Erland
Member
Hi!
Since I started to scan my images I have gone through a Plustek 7300, to a Agfa Duoscan, to now using my girlfriends Pentas KS-1 with a macro.
The plustek really had Amazing detail, can't complain about that, however the dmax and dmin was awful.
The Duoscan had low-low-low resolution (1000dpi) but the feeling of the output was Amazing.
But now, scanning with a dslr, the output is both Sharp and has a wonderful feeling about them.
However, I found something yesterday that got me thinking. I shot a wedding last Winter, and the church, pretty dark, had big Bright windows and the resulting contrast between dark and light Went beyond what my Fuji X100 could capture to retain both highlights and shadow detail. But when I scan my tri-x from the same lighting conditions, I have detail in both. BUT! the scan's histogram does not clip at either the shadows nor the highligts, so it got me thinking if film can capture more dynamic range, but "compress" it to a smaller range of tones?
I am at work, so I can't show any screencaptures.
Kind Regards
Marcus
Since I started to scan my images I have gone through a Plustek 7300, to a Agfa Duoscan, to now using my girlfriends Pentas KS-1 with a macro.
The plustek really had Amazing detail, can't complain about that, however the dmax and dmin was awful.
The Duoscan had low-low-low resolution (1000dpi) but the feeling of the output was Amazing.
But now, scanning with a dslr, the output is both Sharp and has a wonderful feeling about them.
However, I found something yesterday that got me thinking. I shot a wedding last Winter, and the church, pretty dark, had big Bright windows and the resulting contrast between dark and light Went beyond what my Fuji X100 could capture to retain both highlights and shadow detail. But when I scan my tri-x from the same lighting conditions, I have detail in both. BUT! the scan's histogram does not clip at either the shadows nor the highligts, so it got me thinking if film can capture more dynamic range, but "compress" it to a smaller range of tones?
I am at work, so I can't show any screencaptures.
Kind Regards
Marcus
Last edited:
waileong
Well-known
One doesn't scan with a DSLR.
Film has more dynamic range that scanners can capture. That's why film scanners offer multiple pass scanning.
Film has more dynamic range that scanners can capture. That's why film scanners offer multiple pass scanning.
Erland
Member
I am aware that I do not "scan" with a dslr, sorry, I meant digital capturing.
And I do know that film in general has a better dynamic range, however I was wondering if it didn't capture all of it in a linear kind of way. And by such compress the entire capture dynamic range into a smaller range? Like a gamma curve that is less than 1.0?
And I do know that film in general has a better dynamic range, however I was wondering if it didn't capture all of it in a linear kind of way. And by such compress the entire capture dynamic range into a smaller range? Like a gamma curve that is less than 1.0?
Erland
Member
I did read now that film has a pretty straight gamma curve overall, but a more s like curve in the highlights and shadows. that could explain why it can capture a greater dynamic range and still fit it all into a tighter range.
mcfingon
Western Australia
I think you've got the right idea with the S-curve on film, Marcus. B&W film was designed to be low-contrast medium which could render a huge range of tones onto a photographic print, which is relatively high contrast. Tri-X in particular stores a lot of shadow and highlight information in the beginning and end of those S-curves - the toe and the shoulder. I find if I develop film to a low contrast index, I can get pleasing results on digital captures with my Nikon D3200 camera and 40 Micro lens. I can get more highlight detail recorded this way than the digital camera's sensor normally shows.
Scapevision
Well-known
This is true, negative film does compress highlights very well, so appears to capture more than digital sensor, i haven't done anything scientific, nor am I interested. But I am waiting for digital sensors to start acting the same, instead they just act like slide film in terms of highlight retention.
Here's an example of Ektar retaining all the highlight information:
Steaming Lake in Ektar by Dmitrijs Belokons, on Flickr
I believe on digital this would have to be shot in multiple shots and HDR'ed. At least in my experience.
Here's an example of Ektar retaining all the highlight information:
Steaming Lake in Ektar by Dmitrijs Belokons, on Flickr
I believe on digital this would have to be shot in multiple shots and HDR'ed. At least in my experience.
mfogiel
Veteran
As said above, negative film has a S shape response to the light, quite similar to the way in which a human eye evaluates brightness. Digital sensors on the other hand, have a linear response with a kink. This means a digital sensor clips the highlights abruptly above a certain brightness value, while in the shadow the process is "swamped" into the sensor noise, so it is more gradual. In practice, film retains more highlight detail and digital retains more shadow detail, at the same ISO level. Moreover, through the process of pulling, you can further compress a dynamic range on film, to the point where this is probably quite meaningless - I've heard of more than 20 stops recorded on Tmax 100.
willie_901
Veteran
A significant difference is the difference between information loss between film and digital imaging.
