Page 2 of 3 FirstFirst 123 LastLast
Results 11 to 20 of 29

Thread: Flextight Colour Neg Scanning

  1. #11

    Join Date
    May 2017
    Posts
    232

    Re: Flextight Colour Neg Scanning

    Quote Originally Posted by interneg View Post
    Just to annoy those blathering on elsewhere in this thread about IT8's etc, it's possible to match a colour checker (even from an Ektar neg - allowing for the saturation difference!) from a Flextight scan without too much trouble.
    Why would you want to annoy anyone posting here

    IT8.7/2 is for reflective materials to calibrate for the dyes in the print, which are designed on purpose to work under a wide variety of illuminates in the first place so perhaps good enough compared to a macbeth chart which is designed specifically for colorimetric purposes. IT8.7/1 is for transmission targets were the illuminant is more critical, and the dye sets are more likely to suffer from Metameric failure when using the wrong illuminant, they trade this for greater saturation and gamut. Metameric failure is the main reason you need a target with the correct dye set.

    No ones arguing that these targets are not useful, and it is possible to match something that already exists like a negative if you take colormetric measurements of the negative, but it is not quite as easy to match something that does not yet exist such as the resultant print. The calculations and methods being largely proprietary.

  2. #12

    Join Date
    Jul 2016
    Posts
    4,566

    Re: Flextight Colour Neg Scanning

    Quote Originally Posted by Ted Baker View Post
    The calculations and methods being largely proprietary.
    I'd say that the proprietary side of the methods are mostly because practical issues, rather because the calibration nature.

    A 8 bit per channel would require 3D LUT with 256 positions in each direction of the cube, this is 256^3 so some 16 mega cells, each with 4 bytes (3 + 1 dummy because memory alignment ), so 64 MB. This is not much for present computers.

    A 16 bit per channel 3D calibration would require 65536^3 x 8, which is an amount that is crazy big number.

    In the 8 bits per channel calibration you fill the 3D matrix with data from the calibration, from the scanner reading of each color in the calibration you get the x,y,z coordinates and you fill that cell with the reference values in the target.

    Then we have to fill the rest of the cells that were not set with calibration data, this is all the matrix but the calibrated reference points. This process is a common interpolation/extrapolation that takes data from sorrounding places that have reference values, that interpolation can be from a weighted (upon distance) average to splines.

    In this way any RGB combination from the sensor would have and output value in the calibration 3D LUT. So that's straight

    Problem happens when wanting a 16 bits per channel calibration, we cannnot use such an inmense LUT !!

    There are 2 ways I know for that. One is going to a logarithmic scale in a way that it does not degradate the calculations for the low values.

    Another one is having a "virtual" 16 bits LUT in a 8 bits LUT, this is moving the calibrated point to a cell rounding the 16 bits value to a 8 bits RGB input value, and then calculating what it would be the reference (16 bits per channel) output in the new position, by interpolation.

    With that calibration LUT for each 16 bits per channel RGB entry point we interpolate between 8 bits per channel entry points that have 16bits per channel output values, this works perfect.

    I've been typing that code in c++... for 16 bits per channel data with 8 bits LUTs in PCs, and or 8bits data with 5 bits per channel LUTs (32x32x32 levels) for embedded uC.

  3. #13

    Join Date
    May 2006
    Location
    Sydney, Australia
    Posts
    24

    Re: Flextight Colour Neg Scanning

    Quote Originally Posted by Pere Casals View Post
    If "sharpness does seem vary from neg to neg" this can be because curled negatives, you first can try to address that by trying wet mounting or by simply purchasing the V850 film holders that have an ANR glass to ensure film flatness and also to adjust film height.
    I was looking at the ANR glass holders but I can only assume that on a 4x5 neg, they will be magnets for dust and lint. Of course you could fluid mount them as well, but then you have issues with air bubbles and dust etc on the mylar film. Have you had any experience with them?

    Quote Originally Posted by interneg View Post
    In Photoshop you want to use ProphotoRGB colourspace, sample the base density, use that to fill a new layer set to divide - then flatten, & invert. Using a curves layer & clipping warnings, clip black and white points in each of RGB. Now you'll have something that is very close to accurate reproduction of colour within fine colour corrections (curve layer set to 'colour' blending) and a curve layer set to 'luminosity' blend mode for adjusting tonalities.
    That's a great workflow for processing colour neg scans, thank you very much for that advice! My only question about this is that if you're using Flexcolor to scan with, wouldn't it be better to do this on the 3F raw file instead?

