colour negative scanning

[nwilkins]

Hello everyone,

I am wondering if someone can recommend a really good comprehensive resource about scanning colour negatives?

I use an epson V700 and I have tried to do it the way I scan b/w (creating a file with low contrast and as much info as possible) but most times I never end up getting the picture to look as good as the auto levels in epson scan.

So I have been leaving it up to epson scan to do the auto leveling for now, but I suspect it is getting rid of the colour cast by clipping info on both ends of the histogram, which I don't really want.

Also sometimes (especially for a mostly white, snowy scene) the auto levels produces a really poor image with severe colour cast (often reddish).
I downloaded a trial of vuescan hoping that its auto levels could help with the snow scenes but none of the colour film presets (generic or otherwise) was able to produce a decent image.

So I'm wondering if there are any tutorials or books that provide a good workflow outline for scanning the image with as much info as possible and then colour correcting the image, removing the blue cast etc manually.

thanks in advance.

 

[L Gebhardt]

I found this the other day when I was starting to write up how I scan color negatives. The page was written by a member here by the name of pellicle. He seems to do things the same way I finally figured out works best on my scanners. Understanding how the color curves in the film align is critical for getting decent color out of negatives. Give it a read, and if it doesn't make sense ask specific questions here.

 

[nwilkins]

okay thanks a lot Larry - I will have a look at that and post back with questions!

 

[pellicle]

I found this the other day when I was starting to write up how I scan color negatives. The page was written by a member here by the name of pellicle.

and since I seem to have time to examine this forum again please feel free to ask me ... my methods may not be perfect and I continue to learn. So feel free to question what I say and ask for clarification, elucidation or justification.

BTW I just pulled something unbelieveable to me out of an old 110 neg
in my view ...: Film: pulling images out of the mists of time

the neg had been in the camera undeveloped for between 15 and 20 years.

 

[nwilkins]

okay so I'm getting around to doing this now. I am scanning as a positive but I don't know how to make sure I am not clipping any of the channels and still get decent colour.

I am using epson scan and using the levels tool. I set output range to 0 and 255, then moved the black and white points of all three colour channels to the edges of the histogram, then moved the overall black and white points to the edges of the histogram. I ended up with a positive image (before inversion) that looks greeny gray in most places. Do I need to fix this using the midpoint slider? Where should the midpoint slider be on each channel?

thanks!

Nick

 

[pellicle]

Hi

okay so I'm getting around to doing this now. I am scanning as a positive but I don't know how to make sure I am not clipping any of the channels and still get decent colour.

Photoshop provides an excellent tool for this it is the level tool.


Firstly scan with some leeway in the humps of dark and light levels. You can see some of that clearly but if you get it wrong then come back to this point after you've seen this in the next step.

Next, lets assume a 'normal' scene of outdoor brightness, with fully dark and fully illuminated parts. Now for each of the R G and B channels, set the dark and light points. Hold down the Alt key as you move the slider for dark and light points. You will see the entire screen go dark or light and as you move the slider you will see where clipping starts as the change from totally black (or white) to red (green or blue depending).

If it starts off with red (green or blue) in the black (or white) you have already clipped in scanning and go back to scan again with less agressive settings.

Personally I find that setting the dark areas of green and blue to 0 is needed. The red often has its full set in the middle.

I am using epson scan and using the levels tool. ...Do I need to fix this using the midpoint slider? Where should the midpoint slider be on each channel?

the mid point slider essentially adjusts the 'gamma' of the curve. I would recommend doing this much later. At first in epson scan just get it into photoshop without loosing any data.

Trim generously (when in doubt give more) and clean up in Photoshop with better tools.

Next when you get to know a film you can do this by numbers. test a film and see what really really fully over exposed gives (in density of the darkest parts of the film) and you know that's the D-max for that film. Then look at the meerest slivers of murky shadows (the thinnest parts of the exposure just above the base fog) and use those numers.

From then you'll be able to scan by numbers straight into photoshop and reduce your cleanup times for difficult exposures (where everything is dark on purpose or everything is light on purpose).

depending on things (like your image) sometimes things respond well to auto colour balance, sometimes they don't. Essentially you need to get a feel for the film and the lighting (IE in daylight things will need different tweaking to in shadow or indoors). Once you grasp these things you can work quickly (without feeling rushed) to get the image 90% right in moments.

after that, the last 10% takes as much time as you're willing to put into it.

Colour is never easy compared to black and white. Negative is the RAW of the colour world. It may take some time but you'll get stuff that you just can't get out of digital.
HTH

PS: this weekend I tested a Bessa with Negative. I was so impressed I took a digital with my Panasonic GH1. Talk about chalk and cheese.

http://cjeastwd.blogspot.com.au/2013/04/bessa-i-with-colour-skopa-lens.html

as my monitor was not colour balanced I didn't try hard to get the colour right. None the less the differences between renderings and detail are startling.

Film

Digital

On the neg the trunk is about right, as are the fruits on the palm and the greens of the frangipani tree leaves (red channel seems to have the wrong HUE cast) ... yet overall neither represents the scene as it was.

With work either could do well, but the 120 negative strikes me as having better potentials

lastly tweak HUE on the RED YELLOW and CYAN

 

[nwilkins]

okay so in terms of the actual scanning it makes no difference what I do as long as I just don't clip any of the channels? and everything just gets fixed afterwards in photoshop?

 

[Alan Klein]

IS there a way to change Auto Levels at each end of the range before you do the Auto. I other words, tweeak the settings as if you'er going all manual control but then hit auto at the end?

 

[L Gebhardt]

okay so in terms of the actual scanning it makes no difference what I do as long as I just don't clip any of the channels? and everything just gets fixed afterwards in photoshop?

