Ad
Ad
Ad
Pages: [1] 2 »   Bottom of Page
Print
Author Topic: LUT for transforming Negative film to positive  (Read 4255 times)
SZRitter
Full Member
***
Offline Offline

Posts: 156


« on: February 28, 2014, 09:22:05 AM »
ReplyReply

Has anyone seen a LUT for transforming negative film to positive? I am using a DSLR to digitize film (mostly Kodak Portra 160) and this is my last hang up to getting really good images.

If no one can point me in the right direction, Do you think it would work to use a color checker and create a LUT from that? Or even creating a camera profile to just use it automatically with Lightroom?
Logged
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6880


WWW
« Reply #1 on: February 28, 2014, 09:50:16 AM »
ReplyReply

Given the highly variable and special characteristics of negative films I doubt either of those approaches are workable possible and never seen any such user-experience. However, you can develop an Action usable in Photoshop for producing satisfactory negative to positive conversions. Though dated, you may find some useful guidance in my article, particularly for the steps within Photoshop: http://www.luminous-landscape.com/techniques/scanning_colour_negatives_raw_or_not.shtml. Once you do the basic conversion in Photoshop you can import the TIFF or PSD to Lightroom for further work if you prefer using Lightroom.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
SZRitter
Full Member
***
Offline Offline

Posts: 156


« Reply #2 on: February 28, 2014, 11:32:09 AM »
ReplyReply

I just tried your method using Photoshop to get the LAB color for the film base, and it left a very Cyan tint. Not sure if I missed something or not. That said, due to how I held my negatives down (taped onto the surface), I don't have a consistent light. I'm going to try a different method (negative carrier) to hold the negatives and see if that helps.

And am I wrong in thinking that a scanner uses a LUT to transform the different film bases?
Logged
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6880


WWW
« Reply #3 on: February 28, 2014, 11:42:17 AM »
ReplyReply

Scanner software such as SilverFast or Vuescan come bundled with specific film profiles for a large number of different emulsions.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3550


« Reply #4 on: February 28, 2014, 11:43:03 AM »
ReplyReply

Has anyone seen a LUT for transforming negative film to positive? I am using a DSLR to digitize film (mostly Kodak Portra 160) and this is my last hang up to getting really good images.

If no one can point me in the right direction, Do you think it would work to use a color checker and create a LUT from that? Or even creating a camera profile to just use it automatically with Lightroom?

Hi,

A simple LUT will not do it! CN film varies with each processing run.

1. You need to start with a linear gamma capture.
2. Then adjust the White points (deepest shadows) of the three channels, preferably on inter-image blank film (or un-exposed leader), which will neutralize the Mask color.
3. Then adjust the Black points (highlights) of the three channels to render neutral areas as neutral.
4. Invert tonality, negative becomes positive.
5. Apply tone-curve and color adjustments to taste, for which a LUT can be used to create uniform results.

To get the best results, try to neutralize the film base+mask color as much as possible by adjusting the color of the exposing light-source, which will reduce noise in the Blue and Green channel capture, and only fine-tune in step 2.

Cheers,
Bart
Logged
SZRitter
Full Member
***
Offline Offline

Posts: 156


« Reply #5 on: February 28, 2014, 01:03:47 PM »
ReplyReply

I guess I should have said I wasn't looking for a single LUT, but I was just looking for one for the Porta to try.

That said, using a combo of Both of yours, I came out with fairly accurate colors. So, all in all, a success. My thought is if I can export the base curve off of one frame, I should be able to apply it to the whole roll, or rolls (assuming they came out of the same pro-pack and development). Thank you!
Logged
WayneLarmon
Jr. Member
**
Offline Offline

Posts: 50


« Reply #6 on: March 03, 2014, 08:29:47 PM »
ReplyReply

I've been picking away at doing this for a while but don't have a foolproof method that doesn't involve a lot of ad hoc curve twisting.  I shoot RAW and use CS6 ACR.  My first question

Quote
1. You need to start with a linear gamma capture.

OK, I don't understand this.  Or does "linear gamma capture" just mean the RAW file that ACR opens.  Anyway, I open the RAW file in ACR and head straight for (what is now known as) the "Point Curve Tool" in ACR.  Then I adjust the endpoints of the R, G, and G curves and then invert them.  (Well, actually I invert them first.)  Which is what I think you described with your points 2. through 4.  Then I export to CS6 as 16 bit ProPhoto.  I then go into CS6's Curve tool and whale away twisting the R, G, and B curves (and usually try one of the "Auto" options in Curves to see if it auto-magically finishes the conversion.  Sometimes magic happens and I'm done.)  If I know that I have other frames that are similar, I save the Curves settings so I can use it on other images.

