Tweaking my slide film exposures to improve scanning of shadow areas

[philipus]

Advertisement (gone after registration)

Since I began photography in the late 80s I've been a fan of Velvia and I've shot a lot of the ISO 50 variety over the years. I regularly shot it (and had it developed) at EI100 because I enjoyed the extra saturated colours.

 

These days I shoot Velvia 100 (old stock) and 100F and develop at box speed because I find the film saturated enough. I guess when I was of Instagram-equivalent age I saw things differently .

 

I find that my slide films scan well these days (Coolscan 9000 with Vuescan). However, I'm not always happy about the shadow areas so I'm curious about ways "free them up" a tiny bit to assist the scanning process and give me more image information to work with in post.

 

For instance, would exposing the film at, say, EI160 (thus giving it a bit more light) but developing at box speed make things better (or worse)?

 

Thanks in advance for any thoughts

Philip

[gbealnz]

Philip.

not really able to assist with your conundrum, but wanted to add, if you expose at EI160 instead of the suggested EI100, surely you are giving the film "less" light?

Or am I way wide of the mark?

Gary

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[philipus]

Ah yes, thanks Gary. A bit of quick typing there. Thanks for spotting that. I would have to shot at a lower than the box speed EI to give the film more light.

[xalo]

Philip,

 

Not sure if there is a development approach to the shadows in E-6 but would also be happy to learn. My slide film preferences usually go to less satured varieties, but I noted that Velvia (50 in my case) can quickly clog up the shadows.

 

With Provia 100F (my preference) I find the matter less problematic, but still part of the slide film characteristics. In contrasty scenes I seem to have the choice to either overexpose and wash out colours (difficult or impossible to restore) and/or burn highlights or to accept black shades (better). That can work visually, but does not always... selection of a more evenly lit scene is the alternative. For leisurely paced landscapes on medium format slides I have started using ND grade filters (when the scene allows) for a more even exposure level, keeping more leeway (no pun intended) in post. There can be unwanted effects though (unnatural skies etc.)

 

One thing to try (and you may do so already) after the fact is multiple exposure when scanning to get as much detall from dense parts as possible. I bought SilverFast for my Minolta 5400 almost alone for this (and rarely used it), but probably Vuescan or other advanced software has a similar feature. It can also be done manually (never tried I think), by doing separate scans for highiights and shades, then assembling them manually as layers or as HDR in PS.

 

The last option for important images might be using a drum scan or a Hasselblad Flextight. Having recently seen the results from the latter, I must say, it has some appeal. Black will remain black though...

 

Cheers,

Alexander

 

P.S.: I forgot fill-flash or reflector, but that seems not so M...

[philipus]

Thank you very much for your thoughts Alexander. I hadn't thought of the graded ND filters. It could be a good solution where the situation allows for it, such as landscapes. And your explanation about the choices one faces when setting exposure nails it very well, I think; highlights burn in a very "digital" way and leave very little information to be rescued in post. I have found that ColorPerfect's positive mode helps some but is of course limited by the fact that slide film contain much less in the highlights than other film types. The ability of C41 and BW to preserve information in deep shadows or bright highlights is nothing short of amazing, imo.

 

But even on frames where the highlights are quite "blown", and the dynamic range is thus pushed to the left, the shadow areas tend to be quite deep. It depends, naturally, on the scene in question, how contrasty it is etc, but with slide film I often find that it takes a lot of extra time in post (scanning and pp) to dig out the information in the deepest shadow areas.

 

Vuescan does have the possibility to do "multiple exposures", which is called multiple "samples" (Input | Number of samples). According to the Hamrick website this means the scanner makes multiple exposures for each scanned line. This is hardware based and I actually don't know if my 9000 can do this as I've never tried it. It may, if possible, be the best way forward since it would balance time usage and effect/result. I will definitely look into it so thanks very much for suggesting this.