I think that depends on the scanner. Some drum scanners adjust the amplification based on the black and white points. Other scanners do adjust the scan time time or lamp brightness based on settings, so I have heard. But on the simple CCD scanners like the Canon and Epson I have I think it makes no difference. However I still try to get the ranges close, just in case the scanner uses the info to adjust the scan.

 

[nwilkins]

okay thanks - I'll move on to the photoshop manipulations then - I'll be back with questions, but for now:

does it make a difference if you adjust levels in photoshop before or after you invert the image? Is one way easier?

thanks!

 

[L Gebhardt]

No, it shouldn't matter if you invert before or after the levels as far as image quality goes. But I would do the inversion first, and then create a layer for the levels. It will be much cleaner that way.

 

[nwilkins]

thanks for the tip!

 

[pellicle]

okay so in terms of the actual scanning it makes no difference what I do as long as I just don't clip any of the channels? and everything just gets fixed afterwards in photoshop?

as long as you scan as linear positive in 16 bit and don't clip that's right. I'd also put at least the dark point of blue channel hard left on the slider scales as that can make a difference too (green too, but red doesn't need it). I have found on the 3200 for sure this makes a difference to noise in the blue channel.

 

[pellicle]

But on the simple CCD scanners like the Canon and Epson I have I think it makes no difference.

my observation of the operation of the Epson CCD scanners such as the photo perfection range is that if you put it in professional mode and set the levels manually it makes no adjustments. It is quite reproducable and I get the same results when scanning a stouffer step wedge (which is of consistent density) every day I've tried it.

just make sure you're in positive mode not negative mode

 

[philipus]

Umm I might be wrong but I'd say you have to adjust levels after the image's been inverted. It would be really difficult to see the results in an inverted image.

 

[pellicle]

Umm I might be wrong but I'd say you have to adjust levels after the image's been inverted. It would be really difficult to see the results in an inverted image.

Which is why you use the Alt + "move the slider" method in photoshop so's you can see them approaching clipping. Its a very handy tool. Doing it by eye is fraught. In that case (the case of not having photoshop) I'd invert first, be conservative with the levels and use curves to tidy up the shoulder and toe.

 

[philipus]

Yes I agree Alt + slider is a great tool in Photoshop and would make it easier to use Levels on the negative image. I use ColorPerfect though so in my workflow the image would always be inverted before I would use any other tool.

 

[pellicle]

Hi

I use ColorPerfect though so in my workflow the image would always be inverted before I would use any other tool.

I would recommend you do some comparisons with fully manual and using Colorperfect and compare noise levels in blue channel and amounts of clipping. Last time I did that I noticed a lot of difference. Perhaps its changed since then.

From my post here

 

[philipus]

Thanks Chris. I'll definitely give that a try on the 9000 (I tried it on my V ED but didn't notice much difference unfortunately).

Cheers
Philip

 

[pellicle]

Thanks Chris. I'll definitely give that a try on the 9000 (I tried it on my V ED but didn't notice much difference unfortunately).

ahh, you're using a Nikon ... I don't think you'll see that sort of difference (but its not expensive to have a look). I believe that the circuits in the Nikon (and when you add in the multi-sampling) are less noisy than the Epson.

That was about Epsons.

As an aside, its been interesting to see how the Epsons have been developing. I've have had (and still own in one place or another) 3200, 4870 and 4990. The differences are interesting.

even further aside is that for 4x5 neg scanning the 3200 (using the green channel only) is a very capable scanner. It'll do 2000dpi all day - every day.

 

[nwilkins]

it seems through my limited practice in the last few days that during the scanning stage it is far better to set the overall black and white points first before adjusting each channel. would you say this is true?

 

[pellicle]

it seems through my limited practice in the last few days that during the scanning stage it is far better to set the overall black and white points first before adjusting each channel. would you say this is true?

to mu understanding (assuming we are talking Epson scan software) as soon as you set each channel black / white points what ever you set for the overall is immediately changed.

I usually set the black points a bit conservative (meaning darker than I can see anything in the histogram hump). I can usually get the absolute blackpoint by including some of the mask ... not that that is very useful.

I always find that Blue has a deeper blackpoint than Red ... which is unsurprising really when you look at the transmission data that the film makers publish:

if you look at the density of blue its about 2.5 at near its most dense (RHS of graph) while red (for the same light exposure) is more like 1.75

remember its a log curve, so 1.75 to 2.5 is a long way

 

[nwilkins]

okay so I have done a few negatives and it seems I am having no problems making sure there's no clipping at the scanning stage. Then I am going into photoshop and using the alt key to set to black and white points right at the limits for each channel, then inverting.

Since I am much more familiar with Lightroom than photoshop, I would prefer to do colour correction in lightroom rather than PS (I have no idea how to do it in PS). However is there anything that I should do from a colour standpoint that is much easier/more powerful in photoshop, before I import into Lightroom and make subtle adjustments?

 

[pellicle]

Since I am much more familiar with Lightroom than photoshop, I would prefer to do colour correction in lightroom rather than PS (I have no idea how to do it in PS). However is there anything that I should do from a colour standpoint that is much easier/more powerful in photoshop, before I import into Lightroom and make subtle adjustments?

I've never used lightroom, I thought its key benefits were for making adjustments to digital images (which will be in character quite unlike scaned negatives) which were captured with consistent lighting values and that you wished to adjust them in bulk.

Why not make the ball park adjustments in PS and then tweak in LR?

 

[nwilkins]

I haven't seen anything in Lightroom that makes me think I shouldn't be using it for scanned film. The only difference seems to be that photoshop is useful for much more drastic editing of photos, which I generally don't need to do. But maybe photoshop has better tools for editing colour at this stage of colour neg scanning? What tool should I be using?