Scanning Ektar

[Alan Edward Klein]

I'm not sure I follow you, but I tried this, starting with this image as Background Layer:

Using eyedropper tool, I copied a sample of orange film base, pasted into a new Divide layer.
Next I created an Invert layer
And finally, I created a Levels Adjust layer. There, I set white and black points and midpoints for the R, G and B channels, and arrived at this:

Is that what you meant?

The colors don't look normal. The silver looks blue and I doubt if the yellow looks like that in actuality.

 

[4season]

The colors don't look normal. The silver looks blue and I doubt if the yellow looks like that in actuality.

The LED light source I used for illumination may not be ideal for use with film. Not entirely neutral to my eyes either, it has a warmish-green tinge to it.

 

[macfred]

I scan my Ektar with Epson Scan software in manual / professional mode with Histogramm Adjustment for red, green and blue channel. The most important premise for a successful scan is a correctly exposed negative.

FUJI GA645Wi - EBC FUJINON 45mm f/4 - Ektar

FUJI GW670iii - EBC FUJINON 90mm f/3.5 - Ektar

 

[MattKing]

I scan my Ektar with Epson Scan software in manual / professional mode with Histogramm Adjustment for red, green and blue channel. The most important premise for a successful scan is a correctly exposed negative.

FUJI GA645Wi - EBC FUJINON 45mm f/4 - Ektar

View attachment 252919

With all due respect macfred, this is not a good scene to use for colour fidelity comparisons!

 

[Lachlan Young]

@4season your scan needs to be made with no autocorrection & no clipping on any colour on the histogram. The mask doesn't look right. And/ or you have crossover issues with the neg. At least that's what I found when I put it into PS.

@Les Sarile My own experience (which generally follows what E4046 states) of Ektar is that it delivers about 7 stops of straight line information, but unlike Portra, should not under any circumstances be overexposed - the blue channel highlights upsweep quite fast in density (beneficial for specular highlights) - if you don't pay attention to this, it'll screw things up badly. Relying on autocorrections & a lack of knowledge as to what Ektar should look like (think Ektachrome-ish, but with better shadows - overall neutral balance, but with saturation) lead to most of the examples shown in this thread suffering from wonky/ wrong colour, or attempts to correct that.

The mask is trivial, but adjusting curves+gamma per-channel is annoying.

If everything is linear enough, and the mask corrected for using a sample & divide method, you should be able to pull in the black point (on RGB on a curves layer, clipping on) to just clip the shadows/ rebate, then set the white point to about 150 units up the scale - ie if 'B' = 30, 'W' at 180 might not be far off - at least for something along the lines of a darkroom print. There will be mathematical reasons why this is, but I can't be bothered to work them out. Unless you have serious crossover issues, you shouldn't need (and indeed it can cause more problems than it solves) to clip the RGB individually. At this stage, you may want to do some minor colour corrections to suit the scene/ light balance of scene &/ or put a curve on it for tonal balance. Unfortunately all colour work beyond basic pre-sets requires a blended understanding of both science & art to get it to work well.

 

[Les Sarile]

@Les Sarile My own experience (which generally follows what E4046 states) of Ektar is that it delivers about 7 stops of straight line information, but unlike Portra, should not under any circumstances be overexposed - the blue channel highlights upsweep quite fast in density (beneficial for specular highlights) - if you don't pay attention to this, it'll screw things up badly. Relying on autocorrections & a lack of knowledge as to what Ektar should look like (think Ektachrome-ish, but with better shadows - overall neutral balance, but with saturation) lead to most of the examples shown in this thread suffering from wonky/ wrong colour, or attempts to correct that.

Perfectly reasonable to stay within one's defined zone. Myself, I like to know what I can work with.

