How is dynamic range actually measured?

[Walter Graff]

Keith, don't take things personally. The smile means I was joking. The joke had nothing to do with you or your idea. It was a fun stab at RED that sold itself as a cheap camera but after you add it all up, it's not so cheap or at least as cheap as a lot of folks on the web believed. That is not bad, just interesting how everyone goes around saying it's a $18k camera.

 

Your idea is fun but frankly, I don't see a need for it, nor it doing anyhting more than what could be done with a proper chart and scope.

 

I do find this infatuation with tools over talent fascinating. The marketing parts of the manufactures has really got the latest generation of people completely spinning their wheels over the hammer rather than the carpenter trying to discover some nirvana in a camera rather than the talent behind it. That is not directed at you either Keith, just a whole generation that sees good footage and doesn't ask the obvious question of who shot it, rather they say what camera did they use.

[Keith Walters]

Keith, don't take things personally. The smile means I was joking. The joke had nothing to do with you or your idea. It was a fun stab at RED that sold itself as a cheap camera but after you add it all up, it's not so cheap or at least as cheap as a lot of folks on the web believed. That is not bad, just interesting how everyone goes around saying it's a $18k camera.

Sorry, at work I have the "show pictures" box unclicked on Internet Explorer, because some sites (including this one) make you download an inordinate number of files before you get to read anything interesting!

Your idea is fun but frankly, I don't see a need for it, nor it doing anyhting more than what could be done with a proper chart and scope.

 

Yes, but I don't have a "proper chart" and scope, andI don't feel like spending all that money if something I can build for $30 will do the same job. Apart from this, very few people in the film industry have any idea what a waveform monitor actually does or what those lines on the screen mean. However, everybody should be understand what my proposed gadget will do.

 

I will probably be able to test it with various Sony cameras and a RED, getting it in front of a Genesis will be a challenge!

I do find this infatuation with tools over talent fascinating. The marketing parts of the manufactures has really got the latest generation of people completely spinning their wheels over the hammer rather than the carpenter trying to discover some nirvana in a camera rather than the talent behind it. That is not directed at you either Keith, just a whole generation that sees good footage and doesn't ask the obvious question of who shot it, rather they say what camera did they use.

Well of course. You can't buy talent with a credit card, but technology is always available on easy terms:-)

Edited February 15, 2008 by Keith Walters

[John Sprung]

The colour of the LEDs is not important.

I just had the same thought. What if you were to use, say, green LED's? You'd be able to see where you max out response in the green channel/layer, and also where you start to get so bright that there's significant crosstalk into red and blue.

 

 

 

-- J.S.

[Keith Walters]

I just had the same thought. What if you were to use, say, green LED's? You'd be able to see where you max out response in the green channel/layer, and also where you start to get so bright that there's significant crosstalk into red and blue.

 

 

 

-- J.S.

I would imagine the colours of the bayer mask would have been "tuned" so that they all go into overload at more or less the same time.

 

All single-chip colour cameras have some variant of the "chroma limiter" which basically shuts down the colour processing as soon as any one of the colour channels approaches overload. Otherwise highlights would be colourful indeed!

 

It's quite common on cheaper digital still cameras for parts of overloaded areas to have a bluish tinge, because silicon is least sensitive to blue light, and so it's the last colour to overload. I think this is regarded as more acceptable, because it often happens with objects sillhouetted against the sky, and a bluish tinge there does not look so out of place.

[John Sprung]

I would imagine the colours of the bayer mask would have been "tuned" so that they all go into overload at more or less the same time.

 

All single-chip colour cameras have some variant of the "chroma limiter" which basically shuts down the colour processing as soon as any one of the colour channels approaches overload. Otherwise highlights would be colourful indeed!

 

It's quite common on cheaper digital still cameras for parts of overloaded areas to have a bluish tinge, because silicon is least sensitive to blue light, and so it's the last colour to overload. I think this is regarded as more acceptable, because it often happens with objects sillhouetted against the sky, and a bluish tinge there does not look so out of place.

OK, they'd overload in unison with white light, say D65 or whatever. But suppose we give them green only. Shouldn't green max out first?

