Ad
Ad
Ad
Pages: « 1 [2]   Bottom of Page
Print
Author Topic: Copy of color film negatives ... Invert possible in ARC?  (Read 14030 times)
WayneLarmon
Jr. Member
**
Offline Offline

Posts: 51


« Reply #20 on: July 22, 2014, 09:44:38 AM »
ReplyReply


Quote

Which is

Quote
    1. Set Input | Media
    2. Insert film with a transparent area
    3. If Input | Lock exposure is visible, clear Input | Lock exposure
    4. Press the Preview button
    5. Adjust cropping if necessary
    6. If Input | Lock exposure is visible, set Input | Lock exposure
    7. Press the Preview button again
    8. If Input | Lock film base color is visible, set Input | Lock film base color

For step #2 above, use a film frame that has an area that would print as pure black for negative, or pure white for slides.

Point 2. is the problem.  When I tried to follow this procedure on the frame I have been demonstrating (above posts), I tried to use the select rectangle to select an area between sprocket holes.  This does not work.  As soon as I press "Preview", the selection grows until it extends into the image proper.  I never see any "Input|Lock exposure" option.

I suspect that the "a transparent area" portion of step 2. means "use a frame that is completely blank", which won't work with the vast majority of rolls of negatives that have no blank frames.  (Both Canon DPP and Adobe ACR let me do a white balance click in an area between sprocket holes and they do do a white balance adjustment.)

Setting that aside, I went back to the "Color" tab in VueScan, clicked on "Default", and then set the film parameters from "Generic" to match the roll of Kodak Gold 200 that I was using.  The main difference is that VueScan defaults to "Color balance: White balance" instead of "Color Balance: Auto Levels" that I had used in my previous demonstration (above post).  I'm not going to show an example of that, because it was "fuglier" than the previous VueScan example I displayed--it had an additional green-cyan color cast.

I also tried clicking the VueScan white balance tool in an unexposed (between sprocket holes) area of the file with even "fuglier" results.   

I'd really, really like to come up with a workflow that uses VueScan because VueScan is a lot cheaper than (perpetual license) full Photoshop.  But after trying to use VueScan for a number of years, I can never achieve results better than the example I displayed in my above post.

(TL; DR section.  VueScan works really well with Nikon scanners because Nikon scanners essentially remove the orange mask in hardware by exposing the R, G, and B channels separately, which has the effect of mostly removing the orange mask.  I had an email exchange with Ed Hamrick several years ago about color quality when converting negatives and he sent me a VueScan RAW file from a Nikon scanner.  Which worked fine with VueScan.  And was almost perfect after opening it in Photoshop and doing a Ctrl-I inversion and a bit of gamma correction.  But Nikon scanners don't exist anymore.  And the kind of scanners that people can buy now (Epson, Plustek, etc.) don't remove the orange mask in the scanner hardware like the Nikon scanners did.  This may apply to other (not made anymore) high end scanners--I only have the single Nikon (VueScan) RAW file that Ed sent me.  Ed seemed to think that the color conversion problem was solved because VueScan worked fine with Nikon scanners.)

Have you used VueScan yourself to convert camera scans of color negatives?  The RAW file of the camera scan I am working with is linked from my Camera Scanning your Negatives page.  (Under "Converting method 1. Using Canon DPP and Photoshop")   If I am overlooking something in VueScan operation, I'd really like to know what I am overlooking.  So I can write up a VueScan camera scanning workflow that people can use.

Alexiz, thanks for the tip about the ColorPerfect plugin.

Wayne

Logged
BartvanderWolf
Sr. Member
****
Online Online

Posts: 3649


« Reply #21 on: July 22, 2014, 10:18:43 AM »
ReplyReply

Which is

Point 2. is the problem.

The space in between images is the most transparent area of that film (includes the effects of aging and fogging) with those particular processing conditions of the Lab.

Quote
I never see any "Input|Lock exposure" option.

Doing another Preview will make these menu options become available if I remember correctly. Of course only if the scanner allows to control the individual channel exposure times.

Quote
I also tried clicking the VueScan white balance tool in an unexposed (between sprocket holes) area of the file with even "fuglier" results.

White balancing is done on something that is neutral grey in the preview, NOT the film base color!  

Quote
Have you used VueScan yourself to convert camera scans of color negatives?


Yes, I've scanned a lot of negative and slide film on various Nikon Coolscan models, a Minolta DIMAGE Scan Elite 5400, and Epson V700 (and several other fladbeds before). Always worked like a charm, but it works best (noise performance wise, not necessarily color) if the R/G/B channels can  be individually exposed longer to neutralize the background. Alternatively one can use a filter to modify the spectral distribution color of the exposing light-source.

