Jump to content


Film to Digital - Dynamic Range


  • Please log in to reply
13 replies to this topic

#1 Daniel J. Ashley-Smith

Daniel J. Ashley-Smith
  • Guests

Posted 26 September 2005 - 05:02 AM

I was thinking, apparently film has almost "infinite" dynamic range, why not:

Scan the negative in normally
Then under expose the negative, and scan in all the burnt out detail
Then over expose and gain all the shadow detail

That way you could have a 20bit digital image easy.

Would that work? Has it been done before?

Thanks,
Dan.
  • 0

#2 Stephen Williams

Stephen Williams
  • Sustaining Members
  • 4708 posts
  • Cinematographer
  • Europe

Posted 26 September 2005 - 05:05 AM

I was thinking, apparently film has almost "infinite" dynamic range, why not:

Scan the negative in normally
Then under expose the negative, and scan in all the burnt out detail
Then over expose and gain all the shadow detail

That way you could have a 20bit digital image easy.

Would that work? Has it been done before?

Thanks,
Dan.

<{POST_SNAPBACK}>



Hi,

Yes its done it does work. Scan at a higher bit depth per channel would help too

Stephen
  • 0

#3 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 19769 posts
  • Cinematographer
  • Los Angeles

Posted 26 September 2005 - 10:17 AM

As I understand it, if the information is on the negative, you don't need to scan it twice, you just scan all the picture information and then compress or stretch the gamma to what looks good to you, unless you want a very flat-looking master scan and make a color-correction version off of that for dailies & editing.

In terms of shooting a shot twice, once underexposed and once overexposed to record more information, of course the problem is you'd have to shoot both versions simultaneously with something like a beam splitter to get them to line-up & match.

There's a difference between doing a low-contrast scan that holds all of the exposure information and doing something that looks good for video formats & displays, which actually requires that you throw away some information to get something that looks normal in contrast -- just like when you make a print for projection, you deliberately throw away some exposure information on the negative in order to get good contrast and rich blacks.

Just depends on what purpose this scan is for -- video display, for a D.I., etc.
  • 0

#4 Dimitrios Koukas

Dimitrios Koukas
  • Sustaining Members
  • 569 posts
  • Cinematographer
  • Athens, Greece, London UK

Posted 26 September 2005 - 11:40 AM

As I understand it, if the information is on the negative, you don't need to scan it twice, you just scan all the picture information and then compress or stretch the gamma to what looks good to you, unless you want a very flat-looking master scan and make a color-correction version off of that for dailies & editing.

In terms of shooting a shot twice, once underexposed and once overexposed to record more information, of course the problem is you'd have to shoot both versions simultaneously with something like a beam splitter to get them to line-up & match.

There's a difference between doing a low-contrast scan that holds all of the exposure information and doing something that looks good for video formats & displays, which actually requires that you throw away some information to get something that looks normal in contrast -- just like when you make a print for projection, you deliberately throw away some exposure information on the negative in order to get good contrast and rich blacks.

Just depends on what purpose this scan is for -- video display, for a D.I., etc.

<{POST_SNAPBACK}>


What would happen if you take two tapes, hard disk media or whatever and then make them two layers in the editing that will blend together as an 100% superimpose?
Dimitrios Koukas
  • 0

#5 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 19769 posts
  • Cinematographer
  • Los Angeles

Posted 26 September 2005 - 12:15 PM

If your scan contains all the exposure information on the negative, I don't see why that's necessary. I'm sure for some effects & color-correction work, being able to adjust highlights separately from shadows is useful, but I believe that is done anyway with a typical 10-bit Cineon log scan.

The main thing to remember is that you often don't want to have the full exposure range visible because it can look too flat. Even with the HD features I shoot, I spend more time adding contrast to the image in post than taking it away. I like contrast I guess...

I assume a 20-bit scan wouldn't have more dynamic range, only finer tonal gradations.
  • 0

#6 John Pytlak RIP

John Pytlak RIP

    (deceased)

  • Sustaining Members
  • 3499 posts
  • Industry Rep
  • Rochester, NY 14650-1922

Posted 26 September 2005 - 01:42 PM

If you look at the sensitometric curve (tone scale) of a modern color negative film, you can see the tremendous latitude of the film:

http://www.kodak.com...5218CamStop.pdf

With a color negative film, separate emulsions (either blended or as separate layers) are used to capture the dark (fast emulsions), middle (mid speed emulsions) and highlight (slow emulsions) portions of the tone scale. One exposure captures the entire range, well over 10 stops.

Your idea of using three separate exposures, but changing exposure to fit a digital sensor's sensitivity works, but not easily for moving images.
  • 0

#7 Saul Pincus

Saul Pincus
  • Basic Members
  • PipPipPip
  • 114 posts
  • Director

Posted 26 September 2005 - 01:44 PM

If your scan contains all the exposure information on the negative, I don't see why that's necessary. I'm sure for some effects & color-correction work, being able to adjust highlights separately from shadows is useful, but I believe that is done anyway with a typical 10-bit Cineon log scan.

The main thing to remember is that you often don't want to have the full exposure range visible because it can look too flat.  Even with the HD features I shoot, I spend more time adding contrast to the image in post than taking it away.  I like contrast I guess...

I assume a 20-bit scan wouldn't have more dynamic range, only finer tonal gradations.

<{POST_SNAPBACK}>


HDR (High Dynamic Range) still photography is very common these days. Usually it's used by visual effects houses when they want to capture the textures or geometry of an existing set, but retain the look established by the DOP. The advantage is that the VFX people can work with something that's been expertly lit by the main unit DOP while maintaining control over manipulation of that image for effects purposes.

It's a little early, but it's only a matter of time before true HDR technology evolves reliably beyond the realm of still photography. Perhaps it already has?

Saul

Edited by Saul Pincus, 26 September 2005 - 01:45 PM.

  • 0

#8 John Pytlak RIP

John Pytlak RIP

    (deceased)

  • Sustaining Members
  • 3499 posts
  • Industry Rep
  • Rochester, NY 14650-1922

Posted 26 September 2005 - 02:00 PM

HDR (High Dynamic Range) still photography is very common these days.  Usually it's used by visual effects houses when they want to capture the textures or geometry of an existing set, but retain the look established by the DOP.  The advantage is that the VFX people can work with something that's been expertly lit by the main unit DOP while maintaining control over manipulation of that image for effects purposes.

It's a little early, but it's only a matter of time before true HDR technology evolves reliably beyond the realm of still photography.  Perhaps it already has?

Saul

<{POST_SNAPBACK}>


The Kodak "Perfect Touch" printing/processing system uses cutting-edge HDR image processing to extend dynamic range:

http://www.kodak.com...ectPB2003.shtml

Image Processing: High-speed algorithms that operate sequentially and together analyze film order characteristics and scene content of each image, to automatically determine the best image enhancement in terms of color accuracy and density balance (overall lightness or darkness). They also bring out detail in the shadow and highlight areas, and optimize sharpness level. In fact, algorithms offer control over each pixel in an image, which is typically scanned at 4.5 million pixels, as well as 4,000 gradations of color and density.


  • 0

#9 Dimitrios Koukas

Dimitrios Koukas
  • Sustaining Members
  • 569 posts
  • Cinematographer
  • Athens, Greece, London UK

Posted 26 September 2005 - 02:13 PM

If your scan contains all the exposure information on the negative, I don't see why that's necessary. I'm sure for some effects & color-correction work, being able to adjust highlights separately from shadows is useful, but I believe that is done anyway with a typical 10-bit Cineon log scan.

The main thing to remember is that you often don't want to have the full exposure range visible because it can look too flat.  Even with the HD features I shoot, I spend more time adding contrast to the image in post than taking it away.  I like contrast I guess...

I assume a 20-bit scan wouldn't have more dynamic range, only finer tonal gradations.

<{POST_SNAPBACK}>


Thanks,
I am crashing the black a bit and tone the highlights for tv adverts, and i this on a Cintel's C-Reality or Ursa Gold, daVinci's 2K or Renaissance 888 color correctors at 4:4:4 , and I know the capabilities, just wondering if there's another option in terms of an experiment and if someone have done anything with a different result. SevenStars here in Greece is a brand new laboratory set-up by Eastman Kodak.

Dimitrios Koukas
  • 0

#10 Saul Pincus

Saul Pincus
  • Basic Members
  • PipPipPip
  • 114 posts
  • Director

Posted 26 September 2005 - 03:44 PM

The Kodak "Perfect Touch" printing/processing system uses cutting-edge HDR image processing to extend dynamic range:

http://www.kodak.com...ectPB2003.shtml

<{POST_SNAPBACK}>


Very interesting demo, John.