 

[Bormental]

If everything is linear enough, and the mask corrected for using a sample & divide method, you should be able to pull in the black point (on RGB on a curves layer, clipping on) to just clip the shadows/ rebate, then set the white point to about 150 units up the scale - ie if 'B' = 30, 'W' at 180 might not be far off - at least for something along the lines of a darkroom print. There will be mathematical reasons why this is, but I can't be bothered to work them out. Unless you have serious crossover issues, you shouldn't need (and indeed it can cause more problems than it solves) to clip the RGB individually. At this stage, you may want to do some minor colour corrections to suit the scene/ light balance of scene &/ or put a curve on it for tonal balance. Unfortunately all colour work beyond basic pre-sets requires a blended understanding of both science & art to get it to work well.

Thank you. I will try. Right now I have two methods. The "quick & dirty" works well 80% of the time:

• Remove the mask via "mask layer divide".
• Hit "auto-levels" in Affinity Photo.
  
• Pull down the blue channel in the curves tool
• Pull down the green channel in the curves tool to place it roughly between blue & red
• Adjust master gamma

This is super quick, I even automated it, and it 80% of the time it gives me quite pleasant default. From there, I adjust saturation, white balance and I'm done in less than a minute.

But the remaining 20%... ughh. I've copy-pasted your suggestions into my notes.

 

[Tom Kershaw]

This is super quick, I even automated it, and it 80% of the time it gives me quite pleasant default. From there, I adjust saturation, white balance and I'm done in less than a minute.

You might want to try ColorPerfect - the download (Photoshop / PhotoLine plug-in) works as a demo version so you can try out the conversion. It gives different results compared to curve manipulation; perhaps at somepoint I'll do a detailed comparison. www.colorperfect.com

 

[macfred]

With all due respect macfred, this is not a good scene to use for colour fidelity comparisons!

The skin tones are perfect, aren't they ?

 

[Bormental]

You might want to try ColorPerfect - the download (Photoshop / PhotoLine plug-in) works as a demo version so you can try out the conversion. It gives different results compared to curve manipulation; perhaps at somepoint I'll do a detailed comparison. www.colorperfect.com

Tom, I migrated off Lightroom/Photoshop after Adobe stopped selling them, can't use plugins

 

[Tom Kershaw]

Tom, I migrated off Lightroom/Photoshop after Adobe stopped selling them, can't use plugins

I'm using ColorPerfect hosted within PhotoLine - www.pl32.com - I stopped Adobe a while ago as well.

 

[albireo]

Example inversion using Vuescan's advanced workflow, as mentioned in my earlier post, quoted below.

Agfa Isolette I 6x6 folding camera
Agfa Agnar 85mm f/4.5 triplet lens
Epson V550
Ektar 100 120

I use Ektar in medium format almost exclusively, and find it easier to scan than Portra 160, and about as easy as Fuji 400H.

I use Vuescan and would recommend inverting using the 'Vuescan Advanced Workflow' as summarised on the following page

http://benneh.net/techshit/better-colour-neg-scanning-with-vuescan/

Works wonderfully well giving me superb colours (even though the author says he doesn't use the method anymore)

 

[GLS]

You might want to try ColorPerfect

Agreed. It is what I have used for all my C41 colour inversions (from DSLR scans) for a while now. It works very well, including with Ektar.

 

[Helge]

The mask varies with the image - the differing dyes present in the image result in a mask that differs in that part of the negative.
If there is cyan dye at location in the negative (which corresponds with red in the subject) the mask will be different than if there is blue dye at another location in the negative (which corresponds with yellow in the subject).
The mask doesn't just vary with density, it varies with subject colour.
For that reason the sampling done as part of the approach that uses inversion layers and a divide blend doesn't accurately neutralize the mask.

Not to put to fine a point on it, but AFAIK the only correction needed for RA4 prints are the filter pack.
The RA4 paper doesn’t to my knowledge contain any correcting response curves in the different layers for the orange mask.

That means that a simple global neutral filter should be enough.

The mask does vary over the image, but so does the wrong response in the dyes it corrects.
So it all comes to linear global cast.

Correct?

 

[MattKing]

The mask does vary over the image, but so does the wrong response in the dyes it corrects.
So it all comes to linear global cast.

Correct?

Correct.
I'm not sure though that sampling in the rebate area provides the necessary correction.

 

[Bormental]

It does. Because division or multiplication (you can do both, its the same thing) lead to the desired result. The masked areas are affected by a greater/smaller degree as opposed to subtraction. Lachland recently sent me a RAW negative scan from an Imacon and it's been a breath of fresh air to experiment with it (as opposed to camera capture). You really see how colors "snap" into place when the mask is correctly removed.

 

[MattKing]

I'm concerned about the sampling point - an area with no imaging dyes - not the method.

 

[Bormental]

Yes that's a valid concern. What I usually do is to take a separate shot of the blank area on a roll, then crop it to have 100% orange mask on it, blur it for a good measure and sample that (using that value for that one roll). However, what I later found is that minor variations in sampling are not even noticeable in the end result, so perhaps I've been too pedantic here. I am now on my 5th day of researching this and right now my attention is 100% on light sources. CRI of 95 may not be good enough for DSLR scanning.

 

[MattKing]

I am now on my 5th day of researching this and right now my attention is 100% on light sources. CRI of 95 may not be good enough for DSLR scanning.

Just use a tungsten or halogen source or an electronic flash - both have a CRI of 100.
Other sources will have discontinuities in the spectrum, no matter how high the CRI. It is the discontinuities that create the problems.
With respect to the sampling, it might be useful to have a standard, re-usable, unfocused grey card image file from each type of film scanned, and to use that file each time for the division. I expect that is what the film profiles essentially do.

 

[Bormental]

You've guessed it. I am building a contraption with a purpose is to combine the flash and my copy stand. I include a shot of a color target under 1pm sunlight (easy here in CA) with grey patches into every roll. It doesn't make any difference. Sampling the mask off 100% unexposed area is what you want. The grey patches only help you visually confirm that you're doing it right, but honestly it's hard to make a mistake with such trivial operation.

 

[Helge]

It does. Because division or multiplication (you can do both, its the same thing) lead to the desired result. The masked areas are affected by a greater/smaller degree as opposed to subtraction. Lachland recently sent me a RAW negative scan from an Imacon and it's been a breath of fresh air to experiment with it (as opposed to camera capture). You really see how colors "snap" into place when the mask is correctly removed.

How do you explain then that regular RA4 printing is able to remove the mask with simple
subtraction from a static filter?

 

[MattKing]

How do you explain then that regular RA4 printing is able to remove the mask with simple
subtraction from a static filter?

Most likely because an optical subtraction using a filter is equivalent to using division when merging two layers using post-processing software such as Paintshop Pro and Photoshop.

 

[Bormental]

How do you explain then that regular RA4 printing is able to remove the mask with simple subtraction from a static filter?

There are two separate steps involved:

1. Correct for the color bleed/leak between channels.
2. Correct for the tint.

In more detail:

Step 1. The "division/multiplication" happens as light goes through the negative (it gets "divided" by density in each channel). After this process there is no more color bleeding between channels, but you still have blue+green channels overwhelming the red.

Step 2. The paper then uses different (inversely proportional) sensitivity of its CMY layers to correct for the blue/green cast.

Now, when digitizing:

1. This step is done by sampling the mask and doing division or multiplication.
2. This step is done by adjusting black/white points of R/G/B channels.

 

[Helge]

Most likely because an optical subtraction using a filter is equivalent to using division when merging two layers using post-processing software such as Paintshop Pro and Photoshop.

That doesn’t make sense to me.
Whether you do it digitally or optically the math/process should be the same.

 

[Bormental]

Helge, it really is the same process, see my comment above. You're right, the paper doesn't "divide" anything. The division happens as the light goes through the negative in an enlarger.