  4. #14

    Join Date
    Jul 2016
    Posts
    4,566

    Re: Flextight Colour Neg Scanning

    Quote Originally Posted by yelmarb View Post
    I was looking at the ANR glass holders but I can only assume that on a 4x5 neg, they will be magnets for dust and lint.
    To scan (and other tasks), I'd recommend having a really dust free environment. I scan in an small room that only has the scanner and a laptop, I clean air with an HEPA air cleaner: Honeywell 16200. I start it 10 min before and I try to not generate dust from clothes while scanning. No dust.


    Quote Originally Posted by yelmarb View Post
    Have you had any experience with them?
    This is straight but it is matter of practice and watching some tutorials:

    https://www.youtube.com/results?sear...ount++Tutorial

  5. #15

    Join Date
    Jun 2015
    Location
    Glasgow
    Posts
    1,009

    Re: Flextight Colour Neg Scanning

    Quote Originally Posted by yelmarb View Post
    That's a great workflow for processing colour neg scans, thank you very much for that advice! My only question about this is that if you're using Flexcolor to scan with, wouldn't it be better to do this on the 3F raw file instead?
    My own experience is that the 3F is just a straight uninverted unadjusted .tiff file with a proprietary file name - and that making an adjusted 16-bit .tiff in Flexcolor gives a far better inversion - possibly for reasons to do with the analogue to digital conversion or something like that. Essentially, what I am doing in Flexcolor is making the file that a 3F should be, not what it actually is. After all, how many drum scanners offer anything other than a .tiff or similar file? Can't think of any that use proprietary file formats.

  6. #16

    Join Date
    May 2017
    Posts
    232

    Re: Flextight Colour Neg Scanning

    Quote Originally Posted by Pere Casals View Post
    I'd say that the proprietary side of the methods are mostly because practical issues, rather because the calibration nature.
    I think the proprietary side starts with entire analog process from Kodak, fuji and not sure who is left. Neither of them publish the details really needed, though the process is explained in a few books by Kodak employees on how the the negative/positive system has been designed. The best by far is Digital Color Management:Encoding solutions by Giorgianni and Madden it is not cheap but worth it if you want to understand the process better. On the software side most of the solutions aren't all that well published. BTW The process that interneg refers is almost the same as used by colorperfect (who do publish their algorithm). IMHO its a very sound principle but misses a few calibration steps.

    Plus of course in the hay day of analog, high end scanning was primarily focused on chromes, photographers who shot negative stock tended to make prints...

    But back to calibration , putting aside any computational difficulties in the case of a negative/positive system what would you actually calibrate?

  7. #17

    Join Date
    Jul 2016
    Posts
    4,566

    Re: Flextight Colour Neg Scanning

    Quote Originally Posted by Ted Baker View Post
    Neither of them publish the details really needed
    Let me tell my view.

    IMHO the critical process is at exposure time. Then you take spectral information hitting the negative in each image point and you convert that SPD in each point to 3 values. After doing the 1st development of the C-41 you get in the negative 3 values of silver density, one in each of the 3 color layers. This is where true metamerism takes place !!!! In the same way a digital sensor converts the spectum to 3 RGB voltage levels in the bayer pixels. Also the human eye takes the spectrum hitting a point and makes a phototransduction (https://en.wikipedia.org/wiki/Visual_phototransduction), RGB chromophores in the RGB cones filter the spectrum and that ends in "3 electrical signals".

    Kodak and Fuji publish that critical data, Portra 160:

    Click image for larger version. 

Name:	____.jpg 
Views:	7 
Size:	48.5 KB 
ID:	179470

    Then you make the color development stage, and silver is bleached. Those coupler dyes only work like an ICC profile, no information is lost at this stage, spectral information was lost (reduced to 3 exposure values) at the exposure time, the dye coupling only cooks the metameric information for a nice result, that's also linked with the RA-4 particular paper interpertation.

    What I say is that there is no universal calibration than would make Velvia 50 portraits look as if Portra 160 was used, because the captured information has reduced spectrums to 3 color levels.

    In the same way when we have the negative developed we are virtually playing with the metameric conversion done, the spectrums in the negative are from 3 overlaped curves that can be higher or lower, but always having a similar shape, depending on the coupling dyes the film uses, so at the end we have 3 virtual levels for each point.