This last step is I think what you describe as point 5. on your list.  Correct?

Quote
To get the best results, try to neutralize the film base+mask color as much as possible by adjusting the color of the exposing light-source, which will reduce noise in the Blue and Green channel capture, and only fine-tune in step 2.

This I would love to do, but don't know how to do it.  I'm using high CRI daylight balanced CFLs.  I tried using Rosco #3202 tungsten-to-daylight gel in between the CFLs and the negative in an attempt to compensate for the orange mask optically, but I ended up with more residual noise in the converted image than if I just used the CFLs straight and compensated for the orange mask in software (namely, the first step of adjusting the R, G, and G curves in the ACR point curve tool.)

I'm a little queasy about using CFLs, but I've found that I can eliminate a lot of dust and scratches by starting out with 400 watts (equivalent) of CFLs in a softbox and adding several additional stops of diffusion.  I can't get that amount of light with anything tungsten.   I tried a 4900k Solux bulb but that didn't really help the color conversion and it didn't produce enough light for diffusion--I spend hours cloning out scratches if all I use is a single Solux bulb.  A strobe isn't an attractive option, because I do manual focus with 10x liveview.

Ah, the reason I started this thread was because of me trying to come up with a better light source for camera scanning.
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3550


« Reply #7 on: March 04, 2014, 03:14:08 AM »
ReplyReply

I've been picking away at doing this for a while but don't have a foolproof method that doesn't involve a lot of ad hoc curve twisting.  I shoot RAW and use CS6 ACR.  My first question

OK, I don't understand this.  Or does "linear gamma capture" just mean the RAW file that ACR opens.

Hi Wayne,

Linear gamma data is required to do basic image calculations. That means that the image data one adds, or multiplies by a factor, does just that on linear data, not already gamma pre-compensated data. Canon's Digital Photo Professional software for example allows to output linear gamma demosaiced output, just like scanners do on a hardware level. Scanner software then uses that linear gamma data to whitebalance, invert, gamma adjust, tonemap, and color correct for output.

Quote
Anyway, I open the RAW file in ACR and head straight for (what is now known as) the "Point Curve Tool" in ACR.

Unfortunately, the data has probably already been gamma adjusted at this point. The linear tonecurve is applied on top of the gamma compensation, and process version 2012 is highly non-linear in its tonecurve.

Quote
Then I adjust the endpoints of the R, G, and G curves and then invert them.  (Well, actually I invert them first.)  Which is what I think you described with your points 2. through 4.  Then I export to CS6 as 16 bit ProPhoto.  I then go into CS6's Curve tool and whale away twisting the R, G, and B curves (and usually try one of the "Auto" options in Curves to see if it auto-magically finishes the conversion.  Sometimes magic happens and I'm done.)  If I know that I have other frames that are similar, I save the Curves settings so I can use it on other images.

This last step is I think what you describe as point 5. on your list.  Correct?

I'm afraid the processing with ACR is far from ideal, there is too much happening under the hood that is not helpful, where one would prefer only basic linear gamma operations.

The simplest solution is to use an application like VueScan Professional to process a linear gamma Raw conversion (e.g. Canon DPP or similar for other brands, or RawTherapee), as if it were a 'digital negative' such as would be produced by a regular scanner.

Quote
This I would love to do, but don't know how to do it.  I'm using high CRI daylight balanced CFLs.  I tried using Rosco #3202 tungsten-to-daylight gel in between the CFLs and the negative in an attempt to compensate for the orange mask optically, but I ended up with more residual noise in the converted image than if I just used the CFLs straight and compensated for the orange mask in software (namely, the first step of adjusting the R, G, and G curves in the ACR point curve tool.)

To neutralize the film base+mask color as much as possible, one can adjust the total color spectrum of the lightsource with filters, or by varying and combining the exposures for the Red, Green, and Blue channels (the relative balance varies by film/processing cycle), e.g. with a filter wheel. The Nikon scanners e.g. can switch the R/G/B LEDs in sequence, and vary the exposure time for each scanline as instructed by the scanner software (VueScan will allow to set this exposure time per channel). For DSLR captures on should use an application like RawDigger to check for the very best quality exposure per channel.