I was referring more to how soon we might see HDR capture applied to motion image use; i.e. when it might outclass current CCD technology and render the "lack of dynamic range" argument with regard to HD and other video cameras obsolete.

Of course, you'd then need realtime color presets to monitor the image on set, or for dailies, lest everything look too flat like David points out.

Saul
  • 0

#11 Jaan Shenberger

Jaan Shenberger
  • Basic Members
  • PipPipPipPip
  • 262 posts
  • Director
  • San Francisco

Posted 26 September 2005 - 04:31 PM

actually, film emulsion is kinda the basis from which HDR imaging was created. the purpose of it in visual effects is so that compositing & rendering software can understand that there is more luminance range within a texture map or image than what can be displayed at once. though HDR imagery can far surpass it, this is what a neg already does-- contains a wider luminance range than can be printed/displayed at once.

HDR for digital acquisition would be kinda pointless. an increase in the lattitude of CCDs is all you need, because you have an operator or exposure sensors to decide the general luminance level (aperture). i've seen images that use the entire lattitude of a neg (about 14 stops) and it looks aesthetically awkward and very bland. usually, images look best using 7-10 stops of the neg's lattitude. so as long as they can develop CCDs that can capture 10+ stops, then you're fine. HDR images can contain as many as like 1,000 stops or something crazy like that. those are really only necessary for things like image-based lighting.

btw, i'd suggest any DPs to google "image-based lighting" and read up on it. it will soon make visual effects/live action integration even more economical and "realistic". and your actual on-set lighting will essentially be lighting the CGI elements, rather than have effects people recreate your setups from those chrome ball references. this is going to be the biggest advancement in visual effects since 3D motion tracking.
  • 0

#12 David Cox

David Cox
  • Sustaining Members
  • 323 posts
  • Digital Image Technician
  • london, UK

Posted 29 September 2005 - 01:44 PM

...just going back to the original question, the more bits you use the more storage you need and the slower your systems work to handle the extra data. So yes the theory works but in order to balance quality versus budget, its necessary to work with the least amount of data possible. Thats why image formats such as the 10 bit log format came about to get close to what can otherwise be achieved with 16 bits linear - although that is based on the assumption that you don't want to change the image properties too much from their original.

I guess not shooting everything in 48 bits is the same as not shooting soap operas on 65mm film. At the end of the day, the addition to the budget would not be appreciated by the end user (nor the producer!)

David Cox
Baraka Post Production
London
  • 0

#13 John Pytlak RIP

John Pytlak RIP

    (deceased)

  • Sustaining Members
  • 3499 posts
  • Industry Rep
  • Rochester, NY 14650-1922

Posted 29 September 2005 - 02:10 PM

I guess not shooting everything in 48 bits is the same as not shooting soap operas on 65mm film. At the end of the day, the addition to the budget would not be appreciated by the end user (nor the producer!)

David Cox
Baraka Post Production
London

<{POST_SNAPBACK}>


The nice thing about film is that the color depth and tone scale are not dependent on the format. Larger formats offer better sharpness and less grain, but the other good characteristics of film are format-independent.

Has anyone ever shot a soap opera on 65mm? (Excluding "Cleopatra"? ;) )
  • 0

#14 Phil Rhodes

Phil Rhodes
  • Sustaining Members
  • 11943 posts
  • Other

Posted 29 September 2005 - 04:31 PM

Hi,

Arriscan does this.

In a recent presentation, their people claimed that it was done as part of exciting new sensor technology, or something like that. Actually it's done to cover up a glaring deficiency in the sensor technology, but let's not split hairs, eh?

Phil
  • 0


The Slider

Opal

Ritter Battery

CineLab

Aerial Filmworks

rebotnix Technologies

CineTape

Tai Audio

Media Blackout - Custom Cables and AKS

Gamma Ray Digital Inc

Technodolly

Abel Cine

Metropolis Post

Paralinx LLC

Rig Wheels Passport

FJS International, LLC

Willys Widgets

Wooden Camera

Glidecam

Visual Products

Broadcast Solutions Inc

Ritter Battery

Metropolis Post

Opal

CineLab

Gamma Ray Digital Inc

FJS International, LLC

Wooden Camera

Glidecam

Rig Wheels Passport

The Slider

Tai Audio

Paralinx LLC

Abel Cine

Willys Widgets

rebotnix Technologies

Aerial Filmworks

CineTape

Technodolly

Broadcast Solutions Inc

Visual Products

Media Blackout - Custom Cables and AKS