of it.

The low light ability of modern camaras is a fantastic advance.  In film days anything faster than 64 (ASA, as it then was, now ISO) would be seen as a compromise.
I think future improvements could lead to smaller and lighter lenses - I'd really like to see a move away from the mindset that worships 'fast glass' and big f-off willy-waving lenses regardless of expense - if the same or better results can be obtained with smaller, maybe plastic, optics, and a shedload of processor power.  Although I recently bought a Pana GX-7 (because it's out of production and the replacement model is too big) even that is really a bit bigger than my ideal - for me a camera has to be pocketable, or the only step up from that is 'expedition quality'.  Er - unfortunately I don't carry a smartphone. 

[LEE]

I think future improvements could lead to smaller and lighter lenses - I'd really like to see a move away from the mindset that worships 'fast glass' and big f-off willy-waving lenses regardless of expense - if the same or better results can be obtained with smaller, maybe plastic, optics, and a shedload of processor power.

One thing that smartphones show is that you can create a pretty good image using tiny lenses. 

This is the "big lens" (not the selfie lens on the front) on my Motorola MOTO G. (The yellowish dot is the "Flash-gun")



And this is a 6Mp image from it.



BIGGER <<<<<

It's very sharp and would be fine for a 10x8" print.  It's miraculous really.

From the camera-phone thread here are two photos taken on 2 of my mobile phones, separated by 9 years of evolution.




Can they take the miraculous quality of a pin-hole sized lens and scale the miracle up to a larger camera lens?  Maybe the "miracle" is just down to its tiny size and it can't scale (an actual pin-hole camera does a miraculous job without a lens and but you can't scale the hole up to allow more light in.  The miracle doesn't scale).

Maybe there's a reason why a Canon 300mm F2.8 costs £5,000.

[Jaded]

Depth of field.

[clarion]

I've been working on the future of digital photography by taking pictures with my 'Less than a tenner' film SLR - OM40.  Mirror like a safe door slamming, but a perfectly usable machine.  With my 35/2.8, it's nice and clear to focus, though the viewfinder doesn't match up to an OM-1.

[LEE]

Depth of field.

...or lack of.

Software is getting better at spoofing shallow DoF using multiple exposures.

My phone does it (not all that well) and my Canon S120 compact does it (a bit more convincingly) but neither come close to my 85mm f1.8 wide open right now.

I expect it's only a matter of time though. The software just needs to know the distance to all the subject matter in the image.  In theory you could adjust the DoF from 0-infinity and "dial in" the sort of Bokeh you wanted. 
It strikes me as a function of Moore's law, meaning we'll crack it quicker than we think.

Built-in Flash is one area that is generally just awful.  It's rare to see any image that hasn't been totally ruined by a built-in Flash. 
I suppose huge ISO is the only answer.  That may need a sensor technology breakthrough in order to get low-noise at high ISO, on smallish sensors.
Maybe they can crack it with software and better noise reduction techniques.

For example, my Go Pro software can create super-smooth Slow-Mo (FLUX) by interpolating new frames into the original frames, frames that never existed but were generated using the previous and next frames. Maybe software can take a multiple exposure and, assuming noise is not identical in every shot, figure out what is noise and what is not.

[geraldc]

As people who take photos of things, we're at a disadvantage. The majority of photos taken are on phones. The majority of photos taken are by girls. The majority of phone photos taken by girls are selfies.

http://www.telegraph.co.uk/technology/news/10788116/Women-significantly-more-likely-to-take-selfies-than-men.html

A lot of the developments with cameras won't be for us, unless someone here really likes taking selfies.

[LEE]

Depth of field.

...or lack of.

Software is getting better at spoofing shallow DoF using multiple exposures.

This is taken from a tripod using maximum aperture on my compact camera (f/4.5) to get maximum "Bokeh" effect at mid-zoom.
The camera has a max apertaure of f/1.8 but only at a very wide 24mm equivalent.
You can see slight bokeh/blurry background but not much considering the proximity of Mr Monkey. It's a physical limitation imposed by the (small) sensor size.

The following image is using "Background Blur" Scene Mode.  It takes 2 shots in rapid succession and merges them.
It's actually pretty convincing and uses same f/4.5 and same focal length zoom.
I know from experience that "fly-away hair" defeats its ability to separate the 2 layers but it still does a decent job at separating distracting backgrounds.