In digital imaging when one, or all, of the RGB channels becomes clipped, all information is lost. And it gets worse. The data discontinuously goes from containing a high amount of information (at maximum exposure the signal-to-noise ratio is high) to no information whatsoever. There is no in-between. Making matters worse is there are two completely different ways to clip the RBG channels. One is to exceed the electrical charge capacity the sensor wells when the shutter is open. The second is to exceed the maximum voltage threshold of the analog-to-digital converter for any of the channels after the shutter closes.
With film the information loss due to overexposure is a continuous process. This means the consequences of overexposure are different and more forgiving. Chemical development techniques can take minimize the impact of the information loss.
To a much more limited extent overexposure can be semi-continuous in digital imaging. The total information loss does depend the percentage of clipped pixels and only one of the RCB channels can be affected.
I can't think of an objective way to compare the DR of film vs digital. It is clear digital has a higher information content. The effective SNR is higher. For a once in a lifetime situation where spontaneity is important would anyone prefer film over digital in EV = -3 ambient light?[1]
A pure analog work-flow benefits from selective underexposure (dodging) during printing. The dynamic range of the paper and print development chemistry must be relevant too.
A hybrid workflow makes data-driven comparisons even more difficult. Averaging digital scans of analog images does decrease all the random noise contributions form the scanning process by the square-root of the number of scans.
[1] The answer is complicated. It depends on the photographer's subjective views on the aesthetics of grain and the film/lens or sensor/lens. ISO 12233 test charts/wedges and shadow-region rendering quality based on SNR measurements are useful but insufficient to completely asses a photograph's impact.
In digital imaging when one, or all, of the RGB channels becomes clipped, all information is lost. And it gets worse. The data discontinuously goes from containing a high amount of information (at maximum exposure the signal-to-noise ratio is high) to no information whatsoever. There is no in-between. Making matters worse is there are two completely different ways to clip the RBG channels. One is to exceed the electrical charge capacity the sensor wells when the shutter is open. The second is to exceed the maximum voltage threshold of the analog-to-digital converter for any of the channels after the shutter closes.
With film the information loss due to overexposure is a continuous process. This means the consequences of overexposure are different and more forgiving. Chemical development techniques can take minimize the impact of the information loss.
To a much more limited extent overexposure can be semi-continuous in digital imaging. The total information loss does depend the percentage of clipped pixels and only one of the RCB channels can be affected.
I can't think of an objective way to compare the DR of film vs digital. It is clear digital has a higher information content. The effective SNR is higher. For a once in a lifetime situation where spontaneity is important would anyone prefer film over digital in EV = -3 ambient light?[1]
A pure analog work-flow benefits from selective underexposure (dodging) during printing. The dynamic range of the paper and print development chemistry must be relevant too.
A hybrid workflow makes data-driven comparisons even more difficult. Averaging digital scans of analog images does decrease all the random noise contributions form the scanning process by the square-root of the number of scans.
[1] The answer is complicated. It depends on the photographer's subjective views on the aesthetics of grain and the film/lens or sensor/lens. ISO 12233 test charts/wedges and shadow-region rendering quality based on SNR measurements are useful but insufficient to completely asses a photograph's impact.
Hsg
who dares wins
According to Ansel Adam's zone system, negative B&W film has ten stops of dynamic range. Where as today most digital cameras have 12 to 13 stops of dynamic range and 14+ in the case of some of the top end DSLRs.
But the biggest difference between negative film and digital sensors are how they respond to light. For example you can underexpose a digital camera by as many stops as you want, in RAW mode, and then you'll be able to bring back the detail in RAW software. With negative film, you must never underexpose because the shadows get clipped easily and B&W without shadow detail looks compressed and flat.
In your situation, you should have metered the X100 by pointing at the windows and exposed for those highlights, then in RAW software you could easily bring back the shadow detail.
Having said that, B&W film has beautiful highlights, there is no getting around that fact by digital flexibility.
Ko.Fe.
Lenses 35/21 Gears 46/20
Multi-pass or "highlights" mode makes it better in terms of the dark, bright areas on bw film.
Maybe those new DSLRs are capable of the same within one file, or you could take multiple exposures with old camera and combine them (HDR) in PP.
