when will digital be better?

[Frank Granovski]

When will digital be better/higher resolution than film?

The transition from chemical (film-based) photography to digital imaging is a fact. We may deplore this phenomenon, but progress cannot be halted. The technical quality and impact of photographs is still superior to the results that can be generated form the digital process. But the margins are thin and are nullified by the convenience and real time experiences of the digital way of making pictures.

See: http://www.imx.nl

[Wayne Morellini]

I have been sent official documentation, from an official source, on one of the latest, most anticipated, Digital Still camera sensors, also high resolution video capable. Although it doesn't seem to be strictly confidential, I am still treating it in a somewhat confidential manner. I am also familiar with the specs and features of many video sensors out there as well. I can say a few things about such a time coming.

Although I believe that no sensor can cover the full performance range of film in one chip yet (not having concrete facts on the performance range of modern film stocks, and the human eye) individual sensors can now out perform individual film stocks in various ways. This is because you can change in between individual film stocks, while on digital you are usually left with one unchangeable sensor. But there are compromises.

If you look at the modern features of sensors you can see how it does this. Cheap sensors that advertise a latitude of over 20 stops are now available. They use techniques called multi-slope, and autobrite. One is where multiple samples from each pixel can be taken to extend latitude, the other, I think is where the gain of each pixel is adjusted to boost light sensitivity or reduce the charge from the light coming into the pixel. Pixels have a limited range of charge they can store, often proportional to the size, and these techniques can clear away that charge and make room for more, allowing the pixel to work over a higher latitude, and even adjust the gamma curve received to me more film like. The problem with this is that these techniques compromise other parts of film quality, so require careful attention to setup. I have seen colour shift, and washout introduced by these schemes, but much better then without them.

The amount of colour detail and saturation:
On the basis of color, there is a lack, and it is enhanced by processing, but a number of film stock also need to go through a similar process.

Sensors have an advantage is that they are recording average of photons hitting a sensor, with some electrical noise error, there is no intervening variable chemical process which is colour range limited, while being subject to the usual grain and error.

If you look at single chip 3 colour prism alternative sensors you get 3 colours recorded per pixels, without grain. There are two on the market I am aware of, and Fuji is set for a new more organic one next year, which knowing organics might be even more sensitive then modern regular film stock. Currently there is a problem with color noise, with the pre-existing 3 colour alternative single chip, and we wait to see if the new ones are any better.

Where normal digital sensors still do fall down is in terms of sensitivity. There is a max range (under ideal circumstances only) of 96 db (around 16 stops) of sensitivity that has been reported for the Altasens sensor in the year before last. Even though sensors are approaching 70-90% conversion of light photons to electrical electrons, in their peak colour performance regions, of the theoretical max of 100%, this figure is before all sorts of electrical noise introduced by the electrical circuit has obscured a level of the most sensitive detail. So even though the sensor might be able to cover a broader latitude at it's lowest sensitivity, it does not reach down there. However observations a couple of years ago have shown that it is possible to move two electrons with one photon, so there is possible room for more performance in the theoretical maximum (which probably explained whom they got within 90% of the old maximum). Sensors with organic light gathering mechanisms could be one method to do this, with less electrical noise (but you could make a film using these to).


Another area they do fall down is resolution. The higher the resolution of the sensor the less performance it has, because smaller pixel sizes limit the latitude, sensitivity, and colour response. Also the smaller pixels get, the larger the circuits around the pads the pixels use to sens get, causing some light to miss the pad and hit them, which results in a fly screen like affect on fine details and lines of an image. This sometimes causes them to flash, and can be processed down, or out.

But film has grain. Even though sensor resolution might seem to be limited compared to film, and even though it is tempting to judge resolution by film grain size, a grain does not represent the full colour range of the image detail it is recording, the colour detail is more properly represented by a group of grains, and more again to compensate for grain noise, while monochromatic details might be see able within this. So the resolution for sensors is not as bad as it may first seem, unless you are prepared to use, and light, for the most finest grain film (that also effects the films overall performance). Taking this all into consideration, even thought the performance of SHD sensors might not be as great as HD or SD sensors, they are approaching levels where is usable compared to individual film stock. Ultra HD sensors (think resolution approaching on screen film grain size approaching Imax) are probably still a while off this level of performance, unless you are willing to pay for a expensive large format sensor.