The difference in colour balance is down to the fact that image #1 is shot in RAW and converted to jpg with no adjustment.  Image #2 is a straight jpg (a limitation of the Scene Mode).
>>>>> Bigger Glass & Bigger Sensor >>>>Here is f/4.5 on a crop sensor DSLR (Hand held under terrible flouro light)

Here's same camera but with 50mm wide open at f/1.4 (so fairly "Big Glass").  The DoF doesn't even encompass the whole of Mr Monkey's head.


I've got to say that the "Background Blur" on my compact camera has done an admirable job when you compare f/4.5 equivalents.

It doesn't compare to a big piece of glass at f/1.4 though.  I expect that's just a matter of time and processing power though.

I must try my Phone camera's ability to spoof a shallow depth of field.  Watch this space.

>>> Phone Camera >>> NO EFFECTS


>>> Phone Camera >>> EFFECT "Lens Blur"


Opinions? 

I think it's not too bad considering the "infinite depth of field" tendancy of tiny phone-camera sensors.

[fruitcake]

That's a very interesting demonstration, showing how the compact and the phone use software to create effects previously achievable only with hardware. As you say, the results are convincing enough.

[Kim]

The fashion for bokeh is bound to wear off eventually.  As will the fashion for film artefacts in cinema.

[clarion]

Thanks for doing that, LEE.  Very interesting comparison.

[fruitcake]

Another thing that's clear from LEE's post is how close we are to focus-pulling in post: if the compact and the phone use depth data to blur the background in-camera, you'd just need that data to be stored with the image and for the post software to access it, for the user to move the focal plane back and forth after the photograph has been taken.

There are all sorts of other ways the depth data could be applied too.

I imagine a challenge for camera makers has been in getting the depth measurements consistently accurate.

The fashion for bokeh is bound to wear off eventually.  As will the fashion for film artefacts in cinema.

Tastes and expectations change. I guess DoF/background-blur is really a photographer's way of directing the viewer's attention, and other ways may develop.

[LEE]

The fashion for bokeh is bound to wear off eventually.  As will the fashion for film artefacts in cinema.

The fashion for shallow depth of field won't wear off though, it's a key tool in isolating a subject.

What's clear is that it won't be dependent on expensive glass in future, you'll be able to "dial it in" as needed.

It's not difficult to imagine a camera taking, say, 5 images, rather than just 2 and establishing very detailed distance/DoF data.

Interestingly (or perhaps not) my camera uses a different technique to my phone.

1) Camera - Takes 2 identical images in rapid succession.  I imagine is pulls focus between the two images. You hardly notice the 2 images being taken.

2) Phone - You take image #1 and then you follow a a prompt to slowly raise the "camera" a few inches until it takes image #2. It seems it uses the relative movements between near & far objects to calculate distance (because it wouldn't be able to detect the tiny differences in focus).  It's a bit of a faff, quite manual, and involves a few rejected images.
I think they could easily make it seamless by using 2 lenses (to build in the "rangefinding" it seems to use).
You may as well incorporate it into 3D imaging technology if you are going to add a 2nd lens.

[andrew_s]

Another thing that's clear from LEE's post is how close we are to focus-pulling in post: if the compact and the phone use depth data to blur the background in-camera, you'd just need that data to be stored with the image and for the post software to access it, for the user to move the focal plane back and forth after the photograph has been taken.

There is no actual depth data.
You can take a photo focussed on something close, and one at infinity, and compare the two to infer from the known DoF characteristics how far off something was likely to have been, and do the same thing with a 3rd photo at or near closest focus for near field blur.
Having got some estimate of distances, you can then add appropriate levels of extra blur to appropriate parts of the image, but deblurring something that was originally fuzzy to get apparent focus at a particular point is a whole extra ball game.

Post-exposure focus is available though, but it requires dedicated hardware. It works by capturing light direction as well as brightness.
https://illum.lytro.com/illum
(price between approx. $425 and £700, depending where you buy)

Another option is to just take a whole sequence of photos at different focal points, and select from them afterwards, combining in extra blur from photos taken the "wrong" side of the selected focal point. Similar is done in macro work for extra depth of field (focus stacking). A potential problem is that a great many lenses have different focal lengths (and thus field of view) at different focal distances

[fruitcake]

OK. I see how it works now. I guess the compact's method (taking two images and creating a composite) seems more robust and should work for anything where the subject isn't moving or the lighting changing. The camera-phone method (requiring a change of position) seems heavily reliant both on the user's action and on nothing changing during the interval, so I can see how that generates dud images.

