How's my colour neg scanning and Post Processing?

[Ranchu]

White point

White point

The scanner translates the white point and primaries of the film into the white point and primaries of your output space.

The white point of negative film is orange, not white. The relationship between the orange WP and the orangish primaries, and the orangish primaries to each other has to be accurately known before any correction for color or the mask can be done, because the corrections take place in an RGB environment that assumes 128, 128, 128 is neutral. With negative film it's not neutral, neutral is something else.

So the most important thing is that whatever variation of orange everything in the picture is, it is accurately recorded. In order for that to happen, the software must know the sensor response when it's hit by the light source in the scanner. If the software knows this, it can define the scanner's characteristics and then the scanner's relationship to the white (orange) point and (orangeish) primaries of the film, and translate them into the output space in their proper relationship to each other. If it doesn't, it's a crapshoot because nothing is defined.

 

[Pete B]

Tab until you get to the L, a, and b inputs and zero out the a and b. You should now have an L reading of that spot you option clicked. Hit enter twice, then click the mouse. This will set the spot you just clicked to neutral. Instead of doing the mouse dance I described above, you could use a color sampler to get the L value and then just click with the white dropper on that sampler after setting the target values to have a and b zero and an L of whatever the sampler reads.

Anyway, a bit confusing, but it seems to work well. At some point I'll write some of this up if people are interested.

Yes, write it up with pictures. I don't seem to have a or b values to tab to:

View attachment 83986

 

[Tim Gray]

Hmm. Don't know why they aren't there. Is that Photoshop or Photoshop Elements? Elements might not have it...

 

[Ranchu]

Using Saturation 0, and brightness might work...or reverse the hue 180 degrees?

 

[Tim Gray]

Yeah I just played around in PS. Setting the saturation to 0 in HSB definitely changes the L value. I'll have to get back to you on this.

 

[Ranchu]

No problems, Tim.

🙂

 

[shadowfox]

Serious?

I am serious.

The original color cast adds to the cold atmosphere of the scene, I wouldn't have tweaked it.
Your modification does not enhance the original, in my personal opinion.

Do you really think photos without color casts is *always* better?
If so I think you'd be much happier shooting digital. 🙂

 

[tokengirl]

Thanks for that Tokengirl. I do use similar sliders in NX. Unfortunately, as I increase magenta to remove green the blue cast to the grey/brown hat remains. It was this issue that made me start this thread. It's as though the whole image is cross-processed somehow. Often these sliders work the trick but for some reason not these. Now where's that brick wall? Ah, yes....:bang:

Pete

Well, if you want to remove a blue cast, then you need to basically add yellow - by increasing the temperature of the white balance. See below:



The thing is, when you do that, then the skin tones can get funny if you're not careful. All films are different, they all have their unique characteristics, and none of them accurately records every single color. So sometimes you have to make a choice - in this case I think it's either nice pink skin tone, or a more accurate representation of the hat's color. I don't know about you, but for me, skin tones are more important - most people that will see this photo have never seen that hat in person anyways so they'll never know the difference. But everybody can see when someone's skin is too yellow because they look ill.

 

[Pete B]

Yes, I agree with you Tokengirl. I would aim for skin tones normally. I think you're right that getting all the colours decent with this image is going to be difficult. I actually like the cross processed look just as Shaddowfox but I guess I'm saying that I'd like to get a normal looking image ideally and let any subsequent crossed-processed look be my choice.
I went back to look at the original file on my system and reduced the green with the sliders just as you suggested and there is indeed an improvement. Thanks for taking the time to comment and give me some examples and pointers. I appreciate it.
I PMed a guy on another forum who is also very knowledgeable, and he pointed out that Fuji 800Z might not be the easiest film in the world to scan and perhaps I might be aiming a bit high.
Thanks again.
Pete

 

[Pete B]

Well, just an update. Following our discussion, today I decided to have a go with Epson Scan (which didn't take long before I felt like throwing the scanner out of the window) and SilverFast. I'd always been reluctant to put SilverFast on my system as I'd seen reports of it crashing constantly, but I also liked the sound of NegaFix. I have to say it's been a very pleasurable experience. Whereas this image went through Vuescan, into Elements and opened up with ColourPerfect before being saved and further modified in NX,


View attachment 84000




Image 2 is straight out the scanner with one click of the grey point in the sky with NX:


View attachment 84001

The attachments are a little small for you to appreciate but I can tell you the Silverfast version looks great. All scans came out better than those from my previous laboured workflow. At last, now I feel I'm starting to see the film's characteristics.
I know what I'll be doing for my next scanning session. Thanks for all the pointers everyone.
Pete

 

[Tim Gray]

Wow, I like the first one more 🙂 I like the contrast in the face on the second version a bit more, but for skin tones, I find it too pink or something.

But all that counts is that you are happy. 😀

 

[Pete B]

Wow, I like the first one more 🙂 I like the contrast in the face on the second version a bit more, but for skin tones, I find it too pink or something.

But all that counts is that you are happy. 😀

Oh, yes it's a bit magenta. The point is to get to this point previously involved loads of time. 2 more clicks in NX got me

View attachment 84007


Which is pretty good I think for a minute's work. Fine for me anyway!

I felt the first image had a bit of a green tinge.
Pete

 

[Tim Gray]

No doubt. You know what works best for you.

 

[Pete B]

No doubt. You know what works best for you.

Do you still prefer the first? Do you suspect my monitor isn't calibrated (you might be right 😱).
Pete

 

[Ranchu]

Looks good! I'm curious what the other one that's been in the thread looks like?

 

[Tim Gray]

Nope not all. I know my monitor isn't calibrated. If the new method is much much faster than it sounds pretty good.