 

Another thought -- If in fact you can sync to the camera, you only need one LED, and can eliminate the LED matching issue. Make your device output one blink per frame, modulating both current and pulse width. It could start with the maximum exposure, and go down a stop with each successive frame until you have very little, then go completely dark for enough frames to make a total of 24. Let it repeat that cycle over and over. With film, you could put the negative on a densitometer. With data, you could look at the frames as individual files.

 

Yet another idea -- Eliminate the use of a lens on the camera, and just build the one-LED version with a PL mount. That might not matter much technically, but as a marketing idea it buys you plenty. There's a general mindset that anything with a PL mount on it must be worth thousands of dollars.... ;-)

 

 

 

-- J.S.

[Stephen Williams]

There's a general mindset that anything with a PL mount on it must be worth thousands of dollars.... ;-)

 

-- J.S.

 

Hi John,

 

$6500 if the glass is Sigma!

 

Stephen

[Keith Walters]

Another thought -- If in fact you can sync to the camera, you only need one LED, and can eliminate the LED matching issue. Make your device output one blink per frame, modulating both current and pulse width. It could start with the maximum exposure, and go down a stop with each successive frame until you have very little, then go completely dark for enough frames to make a total of 24. Let it repeat that cycle over and over. With film, you could put the negative on a densitometer. With data, you could look at the frames as individual files.

Yes but cameras like the Genesis of the RED do not have a sync output that directly relates to the actual frame rate of the sensor. It's ironic that I have built 24 frame video converters that can be easily synced to popular models of film camera, but not to most HD video cameras! I was actually called on to help figure out a way to get a sync signal out of the Genesis for "Superman Returns" in the early days ,but as far as I was able to determine there was no way of doing it.

 

Interesting development in the White LED department. The White LEDs used in a lot of the latest model Christmas lights seem to be extremely well matched for brightness, as well as being dirt cheap!

 

Yet another idea -- Eliminate the use of a lens on the camera, and just build the one-LED version with a PL mount. That might not matter much technically, but as a marketing idea it buys you plenty. There's a general mindset that anything with a PL mount on it must be worth thousands of dollars.... ;-)

-- J.S.

I have actually done exactly that, with RGB LEDs, for setting up colour video taps. Not only can you have stable 3.2K and 5.6K references at the flick of a switch, you can do real quantative sensitivity measurements, and "fine tune" the circuitry to get the last ounce of noise performance.

 

The mount does not have to be anything special since there is no focussing involved.

[John Brawley]

r "Superman Returns" in the early days ,but as far as I was able to determine there was no way of doing it.

 

 

Couldn't you just use the black burst from the actual output signal ?

 

jb

[David Auner aac]

Couldn't you just use the black burst from the actual output signal ?

 

That was my first thought too! Wouldn't black burst be enough?

 

Cheers, Dave

[Keith Walters]

That was my first thought too! Wouldn't black burst be enough?

 

Cheers, Dave

"Black Burst"?

On an HD camera?

 

HD does not have a colour burst, and the sync pulses on the HD component outputs are nothing like those of ordinary composite video. The Digital outputs are totally off limits as far as I am concerned, so I am not sure what you are talking about.

 

Getting my setup to work would have involved borrowing a very expensive HD camera for a while so I could work out what was required to sync to it, and that was simply not an option. Now I could just buy a sub-$100 DVD player with HDMI upconversion and use the HD component out of that for my experiments, but the need for 24 frame video seems to have died off with the advent of extremely cheap LCD TVs!

 

Black burst or just plain composite video from SD cameras works a treat on my converters, both 25 and 30 fps. But in my experience with Broadcast HD cameras, if they have a Standard Definition output at all, its frame sync phase bears no definite relationship to the frame rate of the CCD chips. This is particularly so when shooting 24p, since there is no such thing as 24 frame PAL (except for what comes out of my converter boxes of course.)

 

Frankly I don't know enough about the Genesis, and I've never been able to get close enough to one to find out much. After Superman Returns it stopped being flavour of the month here, and PV never responed to our requests for information.

