For those who are still scanning film, especially color reversal, I would like to know your opinion on this. I have always found digital color not as attractive as that from scanned Provia or Velvia while digital seems to be better in sharpness and grain/noise (well there is also the DR which is still a controversial subject).

I have noticed that in digital, reds, blues, and greens contain higher levels of the other 2 complementary colors, which produces a somehow washed out rendition. Scanned film tends to produce purer, reds, blues and greens with much less of the other colors. RGB channels in digital are somewhat overlapping, while scanned film tends to show more separate, or not as homogenous RGB channels. Scanned film gives deeper reds, blues and greens.

The reason I believe is that digital tries to keep all RGB channels within the limit of the color space, similar to a relative colorimetric rendition. This reduces noise and creates "dull" colors. As soon as you try to make digital look like Velvia for instance, the noise quickly gets incontrollable.

Am I right in this? And if yes, there must be a way to get film colors from digital, by somehow achieving an absolute colorimetric rendition. How to achieve that? (Though I expect it will be very noisy) (I also suspect that the extra noise of the Sony A900 is caused by this as well, as I and many others here have raved about it's film-like color rendition).

Your comments would be greatly appreciated, as I have probably gotten it all wrong

Neither 'film' nor 'digital' can be easily categorized. Better to cite specific examples. I find the DMR's color quality exceptional, like the Kodachrome II of our distant memories but richer, more accurate, less grainy and faster.

As telyt questions, please be specific.

One man's richness is another's garishness. I've produced Velvia style colors controlling noise. Generally I consider Velvia garish for what it was intended, landscapes. By many, if not most, it is considered unusable for portraits because of the off skin colors.

Plus I believe I can produce richer colors. But then some do not like my saturated colors. And today's style seems to be veering away from saturated colors which is what you may be interpreting as dull and washed out.....

Nowdays we can produce virtually any color, not limited to the chemistry on film or paper. And with today's cameras, noise has become a non-issue... unless you're shooting 3200 and above. Try that with film....

It's all in the post production after RAW conversion as most RAW developers will produce near identical looking images but film stock varied much more in it's colour and tonality variations. Most film scanners with the exception of high end drum scanners produce pretty crude colour and tonality compared to both good direct digital capture and conventional wet darkroom printing.

Ok, I'm talking here about the blue skies and deep reds that reversal film is known for. All my Canon DSLR that I've owned since my 300D and all Nikons judging from net samples show pinkish light reds and light blue skies most of the time. I made 2 observations comparing digital files of both. First, histograms look different. DSLR histograms show somewhat compacted RGB peaks which are very close, while film histograms can be very distant apart, while the picture looks perfectly balanced. Hence talking about relative vs. absolute colorimetric. Second, pure reds such as a red shirt or flag can contain significant green and blue in DSLR files if you use a dropper to measure the RGB values, while with film I have measured almost pure monochromatic color in some cases with a high R value and almost zero G and B values.

The closest I can get to reversal film look is with picture styles portrait and faithful but all picture styles apart from neutral introduce significant more noise.

Now how to interprete these 2 observations is the question here.

Reds are an issue for any Bayer Sensor camera. The problem is twofold. One is that common light-reds are often in the region partially cut off by the filtration as well as the fact that the red channel is the noisiest and also the most likely to blow out.

Blues are another thing entirely. I've never had issues getting good pure blues, especially deep ones with my Nikon's or Pentax's.

Just as the film you like was manipulated to give colors with a lot of "punch" you cab do the same thing with profiles in the digital world. It's just been done for you with the film you like.

Some cameras that are not used commercially can have a standard setup to produce high color.

An accurate white balance helps produce clear colors.

i think this 98% depends on the scanner used, along with the scanning practices, and 2% depends on the film. Almost a completely pointless discussion, unless you are talking about "All things being equal" as in everybody in the world owning a Coolscan V or something.

mawz wrote:
Reds are an issue for any Bayer Sensor camera. The problem is twofold. One is that common light-reds are often in the region partially cut off by the filtration as well as the fact that the red channel is the noisiest and also the most likely to blow out.

Blues are another thing entirely. I've never had issues getting good pure blues, especially deep ones with my Nikon's or Pentax's.