Quote
I'm a little queasy about using CFLs, but I've found that I can eliminate a lot of dust and scratches by starting out with 400 watts (equivalent) of CFLs in a softbox and adding several additional stops of diffusion.  I can't get that amount of light with anything tungsten.   I tried a 4900k Solux bulb but that didn't really help the color conversion and it didn't produce enough light for diffusion--I spend hours cloning out scratches if all I use is a single Solux bulb.  A strobe isn't an attractive option, because I do manual focus with 10x liveview.

Ah, the reason I started this thread was because of me trying to come up with a better light source for camera scanning.

A more diffuse lightsource will reduce the visibility of dust and scratches, but can also add internal film reflections and lens glare. More collimated lightsources will show the dust and scratches more clearly.

Cheers,
Bart
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1675


« Reply #8 on: March 04, 2014, 03:24:04 AM »
ReplyReply

A more diffuse lightsource will reduce the visibility of dust and scratches, but can also add internal film reflections and lens glare. More collimated lightsources will show the dust and scratches more clearly.
Makes one wonder what one would do if the images are of "value for mankind".

Multiple shots using point-sources of light at various angles, doing some smart local blending where scratches are sought suppressed?

-h
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3550


« Reply #9 on: March 04, 2014, 03:54:18 AM »
ReplyReply

Makes one wonder what one would do if the images are of "value for mankind".

Multiple shots using point-sources of light at various angles, doing some smart local blending where scratches are sought suppressed?

Hi,

One would use a more diffuse lightsource to also reduce grain structure (Callier effect), but to suppress scratches a fluid mounting technique works best. Dust should of course be avoided as much as possible (antistatic pre-treatment).

Cheers,
Bart
Logged
WayneLarmon
Jr. Member
**
Offline Offline

Posts: 50


« Reply #10 on: March 04, 2014, 10:32:25 AM »
ReplyReply

Quote
Linear gamma data is required to do basic image calculations. That means that the image data one adds, or multiplies by a factor, does just that on linear data, not already gamma pre-compensated data. Canon's Digital Photo Professional software for example allows to output linear gamma demosaiced output, just like scanners do on a hardware level. Scanner software then uses that linear gamma data to whitebalance, invert, gamma adjust, tonemap, and color correct for output.

I'd like to do this (time consuming for a complete directory) step with DCRAW, because it is mechanical and I can easily script it.  Is this the same as

DCRAW -4 -T -o 4 %FILE%

-4       Linear 16-bit, same as "-6 -W -g 1 1"
-6       Write 16 bit instead of 8 bit
-W       Don't automatically stretch the image
-g 1 1   Set custom gamma curve.
-T       Write TIFF instead of PPM
-o 4     Set output colorspace to ProPhoto
%FILE% is the RAW file


And then I open it in PhotoShop and do your steps 2. through 4. (adjust R, G, and B curves and then invert.)  But does

Quote
Apply tone-curve

mean "do an additional adjustment in curves to convert from gamma 1 to gamma 2.2.  Or whatever looks best."

For the rest, I'm trying to come up with a reasonable workflow.  Does

Quote
To neutralize the film base+mask color as much as possible, one can adjust the total color spectrum of the lightsource with filters

bring me back to needing a spectroscope?   All I have for filters is are swatchbooks of sample Rosco filters.  If I figure out which filter I need, I order a full size sheet from B&H.   The samples in the swatchbooks a a bit small for real use.

And I'll add for the record that I spent considerable time trying to use VueScan (both with a PlusTek 7600i scanner and with my camera scanned RAW files) and that was no improvement over the results I get using the ACR workflow I described in my initial post.  This included several email exchanges with Ed Hamrick.  It was after I determined that VueScan was a non-starter that I switched to the ACR workflow I described.   (Possibly the reason why Vuescan's built in filters were ineffective is because I am digitizing slides and negatives that are old and faded.  Where do I get IT8 targets that match, say, Kodak Gold 200 that was stored unsleeved for 35 years in a damp basement?)   

Logged
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6880


WWW
« Reply #11 on: March 04, 2014, 10:42:55 AM »
ReplyReply

Where do I get IT8 targets that match, say, Kodak Gold 200 that was stored unsleeved for 35 years in a damp basement?)   