 

Vuescan also includes the ability to have the scanner do several passes (as in making full scans) and then average the results (Input | Number of passes). I know from past testing that this quickly becomes time consuming. And from my earlier tests on my Coolscan V the results didn't wow me enough to use it regularly given the extra time outlay. It might be different on the 9000 since it is a more capable scanner with different light source etc so I will certainly give it a try.

 

Your comment about doing separate highlight and shadow scans and working on them in post reminded me of an old Photo.net article that I had forgotten about called Multi-RAW processing. I'll revisit that too to see what I can obtain. Thanks for reminding me.

 

The trouble with all these techniques, however, is that they require a lot of extra time in post (though the multi exposures option, if possible, seemingly less so). This would make them useful only when absolutely necessary on a particular image that one wants to, say, print to particularly high quality. It would be great if it were possible already at the exposure stage to capture (gah, I hate that word but it seems to work here) as much information as possible in the shadows

 

Again, thanks for taking the time to think about this issue, it's very helpful to me.

best

Philip

[wattsy]

Philip, you might try pulling Velvia a stop or more (shoot and develop at ISO 50 or 25) if lighting conditions make that film speed a practical proposition. This should provide a little more dynamic range but you might find at the expense of introducing unusual colour casts (which might be easily dealt with in Colour Perfect). In truth, I think with Velvia it is best to accept that it is a contrasty saturated film and use it accordingly (i.e. play to its strengths).

[xalo]

Advertisement (gone after registration)

Agreed. Velvia can come up with unique looks in high contrat situations:

http://www.l-camera-forum.com/leica-forum/landscape-travel/308766-fort-coefficient.html#post2556056

 

Philip, I think that slides can keep very smooth highlights when exposed for them.

Alexander

Edited December 1, 2014 by xalo

[philipus]

I've now done a few tests using the below fairly contrasty scene which I hope poses an adequate challenge to test how various scanner functions can obtain more shadow detail.

 

The image was scanned on the 9000 with the following general settings (these are the settings I imagine could impact the results so feel free to ask about other ones; I am trying to avoid a very lengthy post):

 

Input

Media: Slide film

Scan resolution: 2000dpi

 

Filter

Infrared clean: Light

 

Color

Color balance: None

Curve Low/High: 0.251/0.75

Brightness General/Red/Green/Blue: 1

 

I scanned as linear TIFFs for use with ColorPerfect. Each image was simply opened in CS6 and then in ColorPerfect's Pos mode where I just OK'd out without making any adjustments. Then saved as 8-bit JPEG at a compression to meet the forum's requirements.

 

This is the full image scanned with only the above, basic settings, and a crop. Crops of the image scanned with other settings follow below.

 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

[philipus]

Here are the crops.

 

With Input-Fine mode ticked.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

With Input-Multiple exposure ticked

 

With Input-Number of samples set to 3

 

With Input-3 samples AND Fine mode

 

With Input-3 samples AND Fine mode AND multiple exposure

[philipus]

I have to admit I don't see much - if any at all - difference between the base scan and the other scans. Perhaps I made a mistake in the settings or perhaps I need to set a much higher number of passes, I don't know. Possibly it's not a good image for this test?

 

I'd very much welcome your comments and thoughts.

 

Philip

[A miller]

Hi Philip - Thanks for the thread. My set up sounds similar to yours in that I use vuescan and a nikon 9000 coolscan. I find slide film tricky in general, as I do vuescan. Together, my brain nearly bursts thinking of how the scans can theoretically be optimized using the many mysterious setting of vuescan.

 

I don't notice any difference in your crops, and frankly think that there is way too much dynamic range in your image to be able to capture the shadow areas.

 

My workflow is to scan at the max dpi and make sure that I use a light infrared clean only if needed and then I try to get the WB right. I find that if the scanned file comes too cool, the shadows and blacks will have a bluish hue to them which is only removable by desaturating the blues in the image, which is not optimal if there are other blues in the photo that should be retained.

