New from Moviestuff

[Tscan]

I use Vegas 10. I've done a few tests in the past using various methods, including the one you mention. Also tried switching to a camera with an invert feature. The biggest hurdles are getting the right exposure for density, and getting rid of the blue cast. Dealing with correcting either of those issues as a starting point usually doesn't leave enough head room in my gadgets for any further correction. Unless you render it and do a second pass, but it just makes for a cluttery chain and more hard drive space. What we need is something in a NLE or capture device that can do the same type of basic neg inverting that still film scanners do, or even the new cheesey Lomo app that allows you to hank crank film negs into your Iphone.

[carllooper]

I use Vegas 10. I've done a few tests in the past using various methods, including the one you mention.

I've got Vegas. I use it for parsing 3D video files from a Sony 3D camcorder into images I can process further. I believe Sony provide an SDK for writing plugins. I'll have a look into it.

Carl

ps. found it:
http://www.sonycreativesoftware.com/download/devkits
I'll make it a project to build a colour negative to positive plugin for Vegas

pps. the problem with a physical filter (such as a bluish filter) in the optical path of the transfer unit is that it wouldn't work. A filter blocks information rather than transforms the information. The orange cast in the negative blocks a certain amount of bluish light, in the light source, from reaching the sensor (which is why the result looks orange). If you introduce a bluish filter all that would do is block a certain amount of orange light in the light source (in addition to the blue that was blocked) ie. you would end up with with both orange and blue being blocked so next to nothing remaining transferred to the sensor. A black result.

[Nicholas Kovats]

That is as we say...wicked. I happen to use Sony Vegas 11. Go Carl.

[Tscan]

Carl that would be incredible if you could do that! It would be the first one!

[aj]

pps. the problem with a physical filter (such as a bluish filter) in the optical path of the transfer unit is that it wouldn't work. A filter blocks information rather than transforms the information. The orange cast in the negative blocks a certain amount of bluish light, in the light source, from reaching the sensor (which is why the result looks orange). If you introduce a bluish filter all that would do is block a certain amount of orange light in the light source (in addition to the blue that was blocked) ie. you would end up with with both orange and blue being blocked so next to nothing remaining transferred to the sensor. A black result.

The matter has been worked on before. Even on this forum.
This was brought up then: http://old.macedition.com/feat/film/fea ... 030626.php

It is a bit of a perpetual topic. C't Magazin für computertechnik had an explanation in 1998, heft 24 page 240

[RCBasher]

I don't know if this belongs in Roger's thread, but I already solved the negative problem with my lighting and capture system. Firstly the RGB lighting adjusts to take out most of the orange mask to allow optimal sensor illumination for minimal noise, then the capture program does a live invert and there are some fine tuning controls to adjust RGB pregains, gamma, loglin,sigSRGB, etc. such that the saved AVI or still frames are very close indeed requiring minor, if any further adjustment in post.

I'll look to post examples but I'd need permission from the source film owner first.

Frank

[carllooper]

I don't know if this belongs in Roger's thread, but I already solved the negative problem with my lighting and capture system.

That's true. Frank's system is a good solution. An RGB mixer. I'm also wrong about use of a filter not working. I was thinking of a filter placed after the light has already been filtered by the negative, but even that's wrong - before or after it doesn't matter - a filter can work just fine. See below. The only remaining issue would be whether the initial light was biased ie. (not quite white) which can upset a filter solution. An RGB light mixer becomes a good idea because it allows you to create the correct light in the first place, and adjust it as required.

For the rest of us I'll write up the Vegas plugin.

Carl

ps. whether you subtract out colours beforehand or afterwards, the effect should be the same, ie. either way would work - just need the correct colour filter (that admits more than just blue).

Here's a negative (-15, -55, -175) processing white light, resulting in an orange outcome, ie. defines the problem.

LIGHT - NEGATIVE = ORANGE RESULT

R = 255 - 15 = 240
G = 255 - 55 = 200
B = 255 - 175 = 80

Here's the same but interposing a correction filter [-160,-120,0] prior to the negative [-15, -55, -175], the result being a balanced outcome (equal amounts of red, green and blue)

LIGHT - FILTER - NEGATIVE = BALANCED RESULT

R = 255 - 160 - 15 = 80
G = 255 - 120 - 55 = 80
B = 255 - 0 - 175 = 80

And if you swap the order of the subtraction performed by a filter, with that performed by the negative, the result would be the same:

LIGHT - NEGATIVE - FILTER = BALANCED RESULT

R = 255 - 15 -160 = 80
G = 255 - 55 - 120 = 80
B = 255 - 175 - 0 = 80

Or using Franks' system:

PRE-MIXED-LIGHT [95, 135, 255] - NEGATIVE = BALANCED RESULT

R = 95 - 15 = 80
G = 135 - 55 = 80
B = 255 - 175 = 80

And testing the above logic in Photoshop gives a visual confirmation (using subtraction blend mode):



The inverting digitally:




Or in summary - TScan's idea is just fine - I was in error.

[carllooper]

I should add that a filter solution, or a light mixer solution is much better than a digital correction solution because the orange to be digitally removed is taking up a huge proportion of the bandwidth (almost 70%) so by the time you've subsequently removed it (in an NLE) you've also removed 70% of the signal that was encoded.

To put it another way, a filter, or light mixer, removes the orange prior to the remaining light being encoded, so the remaining light (the useful light) has the freedom to occupy the entire bandwidth of the sensor's range, rather than being squashed into using just 30% of it.

Carl

[Tscan]

"because the orange to be digitally removed is taking up a huge proportion of the bandwidth (almost 70%) so by the time you've subsequently removed it (in an NLE) you've also removed 70% of the signal that was encoded."