Quote
The RAW file of the camera scan I am working with is linked from my Camera Scanning your Negatives page.  (Under "Converting method 1. Using Canon DPP and Photoshop")   If I am overlooking something in VueScan operation, I'd really like to know what I am overlooking.  So I can write up a VueScan camera scanning workflow that people can use.

I'll have a look at the file.

Cheers,
Bart
« Last Edit: July 22, 2014, 10:21:06 AM by BartvanderWolf » Logged
BartvanderWolf
Sr. Member
****
Online Online

Posts: 3649


« Reply #22 on: July 22, 2014, 11:19:14 AM »
ReplyReply

I'll have a look at the file.

Attached, a copy directly from reading the Raw file as a transparency/negative in VueScan. Selected the left edge to lock film base color. Made a selection which excluded the bottom sprocket openings. Did another preview, and tweaked the Color settings a bit, but I had to guess which Fuji film profile to use.

You can further tweak the film base colors there, the Neutral balances if there was no white to click on the White point, the Black point, the overall brightness, choose the correct film type, etc.

Is is good color? No, but a usable basis from which to tweak in Photoshop.

Cheers,
Bart
« Last Edit: July 22, 2014, 01:26:00 PM by BartvanderWolf » Logged
WayneLarmon
Jr. Member
**
Offline Offline

Posts: 51


« Reply #23 on: July 22, 2014, 01:13:36 PM »
ReplyReply

Quote
Attached, a copy directly from reading the Raw file as a transparency/negative in VueScan. Selected the left edge to lock film base color. Made a selection which excluded the bottom sprocket openings. Did another preview, and tweaked the Color settings a bit, but I had to guess which Fuji film profile to use.

Cool.  Thanks.  I didn't know about the importance of locking in the film base.  And that the area between the frames was better than area between the sprocket holes.

I opened the RAW file in my copy of VueScan, followed your instructions, and got results similar to the image you displayed.  Which is a lot better than I got before (not locking the film base.)

However...I couldn't duplicate this with the image with the ColorChecker chart.  (Converted image is my DPP/Photoshop conversion.)


Apparently the (unexposed) left edge isn't wide enough to make a selection for locking in the film base.  I tried all the variations of selecting the left edge/Preview I could think of (and tried various portions of the right edge) and every time I got way worse results than the VueScan example I posted earlier. 

Could you try this again with the RAW file for the image that has the ColorChecker File?  (Direct link.) 

I'm trying to determine if the issue is that camera scanned images need more unexposed area-between-frames in the camera scanned image in order to be processable with VueScan.    Or if I am still not understanding something.

Thanks.

Wayne
Logged
BartvanderWolf
Sr. Member
****
Online Online

Posts: 3649


« Reply #24 on: July 22, 2014, 01:34:40 PM »
ReplyReply

Could you try this again with the RAW file for the image that has the ColorChecker File?  (Direct link.) 

Hi Wayne,

The ZIP file is damaged, the archive cannot be opened.

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

Posts: 51


« Reply #25 on: July 22, 2014, 08:34:31 PM »
ReplyReply


Hmm, I just downloaded it from the above direct link, copied it to a directory and it unzipped for me

D:\test>ls -l
total 31255
-rw-rw-rw-   1 user     group    32004467 Jul 22 21:05 raw-file-negative-colorchecker-chart.zip

D:\test>md5sum *.zip
7407b08e90e129ce93dfd7c1bb18ad63 *raw-file-negative-colorchecker-chart.zip

D:\test>unzip raw*.zip
Archive:  raw-file-negative-colorchecker-chart.zip
   creating: raw-file-negative-colorchecker-chart/
 extracting: raw-file-negative-colorchecker-chart/60D_11930_converted.PNG
  inflating: raw-file-negative-colorchecker-chart/60D_11930_negative.cr2

D:\test>cd raw*

D:\test\raw-file-negative-colorchecker-chart>ls -l
total 31654
-rw-rw-rw-   1 user     group     4380107 Mar  5 22:24 60D_11930_converted.PNG
-r--r--r--   1 user     group    28032588 Oct 26  2013 60D_11930_negative.cr2


(I use GNU UnxUtils to make MS-DOS be more Linuxy.) I also opened it with 7Zip.  60D_11930_negative.cr2 opened in CS6 ACR.  (The version of the .CR2 file in the .zip file had been white balanced in camera, so it won't open up looking orange.  I undid that white balance for the screen shots because people expect negatives to look orange.  (And then redid the white balance in Canon DPP before exporting as a linear TIFF.)  But this is the original .CR2 file I have been processing for these demonstrations.)

Can you check the file size and md5sum?

Wayne
Logged
BartvanderWolf
Sr. Member
****
Online Online

Posts: 3649


« Reply #26 on: July 23, 2014, 02:01:36 AM »
ReplyReply

D:\test>md5sum *.zip
7407b08e90e129ce93dfd7c1bb18ad63 *raw-file-negative-colorchecker-chart.zip

Hi Wayne,

I downloaded another copy, just like I did for the earlier one that did open.

The MD5 comes out as : 5611f6a417769bdb2c87e38239996e16 , and is clearly not the same. The file size is 21845056 bytes.

<entry name="raw-file-negative-colorchecker-chart.zip"> <size>21845056</size> <mtime>07/23/2014 06:35:29</mtime> <attrib>A-----</attrib> <algorithm>MD5</algorithm> <hash display-as="Hexadecimal" encoded-as="Hexadecimal">5611f6a417769bdb2c87e38239996e16</hash> </entry>

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

Posts: 51


« Reply #27 on: July 23, 2014, 09:18:19 AM »
ReplyReply

Odd.  OK, let's try Dropbox.  Download
https://dl.dropboxusercontent.com/u/30509795/raw-file-negative-colorchecker-chart.zip

I downloaded from this link (using a different browser) into a different directory:

D:\test2>ls -l
total 31255
-rw-rw-rw-   1 user     group    32004467 Jul 23 10:09 raw-file-negative-colorchecker-chart.zip

D:\test2>md5sum *.zip
7407b08e90e129ce93dfd7c1bb18ad63 *raw-file-negative-colorchecker-chart.zip

D:\test2>unzip *.zip
Archive:  raw-file-negative-colorchecker-chart.zip
   creating: raw-file-negative-colorchecker-chart/
 extracting: raw-file-negative-colorchecker-chart/60D_11930_converted.PNG
  inflating: raw-file-negative-colorchecker-chart/60D_11930_negative.cr2

D:\test2>cd raw*

D:\test2\raw-file-negative-colorchecker-chart>ls -l
total 31654
-rw-rw-rw-   1 user     group     4380107 Mar  5 22:24 60D_11930_converted.PNG
-r--r--r--   1 user     group    28032588 Oct 26  2013 60D_11930_negative.cr2


Maybe DARPA needs to go back to the drawing board about this whole TCP thing.  Seems that it isn't quite ready for prime time.

Wayne
Logged
BartvanderWolf
Sr. Member
****
Online Online

Posts: 3649


« Reply #28 on: July 23, 2014, 10:33:03 AM »
ReplyReply


Hi Wayne,

Yes, that one works fine.

Attached an attempt to get something decent out of the file, direct import of the Raw into VueScan, as a color negative, film base color, tweaking a number of Color settings, manually adjusting the film base color further while watching the histogram.

Cheers,
Bart

Logged
WayneLarmon
Jr. Member
**
Offline Offline

Posts: 51


« Reply #29 on: July 23, 2014, 03:14:32 PM »
ReplyReply

This image is tricky.  I think that the film has a yellow-green color cast, courtesy of Kodak.  For example, here is the scan that Dwayne's Photo did when they processed the film



As compared to a shot taken directly with a Canon EOS-M at the same time



I am very familiar with the concrete because it is my front porch.  The concrete does not have an overt yellow-green cast.  The EOS-M version more closely resembles reality.  (But even that has a slight greenish tinge in some places that my eyes don't detect when I look at the concrete.  Maybe some metamerism going on?  I think that the concrete dates from the mid 1940s.  I live in upstate New York (US.))

But this image is representative of ones where I'd spend time fighting in VueScan, only to continue fighting in Photoshop.  After I while, I decided that doing everything from scratch in Photoshop was easier.

One issue we haven't discussed in this thread is that this image was from (reasonably) fresh film, so the film curves as published by Kodak (or by Fujifilm) that VueScan (presumably) use are relevant.  But most of the film that I am processing (family snapshots) are from the 1970s and 1980s and was stored poorly.  So the factory film curves that are baked into VueScan may not be as helpful.   Generally, families don't store film the same way that museums do.

Thanks for the insights on how to use VueScan better.

[Edit.  Changed references from "green cast" to yellow-green cast", after looking at the images again.]

Wayne
« Last Edit: July 23, 2014, 08:11:10 PM by WayneLarmon » Logged
Pages: « 1 [2]   Top of Page
Print
Jump to:  

Ad
Ad
Ad