Summary:

So there you have latitude, colour accuracy, sensitivity, and resolution. You can probably beat many stocks when you average these performance considerations out, but it will be some time before you can beat changeable film stocks. One suggestion I have made in times past, is to have swappable sensors, so that when you have different light levels, latitude ranges, or desire a different performance, you can hit a button, and the current sensor is moved out of the way, and a more suitable sensor is then move into it's place (this absolute beats changing film). For most normal filming (using ND filters as well) this should produce the desired results that you would normally get with a range of film stocks at a fraction of the price and inconvenience. You, however, will still not be able to go as low light, as sensitive, or a as high resolution with the same performance as with film. But then it doesn't matter that much for non specialised work.

However, it is totally possible that a sensor one day could be made that will totally outperform film, by using new mechanisms, which I also have been theorising.

[Kevin Shaw]

In practical terms, digital photography surpassed film photography some time ago, as evidenced by the widespread switch to digital by both consumers and professionals. There may still be some ways in which film is artistically preferable, but it's pretty obvious that most people no longer care and film photography is on its way out for anything other than specialized purposes.

As far as video is concerned, film went out decades ago for most users and is probably on the verge of being replaced for movie production within the next few years. Consider cameras like the SI-1920 or Red for ~$20K and that puts a real challenge to film for anything but the highest-end projects.

[Wayne Morellini]

That is the crux of the matter, TV was good for realtime, and for the quick but nasty. But most TV fiction productions used Film for years, because it was the only one for quality, and the movie industry still uses it for the moment, but things are changing because quality is coming up to more acceptable feature filming levels.

An interesting thing, is we have some old cop shows from the 60's/ early 70's, on here late night, that I think were shot on original B&W video, quiet film like, very nice.

Of coincidence, that two electrons from one photon rise in Quantum Efficiency I talked about, there is an article on it in the new New Scientist:

http://www.newscientist.com/article/mg19025531.600.html

[Frank Granovski]

Red? Isn't that vapourware?

[Kevin Shaw]

Originally Posted by Frank Granovski

Red? Isn't that vapourware?

For now yes, but it shows what may be possible soon at a relatively reasonable price. Digital video has already replaced film for just about everything but big-budget movies, and even George Lucas is shooting digital now. The writing is on the wall, and film of any kind is becoming an anachronism.

[Frank Granovski]

Funny, all the film houses here are still using film. :-)

[Wayne Morellini]

It isn't vapourware until it doesn't get designed into a prototype, it doesn't get made, and it disappears in a puff of vapor ;)

[Frank Granovski]

Do I hear the fat lady singing?

[Wayne Morellini]

Vapor means it had no substance. I just see so many people around here trying to kill products before they even come to market, calling them all vaporware. These sorts of products require significant efforts, significant inputs, and significant development. These sorts of products makes the yearly wage of people knocking it, look like vaporware, even their long term incomes.

[David Tamés]

Originally Posted by Frank Granovski

Red? Isn't that vapourware?

I don't think so, but a very optimistic company and vision, yes, check out this thread: http://www.dvinfo.net/conf/showthread.php?t=66620 as well as this one: http://www.dvinfo.net/conf/showthread.php?t=56300, and by the way, Red is not the only camera company in the "digital cinema" camera space, there's Panavision with their Genesis, Arri with their D20, Dalsa, Kinetta, and one thing you can say, this is where things are headed...

[Nate Weaver]

Originally Posted by Wayne Morellini

and the movie industry still uses it for the moment, but things are changing because quality is coming up to more acceptable feature filming levels.

...and the music video industry, and the commercial industry.

I'm no film zealot, but there's still plenty of film production going on in my neighborhood.

[Wayne Morellini]

True enough for the moment, but the movie industry was the one that everybody could agree on.

[Frank Granovski]

David, thanks for the links 4 those 2 threads.

Yup. All the movie houses here in Vancouver still roll da reel as far as I know.

[John DeLuca]

This is an interesting topic I discussed with my lab owner many, many times over. A quality slide and high end scanner will hold more color gradation and saturation than digital. Slides will not "blob up" colors/saturation like the D2x's color mode 3 for example.

Pumping up saturation in post will just add "dirty water" and blob up/clip color/saturation gradation. Slides still have up to 20points more saturation/color gradation without clipping vs digital.

-John