I agree with your post. Reds seem to be the most problematic, but imo blues to a lesser extent since even when we can get deep sky blues with digital, it'a at the expense of noise (unless you do some further PS work which I dislike). Films like Velvia have the magical ability to create electric blue skies out of boring greyish ones. Someone mentioned on another forum film's dye inhibitors which "inhibit" the formation of color in the other 2 layers, and create this saturated look.

ISO1600 wrote:
i think this 98% depends on the scanner used, along with the scanning practices, and 2% depends on the film. Almost a completely pointless discussion, unless you are talking about "All things being equal" as in everybody in the world owning a Coolscan V or something.



I haven't worked with drum scans personally, but during my film days in the late 90s, I have been published in 2 books where hundreds of my slides were scanned on one. The results were in accordance to what I obtained with my Coolscan 5000. There must ba variations between brands or scanning software, but the observations still apply.

Ed, i don't care what you have had scanned or published. That has no bearing on this debate.

This is simply another film vs. digital, is what it all boils down to. I think it is hard to find a more subjective and difficultly-quantified argument on this forum.

ISO1600 wrote:
Ed, i don't care what you have had scanned or published. That has no bearing on this debate.

This is simply another film vs. digital, is what it all boils down to. I think it is hard to find a more subjective and difficultly-quantified argument on this forum.

What I mean to say is that the observations I made are consistent from drum scans of 10 years ago and recent scans on a Nikon Coolscan. It must be due to film not the scanner or software used.

This is not intended to be a film vs. digital thread as I made it clear since the beginning.

I am more interested on how to make process or manipulate my raw files to give me similar saturations to Provia or Velvia, if this is ever possible.

There are saturation and contrast controls in all RAW processors and these days there is the option of using different colour output profiles with C1 and ACR - these can dramaticaly change the look of the image much more than differences between film stocks. If this is not enough there is a miriad of colour and tone controls in Photoshop to change individual colours to anything you desire. In addition there are 3dr party plugins that will emulate old film stock colour response. It's nothing to do with digital vs film as a capture medium and everything to do with how you process your images.

shirozina wrote:
There are saturation and contrast controls in all RAW processors and these days there is the option of using different colour output profiles with C1 and ACR - these can dramaticaly change the look of the image much more than differences between film stocks. If this is not enough there is a miriad of colour and tone controls in Photoshop to change individual colours to anything you desire. In addition there are 3dr party plugins that will emulate old film stock colour response. It's nothing to do with digital vs film as a capture medium and everything to do with how you process your images.

I know what you mean but:

1. Contrast and saturation: All what they do imho is blow out the red channel.
2. C1: No experience.
3. ACR: I have used it in the past, and found that I could probably achieve what I want with the color tab where you can adjust saturation and hue of each color individually. But I was not successful at that. Seems too complicated for me.
4. 3rd party plug-ins: I have purchased in the past FM's velvia action, but all what it did is increase saturation using the channel mixer and a slight curve. I'm not sure what other ones there are out there.

According to my own recent experience, the most infamous portrait picture style in DPP 3.5.1 produces the closest colors that I could get to the Fujichrome family. It doesn't work all the time but I'm just starting to discover it. I especially find the reds to be deep and purer than other styles (and they don't blow out as quickly). Blues and greens are also quite attractive, but yellows can easily turn greenish.

edwardkaraa wrote:
The reason I believe is that digital tries to keep all RGB channels within the limit of the color space, similar to a relative colorimetric rendition. This reduces noise and creates "dull" colors. As soon as you try to make digital look like Velvia for instance, the noise quickly gets incontrollable.

You're on the right track. Manufacturer choice of filtration for the Bayer pattern will tend to favor long tails in the R, G, and B response. This maximizes the overall sensitivity (V channel) at the expense of color purity. Representative response curves can be found from various chip makers. The effective R, G, and B channels must be extracted with a 3x matrix multiply: in essense, a saturation op. Noise will be introduced - more so if colors are impure from the sensor. An interesting experiment would be to take three shots of a (static) scene, each using a narrow-band color filter before the lens. The resulting image may appear superior. Alternatively, one can simply reduce noise in the H and S channels with image averaging. It helps to have the stacking and demosiacing software lying around (as I do).

I expect standard films and sensors to completely cover the visible spectrum, so image "look" is merely a matter of color mapping.

Ref:
http://www.kodak.com/global/plugins/acrobat/en/business/ISS/supportdocs/ColorCorrectionforImageSensors.pdf

Thanks rico. Very interesting article.

So I can emulate film by shooting RGB separately and make my own digital emulsion

Some very early digital capture systems had a rotating colour wheel ( RGB) and took 3 shots - good colour but not practical outside the studio.

Also it would seem that colour fidelity - not just the ultimate saturation but the seperation of hues (which is what Edward is realy concerned about) has been sacrificed on DSLR's to chase the demand for ever higher ISO by reducing the strength of the individual RGB filters over the sensors. MF backs have much superior colour rendering but quite poor high ISO capability and this has been stated as a fact by at least one manufacturer ( although I can't remember which). While customers demand and forums get hysterical ( see the Canon 5D11 master thread) about the need for high ISO capability I can't see this changing.

so basically, what he wants is a Foveon.........

hasn't then been discussed very heavily in past years on FM? Several very fervent Sigma fanboys on here, IIRC.

shirozina wrote:
Also it would seem that colour fidelity - not just the ultimate saturation but the seperation of hues (which is what Edward is realy concerned about) has been sacrificed on DSLR's to chase the demand for ever higher ISO by reducing the strength of the individual RGB filters over the sensors. MF backs have much superior colour rendering but quite poor high ISO capability and this has been stated as a fact by at least one manufacturer ( although I can't remember which). While customers demand and forums get hysterical ( see the Canon 5D11 master thread) about the need for high ISO capability I can't see this changing.

Exactly! The cameras that seem to do well in the color department suck in the high ISO noise department (or even low ISO comparatively). Not only the MF backs but also the Sony A900 that I'm still considering for the better color it produces but I'm turned down by the bashing it is taking for its higher noise levels.

I think the three things which combine to result in the lack of color purity you observe with DSLR sensors are:
1-bayer filter interpolation of color from a b&w sensor.
2-the straight line tonal response of a digital vs analog.
3-AA filters to counteract bayer filter interpolation artifacts.

The best color I have seen from digital has always come from the larger sensors in digital backs OR from similar technology and design decisions employed in smaller sensors. Usually, these sensors have some of these qualities in common:
1-the elimination of or use of extremely weak and or better/more expensive aa filters.
2-higher dynamic range.
3-better low iso image characteristics in exchange for high iso/low noise.

I do think the higher dynamic range property is very important in the ability to use a more extreme curve in trying to get the most out of color purity and tone as they are related.

Just non scientific observations.

A foveon style bw-sensor would be the way to go if you want good bw, and I think that something similar to foveon technology will be the future for color photography too. Getting rid of many of the problems that the bayer array introduces.

ulrikft wrote:
A foveon style bw-sensor would be the way to go if you want good bw, and I think that something similar to foveon technology will be the future for color photography too. Getting rid of many of the problems that the bayer array introduces.

A Foveon-style sensor would be very poor for B&W since all you really need to do for a dedicated B&W sensor is to remove the bayer filter pattern from a standard sensor. Kodak did this with the (rare) DCS760m, which produced superb B&W output.

ulrikft wrote:
A foveon style bw-sensor would be the way to go if you want good bw, and I think that something similar to foveon technology will be the future for color photography too. Getting rid of many of the problems that the bayer array introduces.


Actually, if you really want good B&W, get a a CCD sensor without a RGB filter (a Foveon style filter is irrelevant for B&W, just like the Bayer filter, you don't need 3 different color sensitives at each photosite or alternating photosite to capture B&W). Monochrome versions of most bayer filters are made by sensor manufacturers.

Kodak produced a monochrome version of one of it's early DSLR's and it is, from what I've seen, spectacular and very sharp. Noticeably more resolution that the equivalent bayer-filtered version. Of course, you would have to go back to using color filters to control the tonalities of the image (because you are not capturing colors that you can select later). Of course, the best thing about digital, color digital that is, is that for B&W you don't need to bring a half dozen red, orange yellow, green and blue filters with you, as well as assess filtration and constantly change them out to optimize a particular image before you capture it.

mawz wrote:
A Foveon-style sensor would be very poor for B&W since all you really need to do for a dedicated B&W sensor is to remove the bayer filter pattern from a standard sensor. Kodak did this with the (rare) DCS760m, which produced superb B&W output.


You beat me to it. ;-)