Exactly- by the time you get rid of all the blue after invert from using your color correcting tools, there's no head room left to do any more correction or sweetening of the image. The method really defeats the purpose of negative, because the advantage of negative is supposed to be lots of head room to work with.

[carllooper]

"because the orange to be digitally removed is taking up a huge proportion of the bandwidth (almost 70%) so by the time you've subsequently removed it (in an NLE) you've also removed 70% of the signal that was encoded."

Exactly- by the time you get rid of all the blue after invert from using your color correcting tools, there's no head room left to do any more correction or sweetening of the image. The method really defeats the purpose of negative, because the advantage of negative is supposed to be lots of head room to work with.

Yes. The orange bias in the colour negative is a function of the way in which colour film is manufactured to encode colour. But in film it's a self correcting system. The very bias which causes orange in a negative is also the very bias which "uncauses" it in a print. The bias cancels out. So film to film printing works just fine. But in a film to sensor pipeline, the digital sensor, insofar as it assumes a white light source (broadly speaking) can't perform this cancelling out. The sensor's colour sensitivity has to be re-jigged (basically to match that of film) by either a filter, or a light source mixer.

To put it another way, the problem with a digital sensor is that it's not film.

So back to Roger's great box. If it's to work in the new (or one should say old) age of negative, a good idea would be to have at least a slot for inserting a colour correction filter. Or even simpler: some instructions on how to mod the box for such without risking turning it into an exhibit only item.

The next model could have a switch for reversal/negative (internally moving a filter + software inversion).

And the one after that, for those with bigger wallets, could have a colour mixer built in. I'd love a a kit version you can build yourself. You get shipped the parts and instructions on how to assemble it yourself. Perfect for customisation. Not sure how this interfaces with what Roger was saying about laws regarding disposal of materials in different countries.

But of course Roger's main market is really transferring the backlog of home movies shot on reversal, (more than film makers). Still, would be nice.

Carl

[carllooper]

After a long search found this gem:

http://www.youtube.com/watch?v=-fNH3yX58Hk

The author gets the negative into a ballpark for an even range using a CC 50 Blue and CC 50 Cyan filter (overlapping) to neutralise most of the orange cast, with the remaining tweaks done in the digital stage (ie. following inversion).

If I understand the filter ratings, the effect of such a joint filter would be to cut red transmission down by two stops, the green by 1 stop, and let the blue pass, which amounts to increasing the bandwidth of green by 1 bit, and the blue by 2 bits, over any transfer that didn't do such. Or in terms of the inverted outcome: increase the bandwidth (quality) of yellow by 2 bits, the magenta by 1 bit with the cyan as is, over any transfer that didn't do such. Looking at the ratings again, it's probably better than this (a 50Blue filter corresponds to 1.3 stops rather than 1 stop, the 50Cyan corresponding to 1 stop).

Looking for CC (Colour Compensation) filters found this list:

http://motion.kodak.com/motion/Products ... en2.htm#s3

Carl

[Tscan]

The orange mask is designed for striking prints, but with most films doing digital intermediate and fewer prints showing at theaters, they may do away with the orange masking on the next update (Vision 4). There was a stock designed for digital that didn't have the mask but it kind of flopped in its time. I seem to remember John Pytlak told me once that an 80A blue filter would do the trick, it cuts 2 stops. I have some blue gels matching an 80A that could be rigged into place?

[carllooper]

The orange mask is designed for striking prints, but with most films doing digital intermediate and fewer prints showing at theaters, they may do away with the orange masking on the next update (Vision 4). There was a stock designed for digital that didn't have the mask but it kind of flopped in its time. I seem to remember John Pytlak told me once that an 80A blue filter would do the trick, it cuts 2 stops. I have some blue gels matching an 80A that could be rigged into place?

Yeah for sure. I saw the 80 series when I was looking up filters. The 80A would certainly get it closer than not using it - and the rest can be done in the digital side.

Re. Orange mask. I always understood the orange mask as a function of the way colour filtering is done in the stock. That the camera stock and the print stock have been designed in combination with each other to produce an optimum transfer between them. If so, it may mean a digital-ready stock may not be as optimum as a traditional negative, ie. in the same way that reversal is not as optimum.

Carl

[marc]

I was under the impression, from what I have heard from other sources, that the orange masking layer is what gives negative it's tremendous exposure latitude.

[carllooper]

Having become recently dubious about almost everything lately (my own assumptions the most) I'm somewhat reluctant to post this - but it reads as quite reasonable:

http://photo.net/learn/orange-negative-mask

This seems to fit with what I've understood in the past. According to the article there's a bit of a balancing act done, between neg and print, due to impurities in dyes, to create the best possible transfer of an image between them. Yellow is (apparently) the best dye so it is the primary filter used in colour separation. If required one can then compensate that by using bluer light in the system (the bluer light being easier to make pure than a blue dye in the film). The system is complicated by processing procedures which send biases this way and that, but when it's all worked out for optimum transfer, an orange cast to the negative is the result.

In other words, if I understand it correctly, it's not the orange cast that makes the film better as such. It's a better film which requires the neg have an orange cast.

Until recently I've only ever understood and appreciated the result (cinephilia), rather than the means. I've only just started (literally last week) to learn colour film processing (as distinct from cinematography and digital image processing). In the past I've just let the lab do all the required understanding on film chemistry.

My sci-fi take on film now, is that it's nanotechnology (literally) - zillions of chemical engines building an image orchestrated by electromagnetic waves and optics. I read an article in New Scientist once, about someone who built a computer by throwing together different ingredients in a coffee cup. You then threw in a question in the form of a lump of something, strirred the mixture together, and the result was in the colour of the outcome - the solution to some otherwise intractable mathematical problem.

I read film as doing the same thing.

Carl