Luminous Landscape Forum
Raw & Post Processing, Printing => Digital Image Processing => Topic started by: Jeremy Payne on September 16, 2009, 10:49:50 am
-
Let's say I have three aligned monochrome layers that I would like to transform into the Blue, Red and Green channels of a single image.
Is there an easy way to do that?
-
You can make Alpha channels from Layers using Calculations, but I don't know that exactly what you are asking to do is possible. Why do you want to do this?
-
Let's say I have three aligned monochrome layers that I would like to transform into the Blue, Red and Green channels of a single image.
Is there an easy way to do that?
You can copy a channel and paste it over another channel. This way you can even swap the color channels. (The monochrom layer has a single channel.)
-
Why do you want to do this?
I'm fascinated by the work of Prokudin-Gorskii and I'd like to make my own color image files from scans of his raw negatives ...
http://www.loc.gov/exhibits/empire/making.html (http://www.loc.gov/exhibits/empire/making.html)
I have been having a lot of fun playing with the fully rendered and color versions doing my own retouching and color correction and then printing them ... they look terrific.
I'd now like to try and create my own color images from the positive scans of his filtered original captures.
Here's an example of the scanned negatives:
(http://memory.loc.gov/pnp/prok/01800/01884r.jpg)
-
Fascinating but not clear to me what gymnastics would be needed to get each greyscale image representing one of the primaries into a channel reflecting that primary.
-
You should use the Image>Apply image command. Set the destination layer or channel by making it the working layer or channel. Select the source from the Apply image dialog. The source can be either an entire layer or just a channel from the current document or another loaded document. So you could create a new layer and copy your source layers to individual channels. -R
-
You should use the Image>Apply image command. Set the destination layer or channel by making it the working layer or channel. Select the source from the Apply image dialog. The source can be either an entire layer or just a channel from the current document or another loaded document. So you could create a new layer and copy your source layers to individual channels. -R
Sweet - thank you! I will give it a go.
-
Cool idea. I was living near DC when that exhibit opened, and I got all caught up in it myself. Not sure I ever really understood how he fired off all three shots quickly enough for that to work. At the time, the images with moving water made me think that he had some method whereby he passed a single holder through the camera back almost in one motion. I suppose the curators have a very good idea of the exact method. It's a shame other photographers didn't try the same thing elsewhere.
-
At the time, the images with moving water made me think that he had some method whereby he passed a single holder through the camera back almost in one motion. I suppose the curators have a very good idea of the exact method. It's a shame other photographers didn't try the same thing elsewhere.
The Wikipedia article (http://en.wikipedia.org/wiki/Sergei_Mikhailovich_Prokudin-Gorskii) says,
Any stray movement within the camera's field of view showed up in the prints as multiple "ghosted" images, since the red, green and blue images were taken of the subject at slightly different times.
The exposure time of the frames is likely to have varied, even if the developed negatives were later on similar glass plates. In a letter to Leo Tolstoy requesting a photo session, Prokudin-Gorsky described each photo as taking one to three seconds, but when recollecting his time with Tolstoy, he described a six-second exposure on a sunny day. Blaise Agüera y Arcas estimated the exposure of a 1909 photo taken in broad daylight to have had combined exposures of over a minute, using the movement of the moon as comparison.
(It includes references for that info, but I haven't looked at them.)
-
You should use the Image>Apply image command. Set the destination layer or channel by making it the working layer or channel. Select the source from the Apply image dialog. The source can be either an entire layer or just a channel from the current document or another loaded document. So you could create a new layer and copy your source layers to individual channels. -R
I don't think this procedure would transform the monochrome layers representing R, G and B information into R, G and B channels separating that information accordingly.
-
heres one way how to do it...
[attachment=16618:Prokudin...i_layers.jpg][attachment=16617:Prokudin_Gorskii.jpg
]
-
This is the result after loading the different images into RGB channels and experimenting with a curves adjustment layer.
