Did Kodak color neg lose sharpness when it was reformulated to improve scanning?

[interneg]

You can start with bare consumer films.

Later you can compare Vericolor with Portra that was introduced in 1998 to be the Vericolor VPS and VPL. Probably the early Portra changed soon.

Also you can compare equivalent Fuji films.

Then compare with Velvia.

Which specific films have you looked at? I've worked extensively with all of these & they largely match the characterisation of their data sheets in terms of sharpness. Or have you just looked at a random sample of Velvia & a random sample of Portra and drawn inaccurate conclusions based on assumptions?

[Pere Casals]

Which specific films have you looked at?

All I could, all Portras, Ektar, Fuji 160, Provia, all Velvias, Vision 3 (cinestill)... HP5, TX, TMXYZ, D100, CMS 20, Plus-X, D-X, S-XX, Valca... Rodinal vs Xtol, Xtol stock vs 1:1 . C200, Xtra 400, Gold, Color Plus, Ektachrome... Say that I spend 10min every week in that.

they largely match the characterisation of their data sheets in terms of sharpness.

interneg, I don't want to "lecture" you in how datasheets have to be interpreted, but those film MTF graphs are probably done at 1000:1 contrast, I recently told you that (contrary to lenses) film MTF is highly dependant on target contrast. 1000:1 are 10 stops, a contrast situation that you won't find on 30cycles/mm textures in a negative, by far.

You know that CN film has an extraordinary highlight latitude, obtained by including a share of very, very small crystals, at 1000:1 the MTF graphs shows that.


See datasheet, Image Estructure section, page 4: https://125px.com/docs/film/kodak/e4051-Portra-160.pdf They look politicians, they speak a lot and say nothing .

and drawn inaccurate conclusions based on assumptions?

Those are not my conclusions, at all, I'm not that good, this was shown to me long ago by a technical service boss (highly proficient and technically educated) in the digital minilab sector, I inspected film strips with him in his microscope, let me explain you in how this is done. You inspect areas in the negative that were grey in the scene (concrete, buildings... ) you inspect greys of different densities, and you learn if color clouds overlap more or less for each density level.


This are the key questions;

1) is velvia more difficult to scan than potra (or C200) at (4000dpi effective) high res ? more or less color noise at pixel level ?

2) why ?

3) beyond 1000:1 graphs, is velvia sharper than Fuji 160 in practice

4) why ?


Since digital minilab era all manufacturers started saying that their film was easier to scan (less color noise), but none of them were telling how this was achieved, larger clouds around the crystal.

Not new in photography, no fine grain solvent developer says in the datasheet that you also have less sharpness, so at Kodak/Fuji they didn't have to think much when writing the datasheet.


Anyway, the "larger clouds" vs "easy scan" had a debate long ago, I can't belive you weren't aware. Today nobody complains, darkroom RA-4 is near extinct, but in the 1990s we had a lot of Pro color darkroom labs for wedding, etc,

By then darkroom prints from 35mm film noticed an slight drop in sharpness, while that change was a benefit for the digitals minilabs, less color noise in the scanning, while the slight sharpness loss was solved with some digital sharpening.

[interneg]

datasheets have to be interpreted.

A microscope is not going to help you if the visual density of the magenta dye is changed, but not its spectral response when exposed to the paper, mainly because your eyes do not respond like photographic paper (or a digital sensor for that matter). There were significant changes in visual dye density in the mid 90's onwards, which seem likely intended to shorten print exposure times & reduce paper reciprocity issues.

As for your assertions about sharpness tests, having been & checked what current procedures are, they involve x-ray and visual light exposures (which allows for testing of emulsion turbidity) of a set of known width slits down to 1 micron which are repeated at a very wide array of exposures, each a stop apart - the results are read by microdensitometry, then the RMS grain data is added before an MTF result is arrived at (which also enables a comparison of the behaviour of the emulsion's sharpness characteristics in different format sizes). This is repeated across each of the colour sensitivities & the combined coated package. This seems to be fairly standard industry procedure, but is only a part of a much larger quantitative analysis that is carried out to try and accurately represent the 'information capacity' of the film. MTF data below 30-50% response doesn't tell you much about the sharpness of a material, so it's largely omitted in the datasheet.

Finally, if the practical sharpness of positive materials was as high as you claim, the cinema industry would have adopted them fully a long time ago. They didn't because neg-pos has major sharpness, colour etc advantages. This is because of the potential to use DIR/ DIAR couplers etc in C-41 - vis-a-vis citrazinic acid and other less effective sharpness enhancers in E-6 (which also has to use a highly solvent first developer to work correctly), but there were also highly mathematical studies done which came to the conclusion that neg-pos was better performing across the board. This sharpness behaviour difference is pretty obvious at even moderate sizes, no matter if you optically print or scan on decent kit. Transparencies remained in use for so long because they gave an absolute colour reference for repro, not because they were 'better'.