There is no such thing.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
WayneLarmon
Jr. Member
**
Offline Offline

Posts: 50


« Reply #12 on: March 04, 2014, 01:52:37 PM »
ReplyReply

Code:
DCRAW -4 -T -o 4 %FILE%
didn't work.  The TIFF files aren't convertible into anything usable.  What did work was

1. Opening the .CR2 file from the camera scanned negative in RawDigger.
2. Exporting as a TIFF with 'Raw composite" and "3-channel (RGB) output" checked
3. This gives a very dark TIFF file.  I opened it in Photoshop and
  a. straightened and cropped the frame off (very tricky with an almost black image)
  b. Inverted with Curves
  c. Used Levels to adjust the R, G, and B channels to fill the histogram. 
  c. While still in levels, adjusted the middle slider to be about 0.35 to adjust gamma.  Then OKed out of Levels

The result was pretty close to accurate and could be put into a simple PS Action.  But straightening and cropping the almost black image in step 1. is unworkable.  So I went back to RawDigger and exported to TIFF again and this time I clicked "Autoscale to use 16-bit range."  This gave me an image that is bright enough for straightening and cropping.  I repeated the rest of the steps, as above and got pretty much the exact same results.

This works and I think is giving better results than my old ACR workflow.   (In addition to my old negatives, I also have brand new negatives I shot last fall, with matching digital images.  The image I am testing with includes a Whibal card and a ColorChecker chart, taken on my front porch, so I know right off the bat when a conversion is in the ballpark.) 

But I would like to use DCRAW instead of RawDigger (so I can wrap the RAW-to-linear-TIFF conversion step in a Perl script.)   Does anybody know the options to give DCRAW to achieve the same results I got from RawDigger?  (RawDigger is built on top of DCRAW...)
Logged
WayneLarmon
Jr. Member
**
Offline Offline

Posts: 50


« Reply #13 on: March 04, 2014, 07:50:22 PM »
ReplyReply

I was looking at Raw Therapee for the first time, originally trying to figure out how to convert a Raw file to a linear TIFF file.  (I can't figure this out--I changed the output curve (in the Color Management tab) to linear but it all this does is change metadata that describes the output profile.  The TIFF file itself is unchanged (looks like a normal gamma 2.2 file.)

However...I was searching the Internet to try to find out how to save as a linear TIFF from RT and found some threads indicating that Raw Therapee developers are working on supporting converting negatives directly.  See

Convert Film Negative images to positive

and

Issue 2193:    Preprocession option needed for inverting image colors

These threads are as of February 2014, so they are recent.  I'm guessing that these modifications aren't in the release version of Raw Therapee yet.

How do I save a file as a linear TIFF from Raw Therapee?   And is the process I described in my previous post (exporting from RawDigger) giving me a linear TIFF file? 

Logged
WayneLarmon
Jr. Member
**
Offline Offline

Posts: 50


« Reply #14 on: March 05, 2014, 03:47:50 PM »
ReplyReply

Quote
How do I save a file as a linear TIFF from Raw Therapee?

I still haven't figured this out.  Anybody?

Quote
And is the process I described in my previous post (exporting from RawDigger) giving me a linear TIFF file?

I sort of don't think so.  I installed Canon DPP.  This has an obvious way to save as linear (a checkbox named "linear" that defaults to being unchecked.)  Files I export from DPP don't look like the files I exported from RawDigger.  And don't look like the "linear" files I get from DCRAW (above.)

But files from DPP seem to convert better than anything else I've tried.  I spent the afternoon revisiting camera scanned negatives and converting them again.  In case the original poster is still paying attention, (based on my understanding of Bart's advice) here is my workflow for converting camera scanned negatives to color corrected positives, using Canon DPP and PhotoShop:

1. Load RAW file in DPP.  Click on the "Linear" checkbox in the "RAW" tab.  Then click on the White Balance adjustment and then click on a neutral part of the negative.  In between sprocket holes or in between frames is a good place to click.
2. Switch to the "RGB" tab.  Adjust the endpoints to the black and white point of the histogram.  (I don't know if here is a good place to adjust the individual R, G, and G. black/white points.  I haven't been because it is difficult to see what you are doing when adjusting R, G, and B separately in the DPP "tone curve adjustment."  I adjust R, G, and B white/black points in Photoshop, later.)
3. Optional.  Go into "Tools/Preferences/Color Management" and change to "Wide Gamut RGB"  I've found in the past that colors on negatives sometimes don't fit into sRGB so I use a large color space until I am all done with the conversion and am ready to save.  YMMV.
4. Use "File/Convert and Save..."  Change the "Save as type" to "TIFF 16bit"  Assuming that you are using something like PhotoShop that can deal with 16 bit TIFFs.  The following steps will involve some radical curve twisting and anything 8-bit is likely to end up posterized.  Again, YMMV, but I use 16 bit TIFFs.