    We are not playing anymore with spectrums hitting every image point, but with reduced metameric information.



    Quote Originally Posted by Ted Baker View Post
    what would you actually calibrate?
    IMHO we have 2 cases.

    a) Slides are a reference medium. A calibrated hybrid process should deliver an sRGB file that shown in a calibrated monitor would be as close as possible to the slide viewed directly. Here we have a problem because some Velvia tones cannot be seen in a monitor. Also the Velvia slide can be 3.4D while a monitor usually has 2.0D static contrast (and 200000000000000:1 dynamic contrast )

    b) C-41 is an industrial process intended to end in a print, but we have a necessary post-process to adjust the result. So IMHO for an scanned color negative we haven't a reference. IMHO single choice we have is selecting a nice reference for each film type. This would be taking sound prints from common shots and making the calibration from the print, or from a good digital edition of the scan, adding perhaps an smart equalization.

    A C-41 negative to positive conversion could have some options: Standard, Landscape, Wedding José Villa, Vivid, Soft...

  8. #18

    Join Date
    May 2017
    Posts
    232

    Re: Flextight Colour Neg Scanning

    Quote Originally Posted by Pere Casals View Post
    Let me tell my view.

    IMHO the critical process is at exposure time. Then you take spectral information hitting the negative in each image point and you convert that SPD in each point to 3 values. After doing the 1st development of the C-41 you get in the negative 3 values of silver density, one in each of the 3 color layers. This is where true metamerism takes place !!!! In the same way a digital sensor converts the spectum to 3 RGB voltage levels in the bayer pixels. Also the human eye takes the spectrum hitting a point and makes a phototransduction (https://en.wikipedia.org/wiki/Visual_phototransduction), RGB chromophores in the RGB cones filter the spectrum and that ends in "3 electrical signals".

    Kodak and Fuji publish that critical data, Portra 160:

    Click image for larger version. 

Name:	____.jpg 
Views:	7 
Size:	48.5 KB 
ID:	179470

    Then you make the color development stage, and silver is bleached. Those coupler dyes only work like an ICC profile, no information is lost at this stage, spectral information was lost (reduced to 3 exposure values) at the exposure time, the dye coupling only cooks the metameric information for a nice result, that's also linked with the RA-4 particular paper interpertation.

    What I say is that there is no universal calibration than would make Velvia 50 portraits look as if Portra 160 was used, because the captured information has reduced spectrums to 3 color levels.

    In the same way when we have the negative developed we are virtually playing with the metameric conversion done, the spectrums in the negative are from 3 overlaped curves that can be higher or lower, but always having a similar shape, depending on the coupling dyes the film uses, so at the end we have 3 virtual levels for each point.

    We are not playing anymore with spectrums hitting every image point, but with reduced metameric information.





    IMHO we have 2 cases.

    a) Slides are a reference medium. A calibrated hybrid process should deliver an sRGB file that shown in a calibrated monitor would be as close as possible to the slide viewed directly. Here we have a problem because some Velvia tones cannot be seen in a monitor. Also the Velvia slide can be 3.4D while a monitor usually has 2.0D static contrast (and 200000000000000:1 dynamic contrast )

    b) C-41 is an industrial process intended to end in a print, but we have a necessary post-process to adjust the result. So IMHO for an scanned color negative we haven't a reference. IMHO single choice we have is selecting a nice reference for each film type. This would be taking sound prints from common shots and making the calibration from the print, or from a good digital edition of the scan, adding perhaps an smart equalization.

    A C-41 negative to positive conversion could have some options: Standard, Landscape, Wedding José Villa, Vivid, Soft...
    Pere, I think you are wide of the mark on a few points when talking about a negative/positive system. Digital Color Management:Encoding solutions by Giorgianni and Madden is really worth the effort if you want to understand the science of negative scanning properly (1st edition is cheap second hand). The Reproduction of Colour in Photography, Printing and Television by Hunt while old and pre-dates scanning still explains some of the important differences between a chrome and negative.

    For example the RA-4 paper has different spectral sensitivity than our own eyes, its peak sensitivity to the cyan forming layer is different. As an example, this can be seen in the characteristics curves which are measured used status-m (which is an arbitrary standard a little closer to our own eyes and a typical scanner...) and results in non-parallel lines. When a negative is printed using RA-4 the resultant printing density results in parallel lines for all three layers (assuming the negative is a of neutral grey subject shot in the correct illuminant).