[Pere Casals]

As for your assertions about sharpness tests,

Not my assertions, but what's very well known:


Reference: http://www.tmax100.com/photo/pdf/film.pdf
https://web.archive.org/web/20180731...o/pdf/film.pdf

Look, Kodak VR 100 and VR-G : consumer film of the 1980s is sporting 100 lp/mm at 30% MTF, it not says the exposure, but it's well sharper than (!!!) Pro film of the late 1990s that was designed to be scanned, and sharper than any 2019 CN film, consumer or pro.

To me this is a fact.

A microscope is not going to help you if the visual density of the magenta dye is changed,

Let me reiterate it a last time... take a good microscope and inspect around x800. Inspect gray subjects having density around 0.85D and around 0.25D, you'll see why color consumer film of the 1980s is sharper than consumer and Pro CN film made in 2019.

> Compare consumer Kodacolor VR-G (1988) vs Kodacolor Gold (2019)

> Compare professional Kodak Vericolor III VPS (discontinued 1997) vs Portra 160/Ektar 100 (2019)



Of course, I prefer modern CN film because it's better for the hybrid workflow we have today for color, but 1980s color film was sharper.

For MF and up there is no problem with that lower sharpness. For my Nikon F5 I feel the modern film limitations, even when it's scanned 8000dpi in a drum, don't you ?


_______


Interneg, we have debated too much (since 2 years ago) the changes in the clouds from the film adaptation to the hybrid workflow, if the consumer 1988 Kodacolor VR 100lp/mm rating is not enough for you, then I cannot do more. I cannot belive you were not aware of that situation, really.

Today Kodak has no CN film that is as sharp as the 1980s consumer VR, this is a fact. And this is 30 years later.


_______




Here you see before/after bleach, at left you see the silver clumps, depending on chem diffusion effects (IIRC) (I guess reaction speed may be involved, slower reaction more diffusion) the cloud can be smaller and more intense, or larger and less intense. With larger clouds there is less color noise in the scanner from aliasing, this is evident, if clouds of diffrent colors overlap more then a pixel takes more the same color than the neigbour pixel. If not the neighbor pixel can take a very different color: one pixel takes the magenta cloud and the neighbor takes the yellow cloud, but no pixel takes both overlaped to capture the real subject's color: this is color noise.

Problem was that sensor discretization magnified that noise from aliasing in the scanning, requiring larger clouds for best overall result. Then the Noritsu added some digital sharpening, some powerful image enhancing aesthetics, et voilà, we had amazing color prints, 10" was a lot for consumer market.

Want a big print? then shot 120 !!! Sorry, clouds are for scanning. This was the message in the late 1990s.

[interneg]

Pere - this is why the granularity of the material matters because it affects the information capacity. MTF is signal, RMS grain/ print grain index is noise. VR/ VR-G/ Gold have high MTF but also Print Grain Indices equivalent to or worse than Portra 800. This is ok in a product meant for perhaps 4x enlargement from 135 while retaining good sharpness, not so good if you want to make big prints. Pro Image 100 which is a Gold derivative shows this problem very clearly. The critical discoveries leading to Portra etc seems to have been that a 100%+ response out to 20-25 lp/mm gives the best sense of real 'sharpness', especially in larger formats, while holding the 50% response to about 50-60 lp/mm drastically reduces excessive granularity - to the point that an 8x enlargement has about the same visual granularity as a 4x off Gold etc. This is all visible in optical prints.

[Pere Casals]

interneg, Kodak Vericolor III VPS was not grainy in optical enlargements, it was only grainy in the frontier and in the noritsu, it was the workhorse of many expensive wedding photographers that were machinegunning with it and printing in top notch optic labs, delivering an atonishing work. I've seen several Vericolor wedding albums, today's wedding digital Pros are very far from achieving that level, with all Ps and all those presets they look like rookies, compared.

Don't think VR-G and Vericolor have flawed designs, both were T-Grain, in fact TMax films came from a color technology spin-off. VR-G and VPS were optimal for optic enlargements, Portra and Gold are optimal for scanning, this is what changed the clouds, as the industry changed, just before the big change, the digital dawn was also comming.


The point is that 100lp/mm color performance in the 1980s was killed, and insted you have this wording in datasheets: "Now the PORTRA Films have been reengineered to deliver ... improved scanning performance ..."


IMHO there is no doubt, you have the 1990 "100lp/mm" or you have "improved scanning performance". Today we have the last one.

Anyway at least you should agree that from Vericolor/VR-G to Portra/Gold clouds grew... those 100lp/mm had to have smaller clouds.

[interneg]

Here's the thing Pere: Kodak Gold films were launched in 1988. They will effectively have entered design several years earlier, long before the whole digital minilab thing. One of the design team members was Ron Mowrey who has been quite clear that the only significant adaptations to Kodak films specifically for improving scanability over the years was to the topcoats. Improvements in overall grain size, edge sharpness etc help both digital and optical printing - it's the marketing department's choice as to how to promote this for what post-production purpose. Vericolor 120 had two retouching surfaces, and that is likely what the early digital minilabs struggled with, especially if they had rather hard sharpening in use.