In the scanned images which is the red, green, and blue channel?
(http://img.photobucket.com/albums/v732/glenngaryglenross/applyImageExp_web1.jpg)
-
This is the result after loading the different images into RGB channels and experimenting with a curves adjustment layer.
In the scanned images which is the red, green, and blue channel?
(http://img.photobucket.com/albums/v732/glenngaryglenross/applyImageExp_web1.jpg)
Very cool ... didn't have time last night to play around - was printing up a storm ...
They are Blue, Green & Red from top to bottom.
-
This is the result after loading the different images into RGB channels and experimenting with a curves adjustment layer.
In the scanned images which is the red, green, and blue channel?
(http://img.photobucket.com/albums/v732/glenngaryglenross/applyImageExp_web1.jpg)
Nice try, but not successful. There must be a way to do this, and I thought of some similar approach to what you are displaying here, but these layers are not really substitutes for channels.
-
Jeremy, I went back to the L.o.C website and re-read their process description. What's missing is exactly how they filter the individual grayscale images in a digital process to produce colours amenable to fine-tuning for producing a correct-looking result. Why not send a request to the L.o.C for more detailed technical information on how they did it? It's *probably* not a secret.
-
Hi jeremy
If you do contact L.o.C. they may be able to tell you the colour of the filters used and try to work out the x y values of the dominant primaries.
These then could be loaded in custom rgb working space and saved as a profile in photoshop.
You would in effect be working the the filter colour space which was used to make the separations and the resulting projections.
This might help in colour adjustments.
Ill play around and see what I get.
[attachment=16622:custom.jpg]
[attachment=16623:profile.jpg]
-
hi eaiest way to assemble image is to open the three image file ( blue green red plate from top to bottom)
use marquee tool to select top image (blue)
copy
open new file (rgb untagged)
select blue channel and paste.
save file
repeat for other two plates
then select move tool and align red and blue to green plate
asign a rgb profile of your choice
colour adjust to taste.
-
Here's what I get using apply image loading the aligned gray scale scans into the appropriate rgb channels.
(http://img.photobucket.com/albums/v732/glenngaryglenross/scannednegatives_web1.jpg)
Same image with white and black points set and mild s curve.
(http://img.photobucket.com/albums/v732/glenngaryglenross/scannednegatives_web2.jpg)
-
hi
just did this today from the 68mb file
(aside from the matlab exercise in image registration, im more interested in colour rebuilding methods)
i think i need to contact the LoC as to the method of digital capture. this is an intriguing experiment.
can we reproduce the capture effect from glass plate negatives?
can we determine the filtration used based on the unknown bgr separations?
can we determine the white source of the projections of the tri negatives?
if i could get any of this info i might be able to try run it thru ciecam02
i think basically we have a camera raw file from c1900
any ideas any one?
-
Here's my first pass ... this is fun!
[attachment=16764:1_3.jpg]
-
hi
just did this today from the 68mb file
(aside from the matlab exercise in image registration, im more interested in colour rebuilding methods)
i think i need to contact the LoC as to the method of digital capture. this is an intriguing experiment.
can we reproduce the capture effect from glass plate negatives?
can we determine the filtration used based on the unknown bgr separations?
can we determine the white source of the projections of the tri negatives?
if i could get any of this info i might be able to try run it thru ciecam02
i think basically we have a camera raw file from c1900
any ideas any one?
This is pretty good. Where did you get the 68MB file from? Can you tell us the specific steps you used to produce this?
-
Here's my first pass ... this is fun!
[attachment=16764:1_3.jpg]
Also pretty good Jeremy. Likewise, where did you get the original file and what steps did you implement to get this?
-
This is pretty good. Where did you get the 68MB file from? Can you tell us the specific steps you used to produce this?
