@Steven Lee you've hit upon an incredibly complex topic, and to cut matters short: you're probably never going to find out what's happening.
This is because the issue is multi-faceted and trying to unravel the exact cause of this phenomenon requires rigorous testing and equipment that's likely beyond your means - not to mention that at an early stage you'll probably call it a day and move on. Which I would certainly recommend
The issue at hand is metamerism, or metameric failure, where a hue is represented in a different way depending on who/what observes it and under which conditions. The causes for this metameric failure are very diverse and the issue explodes in complexity because of the several stages of color processing, each of which being prone to variations:
* The nature of the colorant itself
* The lighting conditions
* The recording device (in this case film in a camera)
* Film processing
* Film illumination during scanning
* The scanner/camera used for A/D conversion
* Display of the result (which in turn can explode in several directions)
* Human observation of the result as well as the original during recording
Each of these areas has an underlying complex of causes that can result in hues ending up different than anticipated, and each of these requires a pretty thorough understanding of the underlying physics. Hence my somewhat pessimistic comment that you're probably never going to figure this one out - it would take years of study just to understand what
might be happening, let alone determining what really
is going on. I'm also not in a position to give a full answer (apart from my lack of willingness to write an entire book in this text box!), but I can highlight a few problem areas in particular.
Firstly, there's the question of the nature of the hue itself. You mention a hue as if it's only one, but I can firmly guarantee that the dress and the wisteria flowers are fundamentally different hues. Each colorant (let's say a dye) has its own, specific wavelength absorption spectrum, which is the description of which wavelengths of light it will absorb and which ones it will reflect (resulting in some kind of color). Even if you take two dyes that visibly seem to match in color, their actual absorption spectra can be very different. Here's an illustration:
From here, page 18:
https://www.sto.nato.int/publications/STO Meeting Proceedings/STO-MP-SET-277/MP-SET-277-12.pdf
These are the spectra of four blue pigments commonly used in paintings (especially older paintings). Yes, the hues are different, but the spectra are actually quite
wildly different. Note how the Azurite is a pretty pure blue that reflects anything in the blue spectrum and blocks everything else. Compare this to Smalt, which actually is a blue and red pigment with significant reflectivity in the deep red and infrared region of the spectrum. Indigo does a similar 'trick' and probably reflects a lot of UV as well. All of these spectra are actually quite decent in the sense that they're fairly gradual. Some dye spectra are much more erratic; see for instance the caretoneid spectrum on this page:
https://www.savemyexams.co.uk/a-lev...ss/13-1-5-absorption-spectra--action-spectra/
You may be able to imagine that such differences in reflectance/absorbance have very profound effect if you start overlying them on top of the spectra of (1) different light sources, (2) light-sensitive silver halide emulsions, (3) digital CCD sites and (4) rods and cones in the human eye. Essentially, these complex interactions probably explain most or all of the hue shift that you're witnessing. Let's have a look at some of those emissions and sensitivity spectra:
Spectra of different light sources; taken from here:
https://www.researchgate.net/figure...es-a-Xenon-lamp-b-incandescent_fig1_330372622
For the dress, which will spend at least some of its time being photographed indoors, the above is quite relevant. But even natural light plays tricks on us:
From here:
https://handprint.com/HP/WCL/color1.html
Notice how especially in the blue/violet region, funny things happen depending on the nature of (natural!) light that hits those wisteria flowers.
This is then what your film (Portra 400) sees:
From its datasheet:
https://imaging.kodakalaris.com/sites/default/files/files/resources/e4050_portra_400.pdf
Notice that your suspicion that the film may be blind (my reformulation of your words) to this particular hue is likely not the problem. However, it may record the hue in a way that causes problems/shifts later on in processing. For instance, it certainly is a little more sensitive to certain hues and a little less to others, and especially if the absorption spectrum of the violet colorant (in the dress, in the flower) is pretty lumpy, you could run into odd amplifications of certain hues while others go surprisingly dim. So yes, the film may play a role, but my guess is that this is probably only one of the minor effects.
Of course, the above is the sensitivity of the film (i.e. its silver halides), but its actual hue reproduction (the dyes forming the image) is yet another one:
(From the Portra 400 datasheet again).
Don't try to match the wavelengths above to real colors in your scene - we're looking here at a dye reproduction, including color correction mask, of a color negative film. At this stage, any firm relationship with reality is basically lost. The only way we can recover something resembling accurate color reproduction is through calibration - and accepting that indeed, our recording gamut is
always limited compared to reality.