Darkroom prints would be more outstanding outcome in terms of wedding pictures, if you done it with bw film in addition to digital camera pictures.
As something really special about special day. IMO.
Maybe those new DSLRs are capable of the same within one file, or you could take multiple exposures with old camera and combine them (HDR) in PP.
Darkroom prints would be more outstanding outcome in terms of wedding pictures, if you done it with bw film in addition to digital camera pictures.
As something really special about special day. IMO.
Erland
Member
This is how I often meter when doing high contrast photos. However, since I was around iso 800-3200 depending on whether the clouds covered the sun or not, the shadow noise would be to apparent if I exposed for the highlights. The walls behind the altar, was on digital overexposed, but retained information on film. And both had coverage and details in the grooms jacket.
I love the output from my x100 (the first one), dynamic range is for me sufficient, and the colors are amazing. But How hard I try, I can't get the black and white to get the same feeling as my tri-x. The falloff before clipping at both ends, are just so much satisfying, and often when I display these images to my friends, they are impressed with the quality of light and information, light-wise in the image compared to their own.
I can't say I am good enough in the darkroom to say what looks best on paper, though I can tell a difference in prints using my inkjet. Though This often gives me an image on paper where there is to much shadow or highlight information to show a pop, and instead it looks flat and to "I was trying to print all the information"-like print. YMMV.
I work as a service technician on digital imaging machines, such as printers, copiers, rips and so forth, so I have plenty of experience on how much contrast you can get in papers, compared to the screen, and it's only on screen I think the film has advantage over digital. The paper copy is a compromise in the end, dynamic range wise.
lawrence
Veteran
lawrence
Veteran
That does make sense -- the dynamic range of film depends on which film it is and how it's developed.
Dwig
Well-known
One, he was making a generalization and more importantly, he was referring only to that portion of the D-LogE curve that was reasonably linear.
BTW, all of this "compression" of the ends of film's dynamic range in old (read: ancient) knowledge. Hurter & Driffield worked it out nearly 1-1/2 centuries ago (circa 1876) and for a long time the density vs exposure curves were referred to as "H&D Curves". Check out https://en.wikipedia.org/wiki/Sensitometry
quejai
Established
I think there may be a shortcut to getting better dynamic range from digital cameras, if you're OK with a monochrome image and you shoot with a color filter.
For example, let's say you use a red filter. So that means that the red channel in your image will be exposed as normal. However, the blue and green channels will each be underexposed - one probably more than the other - this means that these channels will preserve highlight detail even if the red is clipped. With the right photoshop technique, you should be able to bring back detail from these other channels.
The end result will be a black and white image that looks just like it was shot on a film camera with a red filter, with an expanded dynamic range to match. You end up with the shadow detail advantages of digital with the highlight compression advantages of film.
This is just a thought by the way, I'm not sure if it would work. But if it does, then it would also work with other color filters. Thinking about how digital sensors (due to the bayer design) have 2x the green sensors as red or blue, perhaps it would be better (i.e. result in shorter exposure requirements) if a green filter was used instead of a red filter.
For example, let's say you use a red filter. So that means that the red channel in your image will be exposed as normal. However, the blue and green channels will each be underexposed - one probably more than the other - this means that these channels will preserve highlight detail even if the red is clipped. With the right photoshop technique, you should be able to bring back detail from these other channels.
The end result will be a black and white image that looks just like it was shot on a film camera with a red filter, with an expanded dynamic range to match. You end up with the shadow detail advantages of digital with the highlight compression advantages of film.
This is just a thought by the way, I'm not sure if it would work. But if it does, then it would also work with other color filters. Thinking about how digital sensors (due to the bayer design) have 2x the green sensors as red or blue, perhaps it would be better (i.e. result in shorter exposure requirements) if a green filter was used instead of a red filter.
shawn
Veteran
" let's say you use a red filter. So that means that the red channel in your image will be exposed as normal. However, the blue and green channels will each be underexposed - one probably more than the other - this means that these channels will preserve highlight detail even if the red is clipped."
That is an interesting thought, especially with a Sigma Foveon camera. The so called red and green layers are already receiving a reduced amount of light relative to the top layer and all layers are full resolution (Merrill and earlier).
With Raw Digger it is possible to export monochrome TIFFs of each layer. From that you should be able to make an HDR image. I will have to play with that a bit.
That is an interesting thought, especially with a Sigma Foveon camera. The so called red and green layers are already receiving a reduced amount of light relative to the top layer and all layers are full resolution (Merrill and earlier).
