Request for assistance with problem restoration.

[LateJunction]

Hello everybody, I have just joined this forum to seek some assistance and advice. Please tell me if this is inappropriate.

For a long time now I have been digitising a large collection of 35 mm slides which are between 40 and 55 years old. Over the past 6 months, I have been struggling with trying to restore the colour and exposure in some defective slides, shot using Fujichrome film in the late 1960s. In general most of my slides have not faded or decayed but about 5 of these FujiChrome films have either being defective films or were not processed correctly. They have a marked purple cast to them, are underexposed by 2 stops or more and there appears to be nothing in the blue channel above the lower mid-tones.

I have not found a way to correct these slides to my satisfaction and would be grateful for some assistance on how to correct a couple of samples. Hopefully, I could then apply lessons learned to the rest of these slides. Is it possible that I can post a couple of sample images here and get some advice?

My main workflow is digitisation at 64 bit RGBI via Nikon LS-50 scanner, driven by Vuescan to produce linear raw scans. These are then infra-red cleaned in Vuescan to produce 48 bit TIFs. Restoration/management is performed in LightRoom/Photoshop using plugins from Nik, Picture Code, CF Systems and others. I am also using RawTherapee to see if I can get more control.

One of the areas where I am not sufficiently well informed is the setting of most of the options in the Color Tab in Vuescan: Color Balance, Neutral R, G and B, Black and White points, Curve Hgh and Low and Brightness settings. I have Sacha Steinhoff's 'Vuescan Bible' but I'm still struggling. Any suggestions?

[Tony W]

Re: Request for assistance with problem restoratio

If you can post an example (or just a section if size is a problem) of the defective/incorrectly processed film I am sure we will try and help.

Be aware if the film is truly missing all data in one of the channels a full restoration may not prove possible

[LateJunction]

Re: Request for assistance with problem restoratio

If you can post an example (or just a section if size is a problem) of the defective/incorrectly processed film I am sure we will try and help.

Be aware if the film is truly missing all data in one of the channels a full restoration may not prove possible

Thanks for this offer. Size is going to be a serious problem here: a single 64 bit RCBI scan is 157 MB or so. The 48 bit RGB file, after 2x TIF size reduction in Vuescan, is about 32 MB. Should I try scanning at only 2K samples per inch and then only a part of the image? This will reduce the size to about 4MB I guess. Or do you require a part of the image at full scan resolution and without compression in Vuescan? Do you require the 64 bit linear raw file before infra-red cleaning or the 48 bit after cleaning?

Yes, I do appreciate that no amount of sophisticated restoration is going to make up for information that is just not there - not in an acceptable way, at least.

[Tony W]

Re: Request for assistance with problem restoratio

At this time just a small section of your scanned image will be fine just to get an idea how bad the 'damage' is. You are limited by the forum limits anyway to 100 Kb for non patron.

Just add an attachment with a JPEG of a section of your image; I would suggest something containing a memory colour e.g. some neutral area, grass and sky, flesh tones etc

[0lBaldy]

Re: Request for assistance with problem restoratio

LateJunction, welcome to RetouchPRO.
Just one thing to help the forum... If you use Photoshop you can post fairly large pictures which are under 100k using the 'save for web' function in Photoshop as described here:
Size, Quality and/or Format your Attached Images.. (Click here)
The save for web function in the latest Photoshop is hidden under File>Export>Save for web (legacy)

How to attach Files/Images to your Posts or Threads:
(Click here)

If you want to post a larger resolution that is over 100K then you can host your image elsewhere and attach a link in the thread (as you did here), but also, always please attach an under 100K version (so the thread remains useful in the future, regardless of external links). (link to the highest resolution.. or the largest size you can upload)

[LateJunction]

Re: Request for assistance with problem restoratio

I am surprised you can find anything useful in a 100KB sample of one of my slides. I notice that 'save for web' results in a size reduction of about 46 times from the original TIF to a JPEG at a quality setting of 60. Even so, with an original TIF at 120 MB, 100 KB in this format amounts to just 0.4% of the 'bits' in the slide. A sample is attached.

The car is a 1968 Ford Cougar in fern green (a deep green metallic). The photo was taken about mid-day on an over-cast day in April in Edmonton, Canada - so quite a low sun, even at mid-day. I am unable to identify anything that is a neutral grey in the whole photo.

I have tried reducing the quality to 1% in 'save for web' but the file size of the whole slide is still 400+KB. I have therefore reduced the file size in Photoshop by a factor of 55, and used a 'save for web' quality setting of 57 to create a view of the total slide which is less than 100KB. That also is attached.

[LateJunction]

Re: Request for assistance with problem restoratio

I am also attaching another scan attempt with Vuescan which has created colours which, in my mind, going back 50 years, are more faithful to the original scene. Again this is a crop which is less than 100KB, at a quality of 60, but drastically size reduced in Photoshop. Sadly I was stupid enough not to take a note of the Vuescan settings.

[LateJunction]

Re: Request for assistance with problem restoratio

LateJunction, welcome to RetouchPRO.
Just one thing to help the forum... If you use Photoshop you can post fairly large pictures which are under 100k using the 'save for web' function in Photoshop as described here:
Size, Quality and/or Format your Attached Images.. (Click here)
The save for web function in the latest Photoshop is hidden under File>Export>Save for web (legacy)

How to attach Files/Images to your Posts or Threads:
(Click here)

If you want to post a larger resolution that is over 100K then you can host your image elsewhere and attach a link in the thread (as you did here), but also, always please attach an under 100K version (so the thread remains useful in the future, regardless of external links). (link to the highest resolution.. or the largest size you can upload)

Thanks for the welcome and also for the references - they are most helpful.

[Tony W]

Re: Request for assistance with problem restoratio

I am surprised you can find anything useful in a 100KB sample of one of my slides. I notice that 'save for web' results in a size reduction of about 46 times from the original TIF to a JPEG at a quality setting of 60. Even so, with an original TIF at 120 MB, 100 KB in this format amounts to just 0.4% of the 'bits' in the slide. A sample is attached......

The reason why a small sample suggested was your OP suggested underexposed with colour cast. As we are looking for channel information (missing or otherwise) even a small JPEG should be sufficient to appreciate the loss of levels and offer suggestions that may help 'fix' the problem.

In looking at these images it seems that all channels contain information sufficient enough to enable a reasonable restoration.

In looking to restore images we are seeking information rather than accuracy i.e. we start with a poor image and aim to improve rather than copy in the scan process and later with post work. I did actually mention this on another forum but the poster seem to want to discuss IT8 targets and scanner profiles which IMHO irrelevant to the job in hand, anyway...

A very quick play with your posted JPEG suggests all should be recoverable to a satisfactory level. This is just one way that you may want to approach the subject if you do not want to scan again and attempt to normalise the image in the scanning.

1. Use a Curves layer to adjust the R,G and B channels for colour and the combined RGB for final contrast tweaks.

2. Add a Hue and Sat layer and increase the overall saturation to represent what you think is best.

3. Optional: As I wanted to exaggerate the green of the car the shadow areas suffered with a slight green cast (my green closer to British racing green ). This fixed by adding a layer set to Color blend and painting over the offending green with grey at lowish opacity until satisfied. NOTE: This may not be required with your TIFF as it could be a JPEG artefact (JPEG compression throws away colour info over luminosity)

While this may not be exactly to your liking I hope it may be somewhere in the ballpark, at least enough to progress the poorer images

You should be able to get much closer in the original scan if you adjust the channels histogram to balance the colour leaving final tweaks for Photoshop. I would suggest experimentation with pulling the white and black points in for RGB channels, making sure you do not clip information (and leave a little headroom for post work)

[LateJunction]

Re: Request for assistance with problem restoratio

Thank you for looking at this for me. Your results, quickly achieved, are most interesting. The sky, background buildings, concrete surface all look much improved in tonality and colour. From the histograms you have kindly provided I think I understand what you have done - do you see why I thought there was no information in the blue channel above the lower mid-tones? Was I correct in this description?

