Camera "Scanning" -- Use Di-Chro Head to undo Color Mask?

[jzagaja]

First digital cameras 17 years ago were appying only gamma to linear capture but later they tried to simulate slide film and now even negative film (high dynamic range) so transfer function is rather unknown unless you use RAW converter and set to proper transfer function. Tonality/contrast is a one thing but color variations are much stronger and depend on light spectra used for illumination, camera colorimetry and RAW converter. Inversion curve you presented is for standard jpeg output from camera with assumption it is true gamma 2,2 without slope limiting. Was generated in my plugin. I tried 80A filters with tungsten on drum scanner but there was little difference. When you have good sensor with high dynamic range than you can correct in post without problem.

 

[aizan]

what's the dichro head + flash + camera setup look like?

 

[ColSebastianMoran]

what's the dichro head + flash + camera setup look like?

I am using a Beseler Dual-Mode Duplicator. Has an illumination system with dials for color equivalent to a DiChro color head. Also had incandescent or flash illumination.

I use flash and +50 Cyan +15 Magenta to almost neutralize the orange mask in unexposed areas (e.g., rebate between frames).

Here's the setup:

DiChro controls are below, out of the frame. The gizmo "Vello" is a remote cord to trigger the flash in the Beseler.

 

[ColSebastianMoran]

Jack's Curve (Transform & Invert) is a Winner!

Jack's Curve (Transform & Invert) is a Winner!

Over next couple days will make some comparisons, but initial tests appear promising, possibly better than the other's I've tried.

Here's the process
- Illuminate a color neg with +50 Cyan +15 Magenta using 5000K flash
- WB from rebate (it is 4350K and -8 tint with my setup, with one typ film)
- RAW conversion with default settings in LR, Adobe "Camera standard" calibration
- Set White and Black points in PS (Levels)
- Compare a) Simply Invert vs. b) Transform and Invert with Jack's curve (Curves, custom preset in PS)
- Curves Option-Auto with Huss's options: Find Dark/Light, Snap Neutrals

With simply Invert:

With Transform and Invert using Jack's curve:

My conclusions:
- Note: this is a test of automatic processing; neither is a finished image
- I can fix either one to be a pretty good image
- The first will require more adjustments and selected area adjustments to fix the strange tonality (e.g., granite is simply wrong and has color cast)
- The second is the closest I've seen to a good automatic result from color neg film.

Thanks again to Jack and Huss. I encourage others to try Jack's custom Transform & Invert curve for PS.

 

[ColSebastianMoran]

First digital cameras 17 years ago were appying only gamma to linear capture but later they tried to simulate slide film and now even negative film (high dynamic range) so transfer function is rather unknown unless you use RAW converter and set to proper transfer function. Tonality/contrast is a one thing but color variations are much stronger and depend on light spectra used for illumination, camera colorimetry and RAW converter. Inversion curve you presented is for standard jpeg output from camera with assumption it is true gamma 2,2 without slope limiting. Was generated in my plugin. I tried 80A filters with tungsten on drum scanner but there was little difference. When you have good sensor with high dynamic range than you can correct in post without problem.

Jack, I cannot argue against any of that.

In my experiments with Camera-scan of color negative, there are some color issues and tonality issues. Your curve seems to help a lot with the tonality issues.

As for color, every film is different, so is every camera. I had been ignoring this, using the simple "Adobe Standard" calibration. In the past couple days, I've learned that "Camera Standard" or a custom camera profile are better, but there is still room for personal preference.

You say your curve was created for jpg at Gamma 2.2 and I'm using it to transform and invert from a RAW conversion in Photoshop. Do you think my use of your curve is consistent and OK, or would you recommend a different curve.

 

[aizan]

I am using a Beseler Dual-Mode Duplicator. Has an illumination system with dials for color equivalent to a DiChro color head. Also had incandescent or flash illumination.

I use flash and +50 Cyan +15 Magenta to almost neutralize the orange mask in unexposed areas (e.g., rebate between frames).

Here's the setup:

DiChro controls are below, out of the frame. The gizmo "Vello" is a remote cord to trigger the flash in the Beseler.

interesting piece of equipment. looks like prices are all over the place ($50-$550). i take it you're using a small flash unit that can fit in the space where the incandescent bulb goes? do the filters let you correct the color even if you don't use a flash?

 

[jzagaja]

My curve sets middle grey correctly and if your camera adds S-curve you will detect it visually. Idea behind my plugin was to automatically batch scan negatives with predefinded white/black point and if necessary correct each frame later in post. It was used in one DTP studio for many years with Sony scanners. It could be now improved with AI algorithms, scopes etc. used in grading soft like Davinci Resolve but must be rewritten in another language.

From my experience Porta was the only film I had troubles with grading. Now it is improved with Kodak Vision cine film.

 

[ColSebastianMoran]

interesting piece of equipment. looks like prices are all over the place ($50-$550). i take it you're using a small flash unit that can fit in the space where the incandescent bulb goes? do the filters let you correct the color even if you don't use a flash?

Aizan, it's an interesting and flexible device created originally for slide copying. The "Dual" in the name is presumably incandescent vs. flash.
- Good, solid fixtures for camera mounting
- Illuminates a 6x6cm opening
- Mount a plate that takes Beseler enlarger negative carriers
- Mount a box that nicely illuminates a 4x5" opening for 6x9 or large format
- DiChro filters
- Switchable illumination: either incandescent (3000K) or flash (5000k)

Note the flash is high voltage. Use a Wein Safe-Sync.

I thought I would just mount a small modern flash in the base, but that looks very difficult. You need to continuous illumination to focus and position.

 

[ColSebastianMoran]

Here's one more sample, shot in Tokyo on an overcast and rainy day in 2010.

Simply Invert, then Curves Auto w/options:

Transform and Invert using Jack's curve, then Curves Auto w/options:

My take:
- Note: these are just "auto" not finished images. Both start with 16 bit file.
- Could make a finished image from either; I worked on the first one to produce a usable image (not shown), but that was quite difficult, the tonality is simply "off" and needs a lot of correction.
- Second (w/Jack's curve) is much closer to "right" and quite close to the mini-lab scan for color. For tonal range, the second is better than the mini-lab scan. Already "close enough" for some uses and much easier to create a finished image from this one.
- While the dark areas are about the same, the midtones and bright areas are much different. That's result of the transform curve.

Conclusion: With these two tools (Jack's curve and Curves Option-Auto Find Dark&Light and Snap Neutral Midtones), I can make "auto" processed camera scans that will be quite usable. Close to and perhaps better than mini-lab scans.

Edit: ColorPerfect PS filter also does a non-linear inversion. At this point, I recommend either Jack's curve or ColorPerfect for camera-scans of color-neg. See this thread from July 2017.

And, this thread from Nov 2017.

 

[ColSebastianMoran]

Update: I've done more experiments with DiChro filtration trying to neutralize the orange mask in C-41 negatives. I see some color strangeness when I add color filtration.

My conclusion: Don't bother. I think electronic flash alone is the best illumination for camera-scan.

My current best solution: good macro lens, electronic flash to illuminate the negative, Negative Lab Pro by Nathan Johnson to invert the image.

 

[ColSebastianMoran]

Here's the comparison. Good macro lens, flash illumination, no-filter (left) vs. Cyan-Magenta filter (right).

 

[jzagaja]

Can make comparison: 1030Ai drum scanner vs. Fuji XT2 on Leica Reprovit.

 

[Ronald M]

After getting it to photoshop, color can be corrected by doing RGB each layer separately.

Open in PS, add a levels layer, from channels open R channel and set blacks to 10, whites to 245, repeat for G and B layers. Open color channel and it will have perfect color .

Do the thing to get rebate to shoot as neutral before going to PS. This balance works for color balance required . Takes time, but works every time.

 

[ColSebastianMoran]

Update -- I'm now liking Negative Lab Pro for Camera-scan conversions of color negative film. There's more on NLP in other threads. The image below is fully automatic in NLP, with only manual a brightness adjustment. Original negative on Fuji-200 exposed in direct mid-afternoon sunlight. Camera-scan with Nikon D-7200.

I wanted to test again the idea of using my DiCho filtration in making the digital camera-scan. As earlier, I think the color is better with straight electronic flash than with flash+dichro filter.

The difference is subtle and I see it mostly in the reds (I find the reds are the hardest to get right in color-neg inversions). Here's the comparison.

 

[ColSebastianMoran]

As before, the reds are muted when I use the di-chro filtration.

 

[predicolous]

thank you for the comparison. I also like the colors without Di-Chro a lot more!

 

[ColSebastianMoran]

The rationale for using the di-chro filter is to bring the blue histogram up. Without the filter, with ETTR for the red channel, the blue channel lags and could produce noise when inverted.

I cannot find any evidence of benefit in using the filter, and for whatever reason, the colors are worse. My net is:
- The range of densities in a color negative is no challenge for today's raw-captures
- I do not know what causes the color weirdness with the filter