I still reckon camera-makers must be working on capturing a depth map, layering that on a single colour map then allowing the user to interact.

[Googles.]

I see Samsung have played with a double sensor, one for visible light, the other for depth.
http://gizmodo.com/5889363/new-camera-sensor-captures-images-and-depth-data-at-the-same-time

And there's some kind of depth camera which enables gesture control in Xbox kinect. 

[Googles.]

Kinect uses an infra-red projector and sensor to build depth maps, so it suffers less ambient light interference.
http://www.wired.com/2010/11/tonights-release-xbox-kinect-how-does-it-work/

[rogerzilla]

On-camera flash is actually good in black-and-white.  It gives an instant, and very punchy, paparazzi effect which is ideal for parties.  You can bounce the flash but it looks crap for that particular kind of shot.

Bonus: red-eye is a lot less of a problem in b/w.

[Exit Stage Left]

I quite like the idea of post-production Bokeh, having looked through a lot of blurry video and stills. Shooting shallow depth of field on the run is a hit and miss affair, half the stuff you get has to be binned. Shooting with two big sensor cameras in low light must be a nightmare when you come to review the stuff, but you can always use a cutaway from the drone, with its fixed-focus plastic fisheye lens.

[Gattopardo]

I hate the word bokeh it grates.

How much was possible in paper development that is easier with technology nower days?

Interesting shots Lee, thank you.

[LEE]

I hate the word bokeh it grates.

How much was possible in paper development that is easier with technology nower days?

Interesting shots Lee, thank you.

I actually swore, in a previous thread, I'd never use the word "bokeh" again.  However it saves me typing "the actual quality and look of the background blur, especially on highlights", because "bokeh" isn't directly referring to DoF.

Bokeh is something that could be managed with software more easily than creating shallow DoF.  Kudos is always given to lenses which create bokeh without revealing any lens diaphragm blades (those pentagons highlights you see when a cheap 5-bladed lens blurs the background).  Software could give you "cheap" bokeh or "expensive" bokeh and anything in between.

"Buttery Smooth" seems to be the ISO Standard for classifying the best "Bokeh".

I'll see what the "Background Blur" software can make of Xmas tree lights this weekend.

As for paper development I'm so long out of swishing paper around in developer and fixer that I don't know how it compares.  I know that many photographers prefer the depth of a chemically developed print on silver-based papers but I'm really happy with professionally (inkjet?) printed digital images.

I honestly never loved anything quite as much as seeing an image start to appear on paper, in my (parents' bathroom) darkroom, but I definitely wouldn't go back to that, not even for nostalgic reasons*

*It would probably scare the new residents for one.

[Kim]

I honestly never loved anything quite as much as seeing an image start to appear on paper, in my (parents' bathroom) darkroom, but I definitely wouldn't go back to that, not even for nostalgic reasons*

It's the sort of thing that everyone should do once.   

[LEE]

On the subject of Software vs "Big Glass" trying to do something nice with background highlights (OK, I'll say it... "Bokeh").

I had a lunchtime portrait session with Monkey. Xmas lights deliberately in the background (though I had to move back & forth to fill the frame with Monkey depending on what focal length lens I was using. The MOTO G phone is very wide angle so I had to get close).
Note. The relative size of the Xmas tree to Monkey gives you an idea of the relative focal lengths.

MOTO G phone - "Lens Blur" effect

Canon S120 compact camera - "Background Blur" effect (f/4.5)

Canon EOS 100D DSLR (40mm f/2.8 @ f/4.5) - no effects, just glass.

Canon EOS 100D DSLR (50mm f/1.4 @f/4.5)

Canon EOS 100D DSLR (50mm f/1.4 @f/22) Small apertures tend to give you a "star-burst" effect.

Canon EOS 100D DSLR (50mm f/1.4 @f/1.4)


Finally, using a "portrait lens", 85mm f/1.8 @ f/1.8.
On my crop sensor DLSR it gives a 35mm equivalent of a 136mm lens which is just sweet for portraits with a blurred background.


Although the phone and compact do an admirable job they really don't tempt me away from a larger sensor and bigger glass in this situation.

[DDCyclist]

He'd have looked much better in the flames from the gas hob.