Anyway, you have given me enough to try additional tweaking of that poor scan output, as well as the 'better' version I uploaded.

In general though, I am of the opinion that it is better to attempt get the best possible results out of the scanning process, rather than relying entirely on pushing the pixels around in Camera Raw/Lightroom/Photoshop. Am I correct in this view ? In that regard could you expand a bit on what you mean by 'normalise the image in scanning' ?

[Tony W]

Re: Request for assistance with problem restoratio

I agree that it is better practice to get the best at the acquisition stage either camera or scan and minimise the work in post.

For me to normalise a scan is to correct at the scan stage for any colour, exposure deficiencies, at least as far as the scan software will allow me to adjust the individual R G and B channels to get the best balance and possibly the composite RGB for contrast. Ideally if possible using a Curves adjustment.

It is a long time since I scanned anything and while I have used Vuescan I seem to revert naturally to Epson scan

Although not a scan the attachment may illustrate my working on an image that is underexposed. Just imagine the left hand column shows a scan of a reversal film and that film isunderexposed by around 2 stops. The RGB histogram show the state of the scan without any intervention.

Wanting to get the best scan out of this image I move the black and white points on the histogram inwards for each channel. I would also be careful to leave a little headroom for further processing and make sure I did not clip any important information. There are/may be limits to the scan software ability to control to a fine degree the image and I would expect to do the final tweaks in PS

[0lBaldy]

Re: Request for assistance with problem restoratio

Just curious.

When scanning Photos, Slides, Negatives, or anything, the purpose is to get the best possible scan of the Original that can be work on non-destructively.

Towards that goal, I have always been told to turn off all automatic scanner enhancement settings and anything that auto fixes. No sharpness, no levels, no tone or saturation adjustment, no brightness or contrast adjustment, no dust & scratch removal, no descreen or denoising or despeckle, no color adjustment, no auto anything. If you make adjustments during the scan, the scanner makes blanketed, overall, averaged out modifications instead of selective modifications and may overwrite/change some useful information that cannot be retrieved . Save the raw scan preferably as an uncompressed tiff or psd at full size.

Once you have the best possible scan of the original all modifications can and should be made non-destructively in Photoshop or appropriate image processing software. If there are a number of slides/photos that may require basically the same modifications, an action could be made to speed the process and save steps.

Am I wrong with these assumptions?

[LateJunction]

Re: Request for assistance with problem restoratio

Just curious.

When scanning Photos, Slides, Negatives, or anything, the purpose is to get the best possible scan of the Original that can be work on non-destructively.

Towards that goal, I have always been told to turn off all automatic scanner enhancement settings and anything that auto fixes. No sharpness, no levels, no tone or saturation adjustment, no brightness or contrast adjustment, no dust & scratch removal, no descreen or denoising or despeckle, no color adjustment, no auto anything. If you make adjustments during the scan, the scanner makes blanketed, overall, averaged out modifications instead of selective modifications and may overwrite/change some useful information that cannot be retrieved . Save the raw scan preferably as an uncompressed tiff or psd at full size.

Once you have the best possible scan of the original all modifications can and should be made non-destructively in Photoshop or appropriate image processing software. If there are a number of slides/photos that may require basically the same modifications, an action could be made to speed the process and save steps.

Am I wrong with these assumptions?

In my opinion: you should not trust my opinion. Having said that I am in agreement with your view. It presents just one difficulty for me in my work flow: dust removal.

If I don't do dust removal in Vuescan (workflow is: scan at 64 bit RGBI; output to raw tif file; use that raw tif file as input to a scan with infra-red cleaning; output to 48 it tif) then 1). I don't know how to use the infra-red channel in Photoshop and 2). Those who do say the results are not as good as Vuescan (but thy don't reveal how they did it).

In addition I have never had a series of slides on which I need to do the same adjustments. Mine are always poor in a unique way!

[LateJunction]

Re: Request for assistance with problem restoratio

Here is another example from this same film where I just cannot seem to get the colour and tonality right. In this example I have not followed the opinion of setting all the scanner parameters to be at 'zero effect' - I have tried to get the most faithful colour output. Even so my efforts in Photoshop using Levels and Curves are still not acceptable. For example if I get the tonality right, the brightness is way off - too dark by a stop at least. If I correct that then the colours shift. if you know the area in the photo you will know that the colour of the road the horses are on was noticeably red - similar to the colour of Pall Mall today, 50 years on

[Tony W]

Re: Request for assistance with problem restoratio

Hi OB,
Interesting comments of which I broadly agree, with just a few reservations, but nothing definitive

....

When scanning Photos, Slides, Negatives, or anything, the purpose is to get the best possible scan of the Original that can be work on non-destructively.

Agreed

Towards that goal, I have always been told to turn off all automatic scanner enhancement settings and anything that auto fixes. No sharpness, no levels, no tone or saturation adjustment, no brightness or contrast adjustment, no dust & scratch removal, no descreen or denoising or despeckle, no color adjustment, no auto anything.

My preference the same all auto fixes off - however with knowledge of the scanner and the media manual intervention is not necessarily a bad thing.

If you make adjustments during the scan, the scanner makes blanketed, overall, averaged out modifications instead of selective modifications and may overwrite/change some useful information that cannot be retrieved. Save the raw scan preferably as an uncompressed tiff or psd at full size.

If you do wish to make adjustments based on the pre scan image you should not encounter issues as long as you set your parameters to include all the data contained in the original. For instance, if you use Curves or Levels to equalize the black and white points for each channel leaving a little room before clipping the scanner should not modify or overwrite any information.
Saving the scan as TIFF is probably the safe option and also there are no problems with the lossless TIFF compressions

Once you have the best possible scan of the original all modifications can and should be made non-destructively in Photoshop or appropriate image processing software. If there are a number of slides/photos that may require basically the same modifications, an action could be made to speed the process and save steps.

It is highly likely that you will end up in some editing software particularly if you have turned off all auto stuff even if it is only for sharpening and noise reduction.

Am I wrong with these assumptions?

I do not think so you just have a slightly different take on certain aspects. As long as the workflow 'catches mice' we may be just splitting hairs suggesting overall superiority of one method over the other. Well at least without some real world examples to prove a particular point

I have to say that I had forgotton that Vuescan offers a 'raw' format option.

While I understand the theory that 'scanner raw' files offer unaltered image data and may offer optimal information. I do wonder in the case of scanners (at least the low end ones such as Epsons I have used) how much value a so called raw offers over a high resolution TIFF image. One particular advantage may be in batch scanning where using traditional methods of optimising an image, contrast density etc through the scan dialogue tools could fall down as inappropriate settings may be applied to other images in the batch.
A raw scan by its nature should solve this potential problem as no alterations made (other than internal stuff over which you have no control) and in addition to the full 16 bit data per colour channel I believe it will also contain the 16 bit IR channel (assuming scanner has the capability).
In the main what I believe you are doing by raw scanning is seperating acquisition from the processing. That being the case I believe that raises some questions:

1. A raw capture does not guarantee an optimal image by itself, either in camera or scanner, at least if the exposure is incorrect. So how is the exposure arrived at, Scanner automatics or user intervention?

2. Is the end result of raw scanning superior to undertaking a high quality scan to TIFF with the image optimised close for exposure and colour balance?

Accepting that YMMV.

If there is a single 'best approach' I would love to see some real world comparisons comparing images