Copying colour sides
|
February 28, 2008, 04:35 PM
|
|
I have a collection of colour slides, which were taken over 25 years ago, some of which I would like to copy and manipulate using Gimp. I have a piece of equipment which I can connect to my camera to hold a slide and which will allow the camera to be focussed. The slides have been stored in closed cases away from the light so I would expect colour degradation to be a minimum. The film used was Kodachrome 25 and processed by Kodak.
So that I may get any items that I shall require and also be aware of any pitfalls I would welcome comments from those who have had experience in slide copying.
|
| |
|
plus
February 29, 2008, 03:36 PM
|
|
|
The only thing I can think of is getting the colour temperature right. But I never have copied slides.
|
| |
|
February 29, 2008, 03:50 PM
|
|
Hi littletank. I've duplicated a lot of old family slides and can share my experience. I also had a slide duplicator that had a T-mount adapter for Nikon. It worked perfectly on my Nikon film camera. However, it did not work on my Nikon DSLR for a couple of reasons. The main problem is that the slide duplicator is designed to provide a 1 to 1 image size for 35mm full frame sensor. On a 1.5x crop factor DSLR, you only see 2/3rds of the image. Also, the simple T-mount adapter will not meter on Nikon DSLRs. Unfortunately, the Nikon system will not meter, even for flash TTL metering without a "chipped" lens unless you have a D200 or above (I have a D50 and a D80).
I found that I get the best results by using a macro lens that goes down to 1:1 magnification to copy the slides. I took the slide duplicator apart and use the just end piece that holds the slides and diffuses the light. I mounted that on a cardboard tube, into which I aim the macro lens focused at slightly less than 1:1 magnification (so I see the whole slide + a little margin to crop). I then use off-camera TTL flash with the flash placed about a foot in front of the slide duplicator. It is important to shield the flash from the macro lens so that you don't get any lens flare.
With normal Flash TTL exposure, the copies look better than the original since the metering of the DSLR is much better than the metering that produced the original slides.
If you have a slide duplicator that is designed for DSLRs with a crop factor and if your camera can meter with it, then you may be able to use the duplicator as it was originally intended.
Best of luck.
|
| |
|
|
February 29, 2008, 03:52 PM
|
|
|
P.S. Regarding color temperature, since you are using flash to duplicate the slides, you can select flash color balance manually. However, I have found very good luck with just using auto-white balance on my Nikon DSLRs. The white balance on slides tends to be wacky a lot of times, and AWB can actually improve on it. If you set the WB to flash, you will essentially duplicate whatever WB was on the original slides.
|
| |
|
|
February 29, 2008, 07:50 PM
|
|
|
Thank you for your comprehensive response. I do not have an off camera flash so I wqill have to consider various light sources. LED or halogen lamps seem a possibility and I will carry out some trials to find which works best with UFraw.
|
| |
|
|
March 03, 2008, 02:07 PM
|
|
|
I have made an attempt to copy some transparencies and have posted 2 images on meetthegimp. Please comment as it is the only way I am able to improve my technique,
|
| |
|
|
plus
March 03, 2008, 04:38 PM
|
|
Without a look at the original it is difficult to say. The colours seem not to be right and the contrast too high. But this can be happened during copying or the slide can be so much degraded in 30 years.
I would be interested in seeing a 100% resolution cut out of the images.
|
| |
|
|
March 03, 2008, 05:39 PM
|
|
|
I agree with you about the colours especially when compared with both a JPEG taken at the same time as the RAW and the original slide so I have returned to the RAW image and tried again. The difficulty, I think, is with the RAW conversion. To get a reasonable white balance the colour temperature setting is over 10000K which seems to result in a highly saturated image. When I make a significant cut in the saturation level the result starts to compare more closely with the JPEG image. Where would you suggest I put the images for you to see and also the 100% resolution you mentioned?
|
| |
|
|
March 04, 2008, 07:14 PM
|
|
|
I have tried again with a different transparency and, again, would welcome comments.
|
| |
|
|
March 16, 2008, 05:18 PM
|
|
|
I have been told that images made from my old slides show a slight blue fringing. This is very apparent when the images are examined at 100%. Careful examination shows that the colour of the fringing goes all the way through the spectrum from red to blue. When it occurs, the fringing follows the edges of straight lines and I have even seen ellipses around circular objects. In some ways this reminds me of halation. Naturally, I would like to stop this happening and I wonder if you have any bright ideas on things that I might try.
|
| |
|
|
March 17, 2008, 09:52 AM
|
|
|
I have been told that this phenomenon is due to chromatic aberation and presumably is caused by the optics in use.
|
| |
|
March 18, 2008, 07:17 PM
|
|
@littletank.
I had a look at your picture to see what can be done about it.
First of all, one has to distinguish between transverse (TCA) and longitudinal (LCA) chromatic aberration. Both are due to the fact that the refractive index depends on wavelength.