    In fact the negative/positive system is also subject to it own Metamerism, unique to the negative/paper combination. i.e. two colors on the negative might appear different to our own eyes but the same to the paper.

    I don't for one moment think that kodak/fuji who understand these relationships and almost certainly done many print density calculations and have developed system's such a PhotoCD, Cineon, fuji frontier, could not produce targets and tools that could help other vendors, or even photographers, hence my comment about it being largely proprietary ...

  9. #19

    Join Date
    Mar 2008
    Location
    Seattle
    Posts
    118

    Re: Flextight Colour Neg Scanning

    Quote Originally Posted by cdavis324 View Post
    You can always scan as "raw" and use colorperfect to invert the negative. It provides the best results with the least amount of work with almost any scanner. The flextight software is powerful, but can be time consuming to get the right color... You really have to go in and set curves, endpoints and such for each image to get it right with flexcolor. Colorperfect alleviates a lot of that work!
    I did a whole show (18 prints) doing pretty much that. After the first couple, I used some presets in the FlexColor software and that got the negs inverted and most of the way there before I pulled them into PS for final color correction. I then dusted and resized output for lightjet prints (30x40 and 40x50) and they looked phenomenal. I tried using the X5 to produce .tiffs and I was never able to get them to color correct properly and they didn't seem as sharp so I went back to 3F.

    I think this might be along the lines of what Bob Carnie was talking about as well.

  10. #20

    Join Date
    Jul 2016
    Posts
    4,566

    Re: Flextight Colour Neg Scanning

    Quote Originally Posted by Ted Baker View Post

    Pere, I think you are wide of the mark on a few points when talking about a negative/positive system.

    In fact the negative/positive system is also subject to it own Metamerism, unique to the negative/paper combination. i.e. two colors on the negative might appear different to our own eyes but the same to the paper.
    Ted, I think I've not expressed well what I wanted to say. Let me try again.

    It is true that the developed negative has an spectral nature, and each stage (negative, RA-4, display illumination) has it's own metamerism... no doubt.

    But let's speak about encoded color information, because this has an impact in how we can calibrate a system.

    We have CMY layers in the developed negative, let's take a layer, the Magenta for example of the Portra 160.

    Can an spot in the M layer have any transmission spectrum ? No !!!! We only have a collection of possible spectrums

    For each silver density (exposure) in the M layer (after 1st developer) we will have a unique corresponding transmission spectrum in the M layer.

    This is a Bijective function (https://en.wikipedia.org/wiki/Bijection), one to one, from a linear space of densities (real numbers) to an space of functions (spectrums). Same hapens when the negative is RA-4 printed.

    If you apply advanced math to that situation it can be demonstrated that a 3D LUT is exactly what maps the changes in the metameric information in all processes happening after exposure. This includes the color development result in the C-41, an RA-4 printing process, scanning and color profiles, palettes and creative 3D LUTs. (Of course this does not consider local edition, dodging, masks etc, only general transformations in the color spaces).

    The books you point are really interesting, but those explains (amazing) how color was managed by analog means.

    In the hybrid process we have the benefit of digitaliazation that allows for easy information abstraction.

    Let me point an hybrid case, Star Wars 7 and 8 was shot in film, and the first thing they did was scanning the film, they made the digital process, and finally they also printed the film for the IMAX release. All the digital process was made in a RGB space, ending in a printed film that had an spectral complexity like the negative had, but nothing was lost when the spectral information of the Vision 3 negative was reduced to 3 numbers. Any lost information was lost at the exposure time.

Similar Threads

  1. Replies: 18
    Last Post: 27-Mar-2015, 13:25
  2. Scanning colour negatives
    By AbsolutelyN in forum Digital Processing
    Replies: 1
    Last Post: 17-Aug-2014, 08:39
  3. Help/advice Scanning w/ Hasselblad Flextight X5
    By DennisD in forum Digital Processing
    Replies: 11
    Last Post: 21-Sep-2013, 12:55
  4. New Scanning Service with Flextight X5
    By John Hanses in forum Resources
    Replies: 4
    Last Post: 25-Mar-2011, 13:25
  5. ColorNeg: NegPos successor for colour neg scanning
    By Leigh Perry in forum Digital Processing
    Replies: 24
    Last Post: 1-Jul-2009, 00:02

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •