RA-4 paper learnings as it applies to scanning color negative film with a digital camera

koraks said:

Yes, I agree - although I'd like to invite @Adrian Bacon to address this in a separate thread so that the present one can remain focused on equipment/tooling choices.

Click to expand...

Ok. I don’t know how much of this is useful as a fair amount of it is relatively difficult/tedious to do with modern image editing tools, but I’ll dispense what I know. A fair amount of it might seem to be stupid simple common sense type stuff, but for some reason there abounds all sorts of terrible advise about it on the internet.

so… C-41 is standardized. What does that actually mean? Well, it means it’s designed to be printed with RA-4 paper. OK, what does that mean? Well, do the manufacturers make RA-4 paper for every color negative emulsion made? Nope. There’s a small handful of RA-4 papers that most all color negative films are designed to be printed on, and the variation between RA-4 papers isn’t that big. So, not to be too blunt about it, RA-4 paper, and by extension C-41 effectively has a set contrast response for each color channel for ALL films. Wait.. what? Yes. Now some films may have more or less contrast, but it’s relative to that standard set contrast and can effectively be ignored unless You want to dial it out.

other points about the contrast: we don’t care what the contrast is as seen by a densitometer. We don’t care what the contrast is as seen by RA-4 paper. We care what the contrast is as seen by our scanning lens and digital camera. How do we get the contrast? I used a combination of kodak control strips (HD-LD) and made careful test exposures that does a similar thing with all currently available film stocks and measured the contrast for each using my scanning setup, then went and calculated the average.

what about the orange mask? What about it? If you set the per channel contrast correctly, then white balance, the orange mask goes away all by itself, just like with RA-4 paper. Think about that and let that sink in.

so from a simple math perspective, it’s: get sample, apply contrast to each channel, invert, set white balance, apply digital exposure adjustment so the important parts of the image are in the visible part of the display. There’s obviously more to final output than that, but that’s the gist In terms of roughly mimicking how RA-4 paper works with a digital solution.

questions?

@Adrian Bacon Thank you for this explanation. This sounds promising!

One thing that I don't quite understand is the difference in spectral sensitivities/dye absorbance of film and spectral sensitivity of digital cameras. There is a good paper by Jack Holm on the subject:


I know I asked this question before (probably more than once) but I would like to know your thoughts on the mismatch in the spectral responses of film dyes and camera sensors. Is this a problem at all and does conversion software needs to address it?

Romanko said:

I would like to know your thoughts on the mismatch in the spectral responses of film dyes and camera sensors. Is this a problem at all and does conversion software needs to address it?

Click to expand...

Let's go back to RA-4 paper. Its spectral response to light is what it is, so it has a set spectral response right? Color negative films all have different masks, which means the dyes also aren't exactly the same between emulsions. If they were, the masks would all be the same, but they very clearly are not. So, since RA-4 paper has a set response, and it therefore can't match all C-41 films because they all have different dyes and masks, then does it seem to make a difference in the ability to make a good looking positive image? It might have some effect, but apparently not enough to matter.

Likewise, a digital camera also has a set spectral response that can't possibly match the mask and dyes of all color negative film, so it's in the same boat as RA-4 paper, right?

Now... All that said, I have no doubt that you can really exacerbate the situation by scanning the film using a super spiky light source or a super narrow band light source. But, all I have to say to that is "why?" Color enlargers use a full spectrum light source. Digital cameras are designed to see the full spectrum of visible colors. Don't gimp it by using a terrible light source. All that does is make your scanning life harder.

You are right, if you don't care if your scans look like C-41 printed on RA-4 (I'm not saying you shoud), nothing really matters.

@Adrian Bacon, in the color enlarger world they used mostly halogen bulbs balanced for 3400K. Is that a good place to start? Or would 5600K electronic flash be better? How would you set the camera daylight, tungsten or some thing else.

I have come to similar conclusions about the relative simplicity and universality of inversion but through flatbed scanning. But I have also found that factors present at the time of exposure can have effects which are far more difficult to control.

Low cri subject lighting or a mixture of lighting with different spectra, and over or under exposure causing one or more color channels to become compressed in the negative itself are two of the most common difficulties I face during inversion. I have found no easy one size solution to either of these problems and generally I solve them by making a lot of masks and tweaking until my eyes become color blind.

These are not technically problems with inversion or scanning themselves but they present and must be dealt with at the inversion phase or else write the image off as a loss for its terrible color.

brbo said:

You are right, if you don't care if your scans look like C-41 printed on RA-4 (I'm not saying you shoud), nothing really matters.

Click to expand...

I've always been confused by the whole effort of trying to match scans to an RA-4 print. They're completely different display mediums that could not be further apart from each other. That seems like an exercise in frustration. If it were that easy, somebody would have already done it. Not even Kodak (the designer and maker of both C-41 and RA-4) has done it, and they know more about this than any of us.

So you match the scans to RA-4 prints if the customer orders prints since the display media is the same and not if customers only wants scans?

Btw, what confuses you in the desire for a scan that looks like an RA-4 print? Why am I shooting film then? Just to spend money? Or do you dismiss it just because it's hard(er)?

ags2mikon said:

@Adrian Bacon, in the color enlarger world they used mostly halogen bulbs balanced for 3400K. Is that a good place to start? Or would 5600K electronic flash be better? How would you set the camera daylight, tungsten or some thing else.

Click to expand...

Full spectrum is actually more important than the actual white balance of the light, though, in the interest of having as many discrete blue sample values as possible, a colder color temperature light is more preferable. If you want to avoid posterization or color banding artifacts, you want at least 10 bits worth of blue color samples. I use a strobe.

For the camera, white balance is mostly digital gain of the red and blue channels, and you want to avoid clipping the green channel. What I do is capture a frame that is completely blown out and use that to set the white balance (via custom white balance) so that the digital gain of the red and blue channels is 1.0 (i.e. no gain). The results in an image that looks very green. This is fine, because now you can use the histogram to set the camera exposure and you can see where the green channel is and whether or not it's clipping, and where the red and blue channels are relative to the green channel. If your camera supports color profiles, use one that has the least amount of contrast added possible, and use the lowest ISO possible. On Canon cameras that's the Neutral Picture Style. I set one of the custom picture styles to use neutral, then go in and turn the contrast all the way down. This gives you the most linear preview which makes it easier to tell where the green channel is actually peaking relative to the actual sensor clipping. Set the camera exposure so that the green channel peaks just before the clipping part in the histogram when capturing a blank piece of film base plus fog. If you have blinkies happening, that's bad.

_T_ said:

I have come to similar conclusions about the relative simplicity and universality of inversion but through flatbed scanning.

Click to expand...

I suspect that a lot of people think that *it has* to be more complicated than that. There's no way it can be that simple. But... From my experience, it's just not as complicated as it seems to be. You need to have the correct per channel contrast, and the correct white balance for the positive image. If you have those two criteria handled, a lot of things become non-issues. The biggest challenge has been that most image editing software out there doesn't have good tools to control those two things.

brbo said:

So you match the scans to RA-4 prints if the customer orders prints since the display media is the same and not if customers only wants scans?

Btw, what confuses you in the desire for a scan that looks like an RA-4 print? Why am I shooting film then? Just to spend money? Or do you dismiss it just because it's hard(er)?

Click to expand...

If the customer orders prints, unless they specify analog RA-4 prints, they get a digital print of the digital scan. So few people are willing to pay for RA-4 prints (most aren't even willing to pay for prints of any kind at all) that it's not even on the list of available things, though my shop has the capability to do it. If somebody asks about it, sure, I can do that, but it's a lot more expensive than just getting the scans and/or digital prints.

Modern digital printers are designed to reproduce what is in the digital color managed environment, which is not an RA-4 print. RA-4 is a unique analog output. If you want that unique analog output from an analog C-41 input, then the simplest and easiest path to that is to actually make an RA-4 print from a C-41 negative. That's not to say you can't make the image in the digital color managed environment be as close as possible to what you'd get if you actually made an RA-4 print, it's just that at some point the time and effort to do so costs more than just making an RA-4 print. If you're a hobbyist at home and don't value your time, then by all means, go for it. Just make sure you go through the effort of actually making RA-4 prints that you can use as a reference first, which is not an inconsequential amount of time, and money.

I'm not confused about the desire of someone to make a scan that looks like an RA-4 print, I'm confused why somebody would pick the path of scanning a negative to try to get there. If you want a digital image that looks like an RA-4 print, it's a lot less pain to just make an RA-4 print from the negative and then scan the print in. It looks like an RA-4 print because it's a scan of an actual RA-4 print. That path is way simpler than going through a mess of digital convolutions to try to mostly get there.

People shoot film for lots of reasons. Some people do it for that unique RA-4 look, but these days, most people do it because they're young and never shot film before and even with digital scans, film offers a unique look that doesn't generally look like a digital camera capture, so while the digital scan doesn't generally look like an RA-4 print, it also doesn't really look like what you'd get if you shot that same thing with a digital camera.