LCA is the phenomenon that different wavelengths are focused on different planes. This cannot be corrected anymore afterwards as far as I know.
TCA deals with light at an angle to the lens. In this case light will be focused on different positions in the focal plane (that is, if LCA is absent of course). Therefore, TCA becomes more prominent at the edges of photos where the angles is the largest (remember, in optics angle is measured relative to the normal of a plane!). This can be corrected.
However, GIMP will be hardly of any use here. One has to deal with lens imperfections for which GIMP has very limited possibilities. The best you can do, in my opinion, is fiddling with the RGB channels, i.e. slightly moving, rotating and scaling them until they match as close as possible. Just use the channels in normal mode with reduced opacity to facilitate adjusting.
I choose another way and did a quick test with Hugin which is built for this kind of job.
- First I decomposed the image into RGB channels. Have a look by the way at those channels and watch the image almost jump when looking at the different channels. This shows to what extent CA has affected your image.
- Each layer was saved as an image.
- In Hugin I corrected these images and saved the result (multilayer TIFF in this case).
- In GIMP I composed those layers again into an RGB image and cropped the result (as a result of the treatment by Hugin layers have inevitably acquired slightly different sizes)
I have uploaded the result here: http://www.23hq.com/Teld/photo/2961851/view-large
I hope you do not mind me having used your image. Have a look at the roof, windows, milk cans or ladder to see that TCA has been greatly reduced. It can be even better.
- First, I had to work on your image in JPEG which is not optimal to say the least when one has to correct TCA. Use RAW or another lossless format like PNG or TIFF.
- Second, I reduced the image and saved again (!) in JPEG to upload them. That does not help either to show the result.
- Last but not least, I am sure I could improve things in Hugin by playing around with parameters. My goal now was just the proof of concept.
I hope it helps, Serge
|
| |
|
|
March 19, 2008, 06:55 PM
|
|
|
Thanks for your comprehensive exposition which is very interesting. I have no objection at all to you using the image but, for some reason, I am not able to see the result because I am refused access to the page. Incidentally I have had my attention drawn to a piece of commercial software which will run in Wine and which is very easy to use. There is a free trial and I have tried it with success. If you want a reference to the software I will supply it.
|
| |
|
|
March 19, 2008, 08:08 PM
|
|
Oops, sorry. The image was still private. I have made it public now. I am sorry for the inconvenience.
Indeed, I am interested in the reference. I am always curious about what is available in graphics land.
|
| |
|
|
March 19, 2008, 08:40 PM
|
|
|
Have a look at www.epaperpress.com/ptlens
|
| |
|
|
March 19, 2008, 11:01 PM
|
|
Thanks for the link.
I noticed that the author uses PanoTools for his panorama's. Well, Hugin is a GUI for PanoTools. Incidentally Hugin has among its proposals for the Google Summer of Code 2008 an enhanced treatment of TCA. At the moment it requires quite a lot of manual work.
|
| |
|
|
March 20, 2008, 02:11 PM
|
|
|
Having put the topic of CA to bed for the time being I would now like to turn to another subject - barrel distortion. Because I need to use a fairly wide angle setting and to focus down to a few cm there is a very noticeable barrel distortion. PTLens is not able to deal with this and I wondered what might be a good way to approach this aberration.
|
| |
|
|
March 20, 2008, 02:37 PM
|
|
My answer is obviously Hugin. That is my Swiss knife for dealing with lens imperfections.
There is also a GIMP plugin, menu Filters -> Distorts -> Lens Distortion, which might do the job as well.
For your information, both can handle also the other radial distortion, namely pincushion distortion.
By the way, are you sure PTLens cannot handle barrel/pincushion distortion? The website says it can.
|
| |
|
|
March 20, 2008, 02:45 PM
|
|
Every time I look at Hugin it seems very formidable so I will have a look at the GIMP plugin. As for PTLens here is a quotation from the author of the software:-
Calibrations in PTLens are based on distant objects, say from 20 feet to infinity, and are not valid for close-up photography. Although I do
calibrate zoom lenses for different focal lengths, I don't have enough time to calibrate for different distances.
This was in reply to an email.
|
| |
|
|
plus
March 20, 2008, 04:09 PM
|
|
|
At least the GIMP filter will be covered in the next episode. Hugin depends on how well I can understand the thing.
|
| |
|
|
March 23, 2008, 02:49 PM
|
|
|
I have uploaded another copied slide and would welcome comments. Barrel distortion was remedied using GIMP and CA using PTLens.
|
| |
|
|
plus
March 23, 2008, 03:15 PM
|
|
|
You have got it right - and you just missed being "live" on the show for 20 minutes or so. (You'll understand on Tuesday ;-) )
|
| |
|
|
March 23, 2008, 05:41 PM
|
|
|
I can't wait but I suppose I have no choice.
|
| |
|
|
plus
March 23, 2008, 09:34 PM
|
|
|
I just showed this web page in the show and you had posted 20 minutes before. ;-)
|
| |
|