 

I scan in fine mode, but I don't make multiple passes to extract the max out of shadow detail b/c I don't find that this really does anything with the coolscan 9000. If anything, it negatively affects the scan b/c the scanned image could appear over-processed in the sense that there is too much RGB grain noise (I find that vuescan also tends to over-process the scan of images with lots of bokeh).

 

I think that, to understand whether you are getting the most out of your slide film, or any film, is to have the film manually printed in an analog fashion, with the instructions to open up the shadows as much as possible. I think this is a reasonable benchmark for assessing whether your scans are getting you better or worse shadow detail than an analog print. I personally would be satisfied as long as the scanner is giving me at least as much detail.

 

I also think that your goals might be at least partially achieved by underexposing your shot by a half stop or so. This should preserve decent shadow detail while leaving the potential to increase the exposure of the light tones in PP, although you may need to do this surgically through local brush adjustments.

 

Thanks again for sharing your controlled experiment and good luck.

 

Adam

[philipus]

Hi Adam

 

I'm sorry for the very long overdue reply. Thank you very much for sharing your experiences on this. And also for confirming my feelings about the crops.

 

I've now had the chance to install Nikon scan 4.0.2 on a Snow Leopard volume. It's part of a project of mine to compare the results from Vuescan and Nikon Scan using a few set image types.

 

While I think I know both programs quite well I wouldn't call myself a real wizard with respect to either. I used Nikon Scan ten years ago on Windows XP so I'm a bit rusty. But, while it is perhaps not as quick to use as Vuescan (due to how the Preview and Scan functions are implemented, see below) it has a few features which are more straight-forward to use. I am thinking here about the various options under Digital ICE, including the shadow/highlight recover tool DEE (Dynamic Exposure Extender in Nikon-speak) and related tools, like the GEM grain reduction tool.

 

I tried DEE on the above image and thought I should post the results. I don't believe Vuescan has this function, strictly speaking, but similar results can perhaps be achieved by various settings on the Color tab or in Photoshop (but perhaps not entirely given how DEE is implemented).

 

The first two were scanned in Nikon Scan at 2000dpi, no ICE (which thus differs to the Vuescan image, which had ICE on the lowest setting). I made a quick (and poor, sorry) colour correction in CS6+Color Perfect to make it easier for the eyes to compare with the Vuescan image.

 

This image has no DEE applied

 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

This image has DEE shadow recovery at 25, highlight recovery at 20 and the threshold at 220.

 

 

And here's the Vuescan image with the magnified parts.

 

 

A few things are obvious.

 

From the two crops to the left it is clear that DEE does pull out a lot from also dense shadow areas. I tried several different settings and it can pull a great deal from shadows, much more than I had expected. The trade-off is exaggerated grain. I expect this to be even more visible in C41 given how the Coolscans handle film grain. It may be possible to some extent to counter the effect by using the GEM function.

 

It is also clear that the Vuescan image isn't too bad as a starting point. It is certainly not as bright in the shadow areas as the Nikon Scan image with DEE applied, but it is better than the image without DEE.

 

The top right corner crop is a 200% crop (all other are 100%) showing the difference in grain from the two scanning programs. It appears that that Nikon Scan produces a slightly different grain structure than Vuescan, compare with the non-DEE image. I expect this is in part due to the fact that ICE was on in Vuescan (on the lowest level) but I don't think this is the only reason.

 

Another difference is the highlight recovery in DEE. That pulls out a bit more detail in the old man's hat (a few more faint details can be seen, plus the hat doesn't blend into the wall like it does on both the Vuescan and the non-DEE images) and it also removes the impression of the yellow wall being entirely burnt out.

 

In use, DEE's highlight recovery is similar to ColorPerfect's.

 

In ColorPerfect, one set's a threshold above which the recovery should work and then pulls a slider. Then one can "recover" highlights in increments of F-stops using a pulldown menu.