Starting with the list from the exhibition ... http://www.loc.gov/exhibits/empire/object.html (http://www.loc.gov/exhibits/empire/object.html) ... I copied the name of the image - in this case: 'Group of Jewish Children with a Teacher'
I went here ... http://lcweb2.loc.gov/pp/mdbquery.html (http://lcweb2.loc.gov/pp/mdbquery.html) and searched for the text.
Following the results leads to this: http://memory.loc.gov/pnp/prok/01800/01861u.tif (http://memory.loc.gov/pnp/prok/01800/01861u.tif) ... the hi-rez B&W triptych.
I opened in ACR and did a little sharpening, opened in aRGB and zoned out the three areas of the positive B&W and made 3 individual aRGB tiffs.
I used AutoPanoPro to align them in a planar projection. I saved them as individual 16 bit tiffs.
I opened the files in PS ... created a new, blank aRGB image ... copied the b/g layer from each file into the appropriate channel on the channels pallet of the blank file and re-touched from there.
It was easier than I thought it would be.
I'm gonna work on other methods of alignment.
-
Also pretty good Jeremy.
Thanks! I need to work on the alignment ... APP gave it too much distortion. I will work on that aspect next ...
As a comparison to mine, here is the image as re-constructed by the LoC:
(http://memory.loc.gov/service/pnp/ppmsc/04400/04442v.jpg)
... and mine:
(http://farm4.static.flickr.com/3505/3946946563_f594a37241_o.jpg)
-
Thanks Jeremy. Nice work. Compares well with LoC, except for that bit of distortion you are working on.
-
Oh - and to copy the Background into the channels of the Blank doc - did you use Apply Image for that step?
-
Oh - and to copy the Background into the channels of the Blank doc - did you use Apply Image for that step?
didn't need to ... just regular copy and paste.
-
Thanks Jeremy.
-
Here's the 'final' take ... whatever that means.
I leveled it and corrected the distortion introduced in the alignment, cropped it square and did some more retouching.
Overall, I'm very pleased with how this came out ... I'll definitely be doing more and printing them.
Anyone have any suggestions for the paper? I have an Epson r2880.
(http://farm3.static.flickr.com/2482/3948042649_1f04d235d0_o.jpg)
-
Fine result Jeremy.
As for paper, it depends on what "feel" you think is right for these images. Let us recall that the originals were projected, so the whole print business wasn't how these images were "purposed" to use contemporary lingo. The essential choice is between matte and glossy (various flavours of each) and you undoubtedly know what the trade-off is. One is nice and artsy and the other has great DMax. I'd suggest you try a print on Ilford Gold Fibre Silk and another on a nice matte photo-rag paper and make a visual/tactile assessment of which medium is a better "fit" for the character and tonal qualities of the images.
-
hi
I got the 68meg files by searching the archive
eg http://lcweb2.loc.gov/cgi-bin/query/D?prok:19:./temp/~pp_JoqR::
and then click on the triptych wich will take you to another page with the links to the different files.
I cut and pasted each frame into a a new rgb file as jermy outlined except that i opened an untagged rgb file and eyeballed the registration.
i then assigned a profile of my my choice.
-
Interesting thread. For alignment, I think you could use Photoshop CS4 if you have it. Drag two of the images on top of the third as layers, and then use the alignment feature in CS4. This would avoid the warping done by a pano stitcher, and then you could copy paste them into the individual channels of a new document.
-
Interesting thread. For alignment, I think you could use Photoshop CS4 if you have it. Drag two of the images on top of the third as layers, and then use the alignment feature in CS4. This would avoid the warping done by a pano stitcher, and then you could copy paste them into the individual channels of a new document.
I did try that as well (CS4) - but for whatever reason, APP worked better - less "CA" and edge halos. I bet it depends on the image.
-
I did try that as well (CS4) - but for whatever reason, APP worked better - less "CA" and edge halos. I bet it depends on the image.
Jeremy,
Good thread. How did you get good alignment between the layers, I tried with APP but the kid to the right of the teacher moved between exposures, did you manage to get APP to fix this for you?