With Raw Digger it is possible to export monochrome TIFFs of each layer. From that you should be able to make an HDR image. I will have to play with that a bit.
Hsg
who dares wins
From personal experience I can relate. Its frustrating to try and emulate the tri-x feel with digital, but that is mainly a problem of form - even if you learn the best B&W digital conversion, you're still faced with the even bigger problem of content...
So, just go for the content and form will follow.
willie_901
Veteran
The Difference Between Film and Digital DR
The Difference Between Film and Digital DR
As indicated in the last paragraph of this post, the exposure characteristics of film and today's digital best imaging technology (CMOS) are very different. Film has a much wider exposure latitude. In the future (5 + years), this difference may be eliminated.
Dynamic Range In Digital Imaging (applies to scanners and cameras; please ignore unless you are interested in what happens when you scan or photograph film media)
In digital imaging the analog dynamic range is a function of the sensor bed's average full-well capacity and by the read-noise level. That is: the analog signal-to-noise ratio determines the datas' DR. In photography the SNR is often described by the misnomer high ISO noise level.
The analog SNR is the SNR recorded when the shutter is open. The full-well capacity is the maximum charge (on average) a sensor site can hold. The read noise is the electronic noise generated by the sensor electronics.
Current CMOS technologies' SNR are limited by the read noise level. The read-noise level of CCD imaging sensors is higher than CMOS technologies. The newest data streams have no significant increase in read noise after the shutter closes since the downstream read noise is much lower than the sensor read noise and it does not get digitized. When the SNR is not independent of the electronic amplification applied after the shutter closes (a.k.a. increased ISO), the camera is ISO invariant. Many brands now offer cameras that are essentially ISO invariant.
After the shutter closes (or scan completes) it is impossible to increase the SNR/DR of a single image. All one can do is render the image using parameters that take full advantage of the datas' inherent DR.
Modifying (filtering) the light before it reaches the sensor could be useful. The light is the signal.
The Future
New quantum image sensor technologies, QIS) are in development by Eric Fossum's Research Group. Fossum invented CMOS sensor technology. The promise of QIS is a profound reduction in read noise to achieve a much higher SNR/DR. The goal of this technology is to emulate film's exposure characteristics (See slide 22 of this presentation).
The Difference Between Film and Digital DR
As indicated in the last paragraph of this post, the exposure characteristics of film and today's digital best imaging technology (CMOS) are very different. Film has a much wider exposure latitude. In the future (5 + years), this difference may be eliminated.
Dynamic Range In Digital Imaging (applies to scanners and cameras; please ignore unless you are interested in what happens when you scan or photograph film media)
In digital imaging the analog dynamic range is a function of the sensor bed's average full-well capacity and by the read-noise level. That is: the analog signal-to-noise ratio determines the datas' DR. In photography the SNR is often described by the misnomer high ISO noise level.
The analog SNR is the SNR recorded when the shutter is open. The full-well capacity is the maximum charge (on average) a sensor site can hold. The read noise is the electronic noise generated by the sensor electronics.
Current CMOS technologies' SNR are limited by the read noise level. The read-noise level of CCD imaging sensors is higher than CMOS technologies. The newest data streams have no significant increase in read noise after the shutter closes since the downstream read noise is much lower than the sensor read noise and it does not get digitized. When the SNR is not independent of the electronic amplification applied after the shutter closes (a.k.a. increased ISO), the camera is ISO invariant. Many brands now offer cameras that are essentially ISO invariant.
After the shutter closes (or scan completes) it is impossible to increase the SNR/DR of a single image. All one can do is render the image using parameters that take full advantage of the datas' inherent DR.
Modifying (filtering) the light before it reaches the sensor could be useful. The light is the signal.
The Future
New quantum image sensor technologies, QIS) are in development by Eric Fossum's Research Group. Fossum invented CMOS sensor technology. The promise of QIS is a profound reduction in read noise to achieve a much higher SNR/DR. The goal of this technology is to emulate film's exposure characteristics (See slide 22 of this presentation).
Hsg
who dares wins
No. What I mean is that if you're using the zone system, you divide the film contrast range or dynamic range by ten, and then you mainly use the five mid zones in your exposure, development and printing.
Hsg
who dares wins
A graduated neutral density filter is a much better solution to dynamic range for landscape shots.
If you underexpose green and blue channel, especially the blue channel is going to become very noisy.