 

DEE lacks the F-stop recovery part but has the threshold and "intensity" settings (though threshold may work slightly differently, I'm not sure).

 

DEE's highlight recovery seems a bit better at recovering details. I am saying this because a problem I have found with ColorPerfect's highlight recovery is that it tends to just "dull" blown highlights on slide film (which is most of what I shoot) - making them slightly gray or less luminous - without bringing out any further detail. On a C41 image one would be able to recover a bit more depending on how blown the highlights are. For many images the extra "bite" given by bright specular or other highlights is more appealing than the dull highlights that ColorPerfect can produce.

 

I suspect the reason ColorPerfect results in dull highlights, particularly on slide images, is that it is applied to an already scanned image. In Nikon Scan, one first has to do a preview, then select the DEE settings and then "redraw" the preview (which is done from memory without a new preview being done). Once this is done, one scans.

 

Interestingly, the Vuescan image appears somewhat sharper even though no sharpening was applied, either in Vuescan or in CS6.

 

Anyway, these are just some preliminary results. I will put together a more exhaustive comparison of the two programs with a set of images and using various settings. I expect that there won't ultimately be a one-size-fits-all solution. Some images simply need more work, and on top of that it also depends on what the image will be used for, one's preferences and things like that.

 

br

Philip

[tobey bilek]

for future films, you can pull process. Use iso 50 and cut first developer time.

 

Details come with gallon Kodak kit . It works well. U used it to make slides that print on Ciba Chrome.

 

For existing films, use HDR. Scan one time for highlights and once for shadows and layer up in photoshop. Use arrow keys as required to move layer one pixel at a time.

[philipus]

Thanks Tobey. I don't develop my own E6 however the "HDR" method may be worth a try.

[wda]

Philip I can't help you directly because, facing the limited dynamic range of Velvia I switched to one of the latest Ektachromes for scanning in my Coolscan. I always use high multi-pass scanning in Nikonscan software and adjust the levels and/or curves before final scanning. Unless you want to project your slides, why not bite the bullet and try a lower contrast negative colour film? You might be pleasantly surprised.

[philipus]

Thanks for the suggestion David. I do use several types of colour negative film as well and do enjoy that very much.

[budrichard]

To put it simply, there are some scenes because of the dynamic range of the scene that exceeds that capability of analog color reversal film to capture with one exposure.

One can use lighting where possible to decrease the dynamic range, fill strobe, large reflecting panels are useful and used by professionals both in analog and digital photography.

Digtial HDR is certainly an option and overlaying digitally two digitized analog scenes another but I prefer to change the lighting and not fight the problem in post processing.

There are some scense that I just don't photograph. -Dick

[250swb]

Philip, I can't help but think you are trying to get too much out of the actual scan and not thinking of it as a starting point. Your scanner is recording a lot of detail in the scan but you aren't seeing it because you aren't bringing it out in post processing. But to see more of the shadow detail in the raw scan try setting your Black clipping point in Vuescan to zero or near zero. It will make the blacks greyer but again you can compensate in post processing.

 

Steve

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[A miller]

+1 for setting the black and white points to 0 in Vuescan as a general matter. Sometimes, though, I will increase the white point one notch to 0.2 in order to pump the natural highlights just a bit. I feel that setting the B and W points to 0 will force the scanner to tame all highlights (i.e., prevent all highlights, even the natural ones, from being blown). I like this as a general matter, as the highlights can usually be restored in a customized fashion in PP and it provides more flexibility. But sometimes the image has naturally blown highlights that I will want to remain in the image, and taming them in the scan will correspondingly mute the vibrance of the other tones, which is just more work to restore in PP.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[250swb]

Yes same here, by default I set the Black and White clipping points to zero and then only tweak the White one if needed otherwise the image can be too flat. My edited version of the image was a simple trip to Viveza (in Nik Suite) and the shadow adjustment tool, all of five seconds effort against the time of scanning and re-scanning to drag shadow detail out (when it is there all along).

 

Steve