In a digital camera, which you're using, funny things also happen; here's the spectral sensitivity of a random CCD (of a Nikon D700 as it happens):
As you can see, there's ample crosstalk between color channels (although this sensor seems to perform pretty decently by all accounts) and there are also distinct non-linearities that need to be somehow 'fixed in post' if we want real-world accuracy. Much of this already happens within the camera itself and/or in any software RAW processor.
Note that you actually have to look at the interactions between the camera CCD, the illuminant used to backlight your film, and the dye absorption spectra of the film. And this in relation to how the dye image actually correlated with real-world colorants under the illuminant at recording time - you still with me?
Then comes another tricky bit, which is how your computer monitor constructs hues. As you know, it doesn't have any violet emitters, so it fakes violet by mixing mostly blue with a little red and maybe a tiny pinch of green (which will be less if it's a really saturated violet). However, the gamut of your monitor is of course also limited - yes, it's really nice, but still limited. And especially in the violets, it's quite limited, although yellows/oranges suffer just as well. This has to do with with the 'saturation cost' of mixing colors from primary (see here for an excellent discussion:
https://www.handprint.com/HP/WCL/color14.html#satcost) - simply put, if you start with red and blue of high saturation, a mixed violet will always be of a somewhat lower saturation. Your real-world violet dye (especially the wisteria; not so sure about the dress...) may be a very high chroma, pure violet. As a result, it will simply be impossible to render with the same chroma/saturation on a computer monitor. Although, as said, those are really good these days. Still not perfect. Btw, this issue of saturation cost is also VERY relevant when you print your images, regardless of printing method. This is also the main reason why high-end printers these days not only print CMYK, but actually have red, violet and green inks added to the spectrum to increase chroma in those areas of the color wheel. (No, Fuji didn't like it when I commented that their reds on Crystal Archive papers can never be as good as some inkjet printed reds - but it's inescapable!)
Now, we've arrived at the really tricky bit, and frankly, I don't know nearly enough of this to say much about it - but it's us, and our own, human eyes. This is a universe of complexity all by itself, with the sensitivity spectra of our rods and cones, but also and very importantly ( & confusingly) the psychology of color perception. There's a running joke between my fiancee and myself about my teal sweater that I consistently call 'green' while she refers to it as my 'blue sweater'. The reasons why we fundamentally disagree which primary or secondary color the damn thing is, is illustrative of the complexity of this topic. For all we know, we may be seeing different colors. Or maybe it's a linguistics thing, mostly. Or .... a thousand other possibilities...
And this is only the story of the spectra. What we have not touched upon are issues of color management and profiling in digital space, issues of fluorescence as
@MattKing hints at (although I think they are fairly insignificant by and large and merge into the issues outlined above; think about it), the problem of that dress not being one particular hue, but actually spanning a significant part of the color wheel, how human memory of a hue is probably not always reliable, how well your C41 processing actually is, etc. etc. etc. The list just goes on.
The tricky bit is of course that all of these interactions are occurring throughout the color reproduction process, sometimes even at the same time.
I'm not sure if this helped. Practically, probably not, because it's not going to solve your problem. But maybe all of the above will somehow help having peace with the fact that color reproduction is a tricky son of a gun, and that it's a pretty deep hole you're digging for yourself if you want to start fussing about it. In all honesty, if you want your violets to render better than on Portra 400, consider shooting digital. For all intents and purposes, it's going to give you better reproduction overall and an easier job getting to screen (or print) what you see before your eyes. As you can tell from the above, with film, there are some additional steps in a film-based process that bring severe complications. Even despite how phenomenally good color film technology (still) is today, it adds a few layers of possible errors especially if you're going to end up with a digital output anyway.
This is not to say that film is somehow dead. Here's a funny anecdote: I was discussing issues along these lines with an all-analog color printer and photographer couple a few months ago. We came to talk about the choice for film (mostly Ektar I think they use for color) and paper (Kodak Endura, probably now switching to Fuji) for their work. They were adamant in their assertion that it was just not possible to accurately capture certain hues from nature (especially plant dyes, coincidentally) using digital technology. That's why they stick to film and paper. I'm skeptical, and I think it also has something to do with not being capable or willing to relearn color reproduction in the digital domain. But it does illustrate that at least for some, and sometimes, the analog method is still a more reliable and easier way to their end result. But you may have to do without the sunlit wisteria violets.