 

Actually the whole 24 frame video saga is somewhat ironic.

I started research on this about ten years ago after a spate of frantic phone calls from a variety of people who had just discovered on-set that you can't just film a PAL TV screen at 24fps.

 

I found a way to modify a cheap video "system converter" (ie one mainly designed for converting VHS tapes) to make it produce 24 frame video which would work on unmodified PAL colour TV sets.

 

However, because most people in this country seemed unaware of the problem, I wrote a little booklet that explained what the problem was, the sometimes tortuous means needed to overcome it, and described how my $200 a day (no operator needed) converter would eliminate all these problems in one go.

But they hardly got any business, and upon investigation I found that by simply alerting people to the problem, I made it largely a non-issue! They were able to either simply avoid the situations altogether, or fall back on the old trick of shooting the screen at 25fps with the speaking actors always having their backs to the camera, and dubbing in their dialogue later.

 

I did get enough jobs to pay for the initial investment but it was never the license to print money I had hoped. I still have a "museum" of working "vintage" 1970s TV sets, but I have never been able to rent out a single one of those. People usually only find out I have them after spending a lot of time and money trying to find a TV repair shop prepared to fix up some old banger they have found by the roadside;-)

[John Sprung]

The mount does not have to be anything special since there is no focussing involved.

Right, but *they* don't know that. That's where you can make some money..... ;-)

 

 

 

 

-- J.S.

[Keith Walters]

Here's a challenge for all you Digital Cinematography enthusiasts! Today I was asked to help out a photographer who was trying to take the photographs for this year's Christmas lights Flyer. (LED Xmas lights are all the rage now. The quality has dramatically improved over the past year, and the price is dropping all the time).

 

They had samples of this year's range set up hung from the ceiling and they look stunning. The effect is quite different from the usual incandescent lights, as the LEDs are highly monochromatic and pin sharp, like stars in a moonless country sky.

 

But no matter what digital camera he used or how he used it, he could not get his pictures to look anything like the real thing!

 

This photo is one I took back in the test lab with a much cheaper Kodak 5 Megapixel compact camera, but his were hardly any better:

 

(They may not look so bad to you because you are not seeing the original image).

 

I have taken a few shots with the 35mm film camera we use for archival purposes, but I won't be seeing the result until the roll of film is finished! However I am strongly suggesting he dust off his trusty old Canon!

 

So, this could be a very interesting "practical" test of digital cameras (moving or otherwise); how does it cope with LED christmas lights?

[John Sprung]

The effect is quite different from the usual incandescent lights, as the LEDs are highly monochromatic ....

That may be the problem. If they're highly monochromatic, they may lie outside the gamuts of both film and electronic cameras. Do you have spectral energy distribution curves for them? I remember finding tabulated values for the CIE 1931 big XYZ functions. With that you could find where they land in CIE little (x,y) space.

 

One thing I did for fun during the strike is cut and paste the spectral and Planckian loci for 1931 2 degree (x,y) into AutoCad format, along with about a dozen sets of primaries and white points for various systems such as 709, DCI, SMPTE-C, etc. I can send you the .DWG file if you'd like.

 

 

 

 

-- J.S.

[Guest Glen Alexander]

One version of this is a box with a row 15 white LEDS, which have been carefully set up so that the brightness of each LED is exactly 50% of the brightness of its neighbour to the left.

 

If the leftmost (brightest) LED is regarded as 0 stops attenuation, then each LED represents another stop darker, until you get to 14 stops (1/16,384th as bright as the first LED)

 

Bonjour Keith,

 

I'm back.. ha ha ha

 

How do you know each is 50% of the neighbor? Did you use a labsphere? Are you using reflectors behind the sources?

 

Why LED's? They're cheap but they suck, their spectrum is garbage. What does the graph of the current draw versus illumination look like?

 

Using 15 incoherent light sources with garbage spectrums, yuk, no wonder you can't get a decent dynamic range.

 

Why get some cheap laser pens, reflector plate, ye old chunk of wood or plate and set up a slit diffraction or lens system.

 

Numerically you are FFT-ing a summation of logs, the diffraction pattern should indicate the range.