I use Photoshop.  Adapt instructions as needed for any other editing program.

5. Open the image file you just exported from DPP in PhotoShop.  Change the little triangle thingie on the status line so it shows "Document Profile".  It should be showing "Wide Gamut RGB (16bpc) if you followed my recommendations.   Straighten and crop away the edge of the frame so all you have is image.
6. Open Curves (Ctrl-M) and choose "Negative (RGB) from the Preset dropdown.  Click "OK."  (The image should change from negative to positive.  But it will be very bright and the colors will probably look wonky at this point.)
7. Open Levels (Ctrl-L). Click "Options" and choose "Find Dark & Light Colors"  Maybe Click "Snap Neutral Midtones".  Click "OK" in the Auto Color Correction Options".   You are still in the "Levels" dialog.  Check each of the R, G, and B channels to see that the endpoints match each histogram.  Then select RGB and adjust the gamma (brightness) by sliding the middle slider to somewhere around 0.6.  Adjust to taste.  Click on "OK" in the Levels dialog.
(You can do step 7. from the Curves dialog if you want.  But checking the fit of each R, G, and B histogram is easier in Levels than it is in Curves.)
At this point it should look close.  But let "Auto" in curves take another whack at it.  "Auto" works differently now that you've adjusted the gamma.  
8. Open Curves again and click on "Options..."  Try each of the top three algorithms, with "Snap Neutral Midtones" selected and deselected.  When you find the one that looks best, click on OK to exit the "Auto Color Correction Options dialog."  You will still be in the "Curves" dialog.
9.  At this point it should look real close.  But you can tweak the individual R, G, and B curves and see if you can improve it.  
10. After this use any other PS tool that will help.  Shadows/Highlights, etc.  Change the color space to sRGB (or whatever you need) as a final step (Edit/Convert to profile).  Change to 8 bits (Image/Mode/8 bits per channel.)  I usually keep the 16 bit TIFF but you probably should also save a copy as 8 bit JPG.

Most of this is mechanical and doesn't need any fine judgement until you get to step 9.  By the time you get to step 9., it should be very close to correct.

Corrections and suggestions are welcome!   Please.
« Last Edit: March 05, 2014, 04:00:47 PM by WayneLarmon » Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3550


« Reply #15 on: March 07, 2014, 07:39:02 AM »
ReplyReply

Code:
DCRAW -4 -T -o 4 %FILE%
didn't work.

Hi Wayne,

That surprises me, although I've not seen the file itself. The '-4' parameter should convert the Raw file into a Linear Gamma 16-bit/channel file, and the '-T' parameter will produce a TIFF file format output. That's exactly what you need.

Quote
The TIFF files aren't convertible into anything usable.

Depends on the steps taken... Maybe you could make an image available?

Using a 'Levels' type of correction you apply on that file, you can neutralize the White point (blank leader or inter-image space) for all three channels. That takes care of the mask. You then neutralize the Black levels for an overall Neutral rendering (although still inverted/negative). You then switch the output levels to invert the image. The simple middle control is not accurate enough for gamma adjustments, so you'll have to use a Curves adjustment to produce a more reasonable gamma compensation. What remains is color adjusting (Hue/Saturation/Lightness).

Quote
What did work was

1. Opening the .CR2 file from the camera scanned negative in RawDigger.
2. Exporting as a TIFF with 'Raw composite" and "3-channel (RGB) output" checked
3. This gives a very dark TIFF file.  I opened it in Photoshop and
  a. straightened and cropped the frame off (very tricky with an almost black image)
  b. Inverted with Curves
  c. Used Levels to adjust the R, G, and B channels to fill the histogram.

This should be close to what you got from DCraw, but with some of the White-balancing and Gamma correction done.

Quote
c. While still in levels, adjusted the middle slider to be about 0.35 to adjust gamma.  Then OKed out of Levels

It is better to use thee Curves dialog for this gamma adjustment. The middle input slider of the Levels control is not very accurate for the purpose of  gamma precompensation.

Quote
The result was pretty close to accurate and could be put into a simple PS Action.  But straightening and cropping the almost black image in step 1. is unworkable.

Not sure what cropping you are doing, but with 'Black Level' subtraction switched on, you only get what you framed in the camera. The image is dark because it has a gamma=1.0 and it is not Whitepoint (film mask) corrected yet.

Quote
This works and I think is giving better results than my old ACR workflow.   (In addition to my old negatives, I also have brand new negatives I shot last fall, with matching digital images.  The image I am testing with includes a Whibal card and a ColorChecker chart, taken on my front porch, so I know right off the bat when a conversion is in the ballpark.) 

But I would like to use DCRAW instead of RawDigger (so I can wrap the RAW-to-linear-TIFF conversion step in a Perl script.)   Does anybody know the options to give DCRAW to achieve the same results I got from RawDigger?  (RawDigger is built on top of DCRAW...)

You may get closer to the RawDigger output by experimenting with the DCraw whitebalancing options, but without gamma pre-compensation..

Just another thought, since you seem to use CR2 files, have you tried exporting a linear gamma TIFF from Canon's DPP application?

Cheers,
Bart
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3550


« Reply #16 on: March 07, 2014, 07:51:35 AM »
ReplyReply

How do I save a file as a linear TIFF from Raw Therapee?

I'd assume a linear gamma profile is required.

Quote
And is the process I described in my previous post (exporting from RawDigger) giving me a linear TIFF file?


As far as I understood what you are doing, it is perhaps already gamma adjusted. However, in the RawDigger preferences you can change settings, so it depends on how you've actually got that set up. Check out the (Display only) options Gamma 2.2, and the (Data processing) 'Subtract Black' and 'Linear Raw Curve' options.

Cheers,
Bart

P.S. I'm still puzzled why you can't get decent results from VueScan. I used to even get decent results scanning negatives as a positive with my Minolta software and using VueScan to process those 'Digital Negatives' with the 'Advanced Workflow' from VueScan (targeted at neutralizing the mask(ed) colors).
Logged
WayneLarmon
Jr. Member
**
Offline Offline

Posts: 50


« Reply #17 on: March 07, 2014, 02:00:30 PM »
ReplyReply

Quote
The TIFF files aren't convertible into anything usable.

Depends on the steps taken... Maybe you could make an image available?

I just did tests and tried all the methods I know of to convert a RAW file to a linear TIFF.  

After loading the linear TIFF into Photoshop CS6, I used the follow procedures in PhotoShop for each conversion.

  • Invert with Curves/Negative (so I can see positive colors from here on out.)
  • Normalize the R, G, and B curves with Levels by going through each channel one by one to set the white and black points for each R, G, and B channel
  • Convert the gamma with Curves by pulling down the composite RGB curve. I noted that each image had a green cast, so I removed the green cast by pulling down the green channel.

Here are a few images:


Canon DPP, white balanced and saved as linear


Converted.


DCRAW -4 -T -o 4 %FILE%


Converted

Exported from RawDigger with '3 channel (RGB) checked' and 'Autoscale to use full 16 bit range' not checked


Converted.


Adobe ACR with the R, G, and B histograms filled with the Point Curve Tool


Converted

And here is the same scene shot with a Canon EOS-M the same time I shot the film (which is Fuji 200.)


EOS-M

More examples and a link to the RAW file are here.

Quote
P.S. I'm still puzzled why you can't get decent results from VueScan. I used to even get decent results scanning negatives as a positive with my Minolta software and using VueScan to process those 'Digital Negatives' with the 'Advanced Workflow' from VueScan (targeted at neutralizing the mask(ed) colors).

Possibly because old school scanners expose the R, G, and B channels separately so VueScan has an easier job.  When I was corresponding with Ed Hamrick he sent me a RAW file of an image that was scanned with a Nikon scanner.  Each R, G, and B channel filled the histogram exactly.  I loaded the Nikon RAW file into Photoshop, inverted it, and it was almost perfect.  RAW files from any of my scanners (Epson V600 and Plustek 7600i) don't look like this.  VueScan doesn't do nearly as well with these RAW images (nor with RAW images from camera scanning.)