I've scanned a lot of VPS and VPL, and not had any significant issues - the granularity is in the ballpark of Portra 400, much as the print grain index suggests. I have little interest in the banalities of wedding photography, but a lot of significant artists used VPL & VPS in a wide range of formats & it still makes better than good prints without pain. The Portras have more saturation all round & a large object sharpness response closer to that of the Gold etc films with a bump in the 10-25 lp/mm range.

It's all about the highest possible signal with the least possible noise. High signal & high noise aren't good, but if reducing the signal slightly has the effect of very drastically cutting the noise, more information can be contained. Noise/ grain usually peaks in the middle densities, and drops towards dmin & dmax, thus higher noise/ granularity will affect how people might perceive the resolved behaviour of an image, even if it has a notionally high resolution at high contrast.

If you are aiming a film at amateurs who shoot 135 & get 6x4's, it needs a rather different blend of sharpness, resolution and granularity compared to a professional film aimed at good tonal reproduction and finished in multiple formats. In one, microcontrast performance is all that matters, in the other, everything from micro to macro scale contrast performance must be considered & kept reasonably consistent, otherwise people complain about why their 4x5's look so different from their 135 images.

[Pere Casals]

Let's review the facts we have:

> No Pro complained about Vericolor III VPS

> Around 1996 all consumer an pro CN films changed, at Fuji and at Kodak

> By then new datasheets said "reengineered for scanning performance"

> At the same time Performance drops from 100lp/mm to 50-75 lp/mm (30% MTF)

> Color clouds became larger and overlaped more



Do you still think that the only change for scanning was in the top coating ? Where Photo Engineer says that?



Look, by 1996 we had Pentium "1" PCs running at 75-100MHz, and a frontier had to digitally process a print in a few seconds, perhaps 8 seconds, as it printed massively. The digital minilab revolution required that change in film.


That 100lp/mm to 50-75 lp/mm drop was painful, if was a 15 years step back to pre T-Grain emulsions. And it was shame. Benefits for the hybrid was the trade-off.

Here's the thing Pere: Kodak Gold films were launched in 1988.

The point is how Gold films changed around mid 1990s.

_________________________


Again, take the microscope...


_________________________


Why a shame? Pros acquired expensive prime and zoom glass that was suitable to take advantage from 100lp/mm film performance, with new 65lp/mm film (Portra NC) all that glass was performing like consumer discount glass and kit optics. The kit Nikon 28-80mm f/3.3D became as good as best pro glass, not suitable for pro usage by performing equal.

Of course, LF fotography noticed less the change, or simply nothing was noticed.

[interneg]

Pere, if you had seen the German language data sheet for the Gold films from which the MTF data for Gold 100 was harvested for that PDF document you cite, you wouldn't be making those claims. The datasheet is explicit that the films have been optimised to ensure good colour and fine grain in scanning. What we can draw from that is that the signal to noise image content at that MTF was not ideal for super low granularity across a range of image formats. And if you have seen or made large optical prints from the film in question you would see why. It's got about as high a level of granularity as Portra 800 - which obviously has a 3-stop speed advantage.

[Pere Casals]

Pere, if you had seen the German language data sheet for the Gold films from which the MTF data for Gold 100 was harvested for that PDF document you cite, you wouldn't be making those claims. The datasheet is explicit that the films have been optimised to ensure good colour and fine grain in scanning. What we can draw from that is that the signal to noise image content at that MTF was not ideal for super low granularity across a range of image formats. And if you have seen or made large optical prints from the film in question you would see why. It's got about as high a level of granularity as Portra 800 - which obviously has a 3-stop speed advantage.

Obviously in color films there is a trade-off: "granularity" vs "sharpness".

Interneg, the key concept is that the optimal trade-off for Frontier/Noritsu scanning is different than the optimal one for optic prints, because scannig delivers aliasing when the clouds and the pixels have similar sizes, so as scanning amplified granularity the best trade-off was shifted in the larger clouds side.

And this ended in less sharp color films, today no color film sports the sharpness it had 1980s VR 100, by far. And belive me, that change was not because Pros were complaining about Vericolor granularity, it was the counter, they complained about the loss in the sharpness of Portra NC, still they found that hybrid processing had powerful benefits in image control and in manpower costs, so there was no way back.

Also the loss in sharpness was partially addressed by digital sharpening, and at the same time it addressed some blur from the taking.


So at least you admit now that color clouds changed in the 1990s, well, now you only should realize what was the reason: hybrid processing required another trade-off.


We also have to understand the computing means of the era, it was not possible to oversample to reduce aliasing because hardware limitations, as several prints per minute were crafted in a digital minilab.

It's got about as high a level of granularity as Portra 800 - which obviously has a 3-stop speed advantage.

First is that granularity is not necessarily bad, and second Vericolor 100 had not that granularity, by far, if well scanned or optically printed, similar to Portra 160 and at least well less than Portra 400:



Vericolor could have granularity in scanners with low dpi with performance that limited by the discretization, and not by the optics.

_________

Probably, with present computing/scanning means it would worth to reengineer the color films for more sharpness, but marked size does not allow for that, they manufacture the receipes they have.