Scanning Consistency (Warning: Pictures)

[tgray]

Advertisement (gone after registration)

Pete, we've talked about this stuff offline. I don't see anything wrong with your method or results. Keep in mind that sometimes the color in an image is just plain awkward, no matter what adjustments you make. I think you've been staring at the picture of the woman in front of the vista for too long - it doesn't look as strange to me as I think it does to you.

 

I do something similar to what Pete is doing for my color negs. When I do it, there is a step before (not shown) where you take the linear gamma scan and change the gamma of the file to 2.2(ish). I'm not sure if Pete is doing that; I think he is. If so, it's really not any different than what ColorPerfect or scanner software does.

 

If you don't like ColorPerfect, don't use it. If you do, use it. I think it gives pretty good color, but has a horrible interface, and sometimes does real funny things to saturation. If you can get more consistent color using your scanner software, Lightroom, or some other product, by all means use it. I find that for me, it's easier and faster to scan as a linear file and do all the work in Photoshop. That way I can run an action set that does most of the work for me. I also find that the more I do it, the better and quicker I'm getting at correcting for color casts in difficult situations. I've done multiple tests, comparing my method, Vuescan, ColorPerfect, and Nikon Scan, and for me, doing it all in Photoshop consistently gives me the best image very quickly.

[Advertisement]

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

[Stealth3kpl]

 

I will do screen shots of my workflow with C41 if you wish.

 

 

That would be great, thanks.

 

 

 

Your issue with the clipped shadows in the example shown above sure isn't intrinsic to ColorPerfect but rather a problem with the scan exposure or the black point setting or both..

 

I think the exposure is good:

 

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!

 

 

 

I look at this guy's portraits shot on Portra 160NC and wonder how he can get such lovely results bringing out the subtle colours of film yet very few people seem to be capable of achieving such nice results. I think I got close here with the Levels technique:

 

 

 

 

People talk about how nice the tones of such a film are compared to another but I'm left thinking, since scanning and use of scanning software is so variable, how can they tell?

 

Pete

[Stealth3kpl]

Pete, we've talked about this stuff offline. I don't see anything wrong with your method or results. Keep in mind that sometimes the color in an image is just plain awkward, no matter what adjustments you make. I think you've been staring at the picture of the woman in front of the vista for too long - it doesn't look as strange to me as I think it does to you.

 

I do something similar to what Pete is doing for my color negs. When I do it, there is a step before (not shown) where you take the linear gamma scan and change the gamma of the file to 2.2(ish). I'm not sure if Pete is doing that; I think he is. If so, it's really not any different than what ColorPerfect or scanner software does.

 

If you don't like ColorPerfect, don't use it. If you do, use it. I think it gives pretty good color, but has a horrible interface, and sometimes does real funny things to saturation. If you can get more consistent color using your scanner software, Lightroom, or some other product, by all means use it. I find that for me, it's easier and faster to scan as a linear file and do all the work in Photoshop. That way I can run an action set that does most of the work for me. I also find that the more I do it, the better and quicker I'm getting at correcting for color casts in difficult situations. I've done multiple tests, comparing my method, Vuescan, ColorPerfect, and Nikon Scan, and for me, doing it all in Photoshop consistently gives me the best image very quickly.

 

Yes, I am doing the 2.2 step (It's step 3 in NX2 - but I couldn't be bothered explaining it so glossed over it). I really only started this thread to see if anyone had an extra trick up their sleeve that I should be doing. I suspect I am doing everything that can be done whilst converting with this method. ColorPerfect sometimes works well (if the exposure is good) but if the exposure is of then it can go out of wack. I'm fairly familiar with ColorPerfect. Perhaps moreso than many.

I have to conclude that I'm doing all I can to colour balance by aligning the red, green and blue histograms, setting black and white points, finding something mid grey, introducing a contrast curve then perhaps messing a little with colour balance sliders to taste. It doesn't sound like there's a secret step that I'm omitting. I, too, often find I prefer the results of this method to the result via ColorPerfect.

Pete

[Stealth3kpl]

The gamma is adjusted for each R,G,B channel to 2.2 (no longer linear Gamma scan)

 

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!

 

 

 

 

 

Pete

[tgray]

Cool.

 

I've got this one shot that I've been using as one of my torture test shots. It's a Portra 800 shot, indoors under nominally tungsten balanced fluorescents, so it has a nice green cast too it. Most of the workflows get a decent version of it with the green cast, which is only natural, since it's presumably in the negative. However, correcting out the green cast is pretty straight forward using the curves method that you outline. You just push the black point of the G layer up a lot higher than you might think you should at first, and push the white point of the G layer up by almost the same amount, so it's not right at the right edge of the G histogram. Of course you could do this correction to a negative inverted by another method, but this method kind of catches it near the beginning.

 

I guess what I'm saying is this: using curves kind of demystifies the whole process, which I find satisfying. It's helped me interpret color negatives better and let's me address color shortcomings easily. I can identify what's not right in elementary terms (too much green, etc.) and once I do that, the solution is straightforward - adjust that curve to achieve the desired result.

[Stealth3kpl]

Although this thread wasn't intended as a comparison of the two techniques, I thought I'd post the levels and the ColorPerfect versions of this file. CP does very well although the sky wasn't blue on the day.

 

Levels version:

 

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!

 

 

 

CP version:

 

[Stealth3kpl]

Advertisement (gone after registration)

Cool.

 

I've got this one shot that I've been using as one of my torture test shots. It's a Portra 800 shot, indoors under nominally tungsten balanced fluorescents, so it has a nice green cast too it. Most of the workflows get a decent version of it with the green cast, which is only natural, since it's presumably in the negative. However, correcting out the green cast is pretty straight forward using the curves method that you outline. You just push the black point of the G layer up a lot higher than you might think you should at first, and push the white point of the G layer up by almost the same amount, so it's not right at the right edge of the G histogram. Of course you could do this correction to a negative inverted by another method, but this method kind of catches it near the beginning.

 

I guess what I'm saying is this: using curves kind of demystifies the whole process, which I find satisfying. It's helped me interpret color negatives better and let's me address color shortcomings easily. I can identify what's not right in elementary terms (too much green, etc.) and once I do that, the solution is straightforward - adjust that curve to achieve the desired result.

 

I think this is why I strayed from using ColorPerfect despite it giving good results alot of the time. I feel a bit more in touch with what's going on and find the process quite therapeutic in the evenings.

 

When, for your Torture Test shot (nice term - like my lady shot), you say "You just push the black point of the G layer up a lot higher than you might think you should at first" do you mean you're clipping it at the black end or that you're moing the whole of the green histogram to the right as shown here:

 

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!

[tgray]

When, for your Torture Test shot (nice term - like my lady shot), you say "You just push the black point of the G layer up a lot higher than you might think you should at first" do you mean you're clipping it at the black end or that you're moing the whole of the green histogram to the right as shown here:]

 

 

I'm clipping it more. It actually moves the whole green channel of the histogram to the left. Which makes sense - there's too much green, so you turn the whole dang thing down.

[Stealth3kpl]

ColorPerfect is a godsend in terms of getting the best results imo, but I sure wish the developer would get some help with standard interface implementation

 

I have chatted with the developer (Christoph) in the past about this issue and he is very aware of it. He's also aware of the need for a decent user manual. I have a number of emails from him over which he has spent a great deal of time explaining various issues. It's commendable that he should spend so much time ensuring that a user is properly informed. I'm very grateful to him.

Pete

[Stealth3kpl]

I'm clipping it more. It actually moves the whole green channel of the histogram to the left.

 

That's interesting. I won't feel I'm doing the wrong thing if I have to resort to that for an image.

One thing I've seen on the web is people altering colour balance by altering the "Gamma" of the individual R,G and B channels as seen in the first part of

. This seems wrong to me(?). This Photoshop app (5 mins in) seems better in that colour balance can be altered in dark/mid/bright tone regions individually (but I'd have to buy Photoshop)

Pete

[tgray]

Yeah, you can do that too. I prefer to start with adjusting just the end points, because the midpoints move when you do that. Then, if you feel the need, move the midpoints/gamma too. I wouldn't feel like you are violating any principles here. If it gives you color you like, then it gives you color you like.

 

Moving just the midpoints/gammas of each curve can help, and can even get you most of the way to your final color, but it leaves the shadows and highlights behind sometimes. If you work with the endpoints first, the midpoints might just come out fine with no need for adjustment.

 

Also, I'm not afraid to clip a channel or two on either/both ends in your adjusting. It's very useful to avoid this in the scanning stage, but in the adjustment stage, sometimes the shadows or highlights *should* be clipped on a channel (or two).

 

In motion picture, these kinds of corrections are often done with a 3 way color corrector, or more simply, with lift/gamma/gain. You can do all this with curves (I think), but I do wish we had lift/gamma/gain controls. I'm color correcting a film short I shot now and those controls make things very straightforward to correct.

[Caterman]

If you want to go deeper into color correction and making your image pop you have to check out Dan Margulis the absolute king of color and the LAB color space, theres only so far you can go in RGB. Getting the color you want is a massive subject!

[Stealth3kpl]

The gamma is adjusted for each R,G,B channel to 2.2 (no longer linear Gamma scan)

 

[ATTACH]245835[/ATTACH]

 

 

Pete

 

Just for clarity this Gamma shift is done before the inversion step.

Pete

[plasticman]

Regarding ColorPerfect - I can live with the 'unorthodox' interface because of the results. I'm actually really grateful for eccentric developers like this: a genius with color, but a cludge with human interfaces - I can live with it because they make something possible that otherwise would take a lot more work and effort on my part.

 

Incidentally, regarding the fill-in of blacks with CP: I also had this problem until I realized you fine-tuned the shadow tails by adjusting the Black Point and then the Gamma to compensate for the overall lightening of the image. Each of these alterations needs to be done with great sensitivity, but the end result is immeasurably better, with no filled blacks.*

 

Edit: I ought to add that this is only necessary in a few cases, and not a procedure I go through with every image: it's something I'll do when there's a noticeable mismatch in the exposure and the default image that I get from CP - usually a default result that looks way too contrasty.

[Stealth3kpl]

regarding the fill-in of blacks with CP: I also had this problem until I realized you fine-tuned the shadow tails by adjusting the Black Point and then the Gamma to compensate for the overall lightening of the image.

 

Much better. I'd been adjusting Black P then white. I'd forgotten about the gamma

 

 

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!

[plasticman]

Much better. I'd been adjusting Black P then white. I'd forgotten about the gamma

 

Well it turns out you were correct and I was wrong! But only partly - I'll explain... or rather David Dunthorn (the creator of ColorPerfect) will explain...

 

After this thread I thought I'd sit down and thoroughly test the procedure, and the results were still not consistent on the filled blacks. (I must stress that I feel that filled blacks are the only weak point in the entire film workflow now - so I really wanted to nail it).

 

So I started an exchange of mails with Mr Dunthorn, where I tried to thoroughly question the different functionalities of the various controls, and their effect on an image.

 

My particular frustration was that the 'Shadow: Degree-Clip' control seemed to have zero effect on the shadow-clipping tail.

Here's David's final answer - it's long, but following his procedure fixed the filled blacks perfectly, without affecting the contrast or tonality of the rest of the image:

 

----v----v---- David Dunthorn's answer ----v----v----

"...detail lost in BP clipping is not affected by the Shadow Degree control. So, when you ask "My main question really is: is the plug-in operating as intended when the Shadow clipping controls have no effect on the image whatsoever," the answer is "yes" but both the question and the answer are very misleading because you are trying to use Shadow Degree on the BPoint adjustment whereas it only applies to the White adjustment.

 

Also, as you found, the number in the box in the Shadow Degree-Clip tracks the percentage of tails clipping and is not a user control. It is in fact an estimate based on the image preview and so will change as the preview image changes.

 

Let me explain again:

 

Your stated interest is in bringing out dark shadow detail... To do that adjust B Point so that its number value is smaller. This will bring out any dark shadow detail that is present. Doing this also effectively adds white to the image, just as though you had painted over the entire image with a light coat of white, translucent paint. This is what you refer to as "losing contrast," but actually it is NOT the same as changing gamma.

 

Now, to remove the white that has been added, use the White control to reduce the amount of White in the image, restoring what you have called "contrast." In general this will make the White number value go slightly negative, showing that you are in fact removing White.

 

Contrary to the situation with B Point, shadow detail lost in removing White IS affected by the Shadow Degree control. So now you adjust the Shadow Degree pulldown and you will see the clipping percentage in the adjacent box decrease. Usually the decrease will not be even because normally you are dealing with a high image noise level in the shadows, but it will decrease and the dark shadow detail you want will become visible.

 

These adjustments are easily and quickly done once you understand them, and of course if you find that you are routinely making similar adjustments you can try adjusting the threshold and have ColorPerfect remember the White and Shadow Degree adjustments from one image to the next. But most people will not find that to be necessary."

----^----^---- David Dunthorn's answer ----^----^----

 

Thanks for starting the discussion - I'm gonna print the answer and pull it out whenever the filled blacks problem arises again. Like pretty much everything in ColorPerfect, step-by-step instructions are the only way to go!

[biglouis]

Firstly, most film users don't use the Nikon software, so your examples are unfamiliar to most users.

 

Secondly, most users, I would think, don't scan as RAW.

 

I have looked at a demo of ColorPerfect and threw it away within half an hour. The interface is so confusing and complicated.

 

I scan as a positive tiff, bring into Photoshop, and do all my adjustments there. I do adjust levels in colour channels when necessary, as well as curves and luminance levels.

 

Alternatively, I scan E6 and don't have any of these problems

 

I am completely with Andy on this.

 

Reading this thread is doing my head in. I'm not convinced that your workflow is not so convoluted that you are beginning to get lost in what you are doing.

 

Take a simple scan in VueScan and use Photoshop to do the rest. How often have you taken a poor digital capture and processed it into something good? All you are doing when you scan is the equivalent of taking a digital capture of your negative. Sometimes it will be spot on, other times so-so. It doesn't matter because the post processing capabilities of PS are immense, if you know how.

 

I've mentioned it several times in other places but what you need is a copy of "Scott Kelby's 7 Point System for Photoshop" and you'll end up with perfect results. You either love or hate Scott Kelby but personally that is the best value for money PS book I've ever bought.

 

Andy, I would benefit from the offer you make later in the thread to show us your workflow. If nothing else, I would like to see how it varies from mine which I suspect is quite similar (except for my love of LAB space for pushing colours!).

 

LouisB

[sfage]

I think I'd like to second what some people are saying. I scan in Vuescan, I do very little, and off I go to PS.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[CharlesL]

For Wrong2, I found that a delicate increase of red at the 40% mark of the curve/histogram improves the foreground. But might want to apply less to the skin.

[Stealth3kpl]

Starting this thread I was hoping someone would show me how to get good results from scanning. I've shown my workflow. It gives me the best results I can achieve a full histogram for each RGB channel. I would appreciate it if someone could show their workflow start to finish so that I can compare. The whole workflow is needed from scanner set up to photoshop manipulation. Thanks.

http://www.l-camera-forum.com/leica-forum/film-forum/169496-scanning-consistency-warning-pictures.html#post1627966

http://www.l-camera-forum.com/leica-forum/film-forum/169496-scanning-consistency-warning-pictures-2.html#post1676308

http://www.l-camera-forum.com/leica-forum/film-forum/169496-scanning-consistency-warning-pictures.html#post1627966

 

Pete