Ed's position is, apparently, that VueScan works with the old school transparency scanners so there isn't any problem and he won't do any additional work on making it convert transparencies (correctly) with the scanners that are currently on the market.  Because most of his customers are interested in making their all-in-ones scan prints, the few customers that are interested in scanning transparencies aren't worth spending much time on.   This was my impression after exchanging a few emails with him.  (I was requesting better controls for adjusting color and compensating for film fading.)  I found that I had to do just as much struggling in Photoshop after using Vuescan than I do now working directly from camera scanned negatives.

They don't make scanners that expose R, G, and B channels separately anymore.  And Nikon and Sony (Minolta) aren't likely to start up the production lines again.  But we have a bizillion DSLR and mirrorless bodies that can digitize transparencies (in less than a second per frame.)  If we can figure out the best way to convert the digitized negatives.

« Last Edit: March 07, 2014, 02:39:04 PM by WayneLarmon » Logged
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1152



WWW
« Reply #18 on: March 07, 2014, 03:32:45 PM »
ReplyReply

Quote
P.S. I'm still puzzled why you can't get decent results from VueScan. I used to even get decent results scanning negatives as a positive with my Minolta software and using VueScan to process those 'Digital Negatives' with the 'Advanced Workflow' from VueScan (targeted at neutralizing the mask(ed) colors).

I was asking the same thing. Something has really changed from the discussions I participated quite a few years back of folks using the scan as positive technique which worked pretty well using my CCFL light source Epson 4870 & Siverfast SE's HDR setting to get a linear tiff for further processing in PS. However I didn't get consistent results between the two film brands of 35mm Agfa Portrait 160 & Kodak HD 400 color negatives.

I left that all behind years ago due to the PITA aspect trying to edit out the color casts and saturation issues between the two film stocks. Now I see YouTube videos of folks using their DSLR's to "scan as positive", getting very good results quite easily with a simple invert and adjustments in levels.

This might have more to do with improved DSLR color accuracy over older scanner color description models, wider gamut capture in combination with ACR's color engine, color managed coding and improved ability to use some of the color information embedded within Canon DSLR Raw files.

In Wayne's other conversion processes there's no mention of source color space and yet the color is reasonably close. It almost seems as if software has standardized on a color description model in interpreting modern DSLR sensors over scanners.
Logged
WayneLarmon
Jr. Member
**
Offline Offline

Posts: 50


« Reply #19 on: March 07, 2014, 04:22:48 PM »
ReplyReply

Quote
I left that all behind years ago due to the PITA aspect trying to edit out the color casts and saturation issues between the two film stocks.

My current procedure is a PITA a lot of the time.  Up until reading this thread and reading what Bart wrote about linear files, I had used the "ACR, adjust R, G, B in the Point Curve Tool" (before finishing in Photoshop) technique exclusively.   It often doesn't work as well as it did in this particular test.

It just so happened that the ACR conversion ended up on the nose.   This time.  I only started with the other methods after reading Bart's posts, earlier in this thread.   I used this image as an example to show how the other procedures fared using simple mechanical processes (no twisting R, G, and B curves by eye, (other than removing the green cast that this particular image had, at the end of the conversion.))  

I did a test yesterday of revisiting older camera scanned negatives and the "export DPP as a white balanced linear file with the histogram stretched" often worked better than my old ACR technique.  As Bart pointed out earlier, ACR isn't linear so I can't do any adjusting on a linear file.   I think that DPP is exporting a linear file when the "linear" box is checked.

Quote
Now I see YouTube videos of folks using their DSLR's to "scan as positive", getting very good results quite easily with a simple invert and adjustments in levels.

Links?  I haven't seen these videos.

Quote
In Wayne's other conversion processes there's no mention of source color space and yet the color is reasonably close. It almost seems as if software has standardized on a color description model in interpreting modern DSLR sensors over scanners.

I don't think that camera sensors are thought of as having color spaces.  When exporting from a RAW converter, I reach for the largest output color space I can because I've found that this reduces troublesome cyan (usually) casts.  Then I whomp it down to sRGB as the final step before saving.

I'll repeat that this thread is my first exposure to the concept of saving a digitized negative as a linear file and converting from there.  Which is why I've had lots of questions.
« Last Edit: March 07, 2014, 04:25:23 PM by WayneLarmon » Logged
Pages: [1] 2 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad