Help - Film Scans need lots of Levels & Curves Adjustment

[digitalirony]

I am new to film scanning and have been wondering about the scans

that I am producing using a

Microtek ArtixScan 4000t and VueScan 8.3.03.<br><br>Every single

scan looks

washed out and I have to apply some level of curves and levels

adjustment afterwards to correct this. The level of correction

required varies from frame to frame but the examples below are

average samples. The level of adjustment required seems a little

extreem.<br>

<br>

Is this normal and if not how do I go about correcting this at the

scanning stage? I have tried making adjustments in VueScan but there

seems to be very little control.<br>

<br>

I have however in my searching found useful tips on locking exposure

and setting the film base colour which have improved my scans.<br>

<br>

I have been scanning at 24bit colour depth because I haven't got the

disk space for 48bit scans.<br>

<br>Any help would be appreciated<br><br><center>

<img src="http://www.digitalirony.com/raw_vs_post.jpg" width="600"

height="407" alt="raw scan vs after levels and curves"

vspace="15"></img>

<img src="http://www.digitalirony.com/raw_vs_post_2.jpg" width="409"

height="600" alt="raw scan vs after levels and curves"

vspace="15"></img></center>

[emre]

It is up to you. Flat scans indicate that you are capturing a wide tonal range. If you set the black- and white points more agressively (i.e., if you clip a little), you will get closer to the end result. If you are working in 16 bits/channel, you can leave it as it is. IF you are working in 8 bits/channel, it is better to clip a little so you have to do less color correction.

 

P.S.: As soon as I saw Algida I said "Turkey!" In fact, that street looks familiar. Where is it?

[erik scanhancer]

VueScan kind of conservatively adjusts your colors and contrast. Personally I think that is

good, because no information will be deleted that way, leaving you lots of space for editing

to your own liking. However, when you use Levels and Curves it is very important to do so in

16 bit (called 48 bit in Vuescan) in order not to lose vital information due to rounding

factors.

 

I don't know for the VueScan version in its basic confection, but there are settings in the Pro

version to determine what kind of output use you want: for editing, for print, etc. This will

make some difference to the initial output.

[William Kahn]

Jon, what you're experiencing is normal. It sounds like you're already doing the right things in the pre-scan setup and, in fact, it's better to make the major adjustments in your image editor than before the scan.

 

From the raw scans, it looks to me like both shots might have been slightly overexposed, anyway. Not unusual when shooting landscapes over water with a lot of sky. The other shot is backlit, and a fill flash might have helped the foreground exposure. I'm not criticizing your work, just pointing out that quite often, in a raw scan, what you see is what you shot. Happenss to all of us!

[digitalirony]

Thanks Emre,<br>

<br>

Both photos were taken in Olu Deniz in August this year. It was my first visit to Turkey and I really enjoyed it. I did quite a few tours but not being in control of the times of the trips, didn't have to best light (still managed to shoot 15 rolls though :) ).<br>

<br>

Already planning a return trip, and now that I know the area a little I can rent a car and visit all the places I wish when the light is better.

<br><hr><br>

Thanks also Erik,<br>

<br>

It looks like I will have to buy another, larger hard disk and more memory. I don't know if my puny processor will be able to handle processing such large files though, but there is only one way to find out.<br>

<br><hr><br>

Thanks also William,<br>

<br>

I thought that it might be normal but wasn't sure. I haven't quite mastered using partial metering on my canon so I never feel confident using it (I never know what in the scene to use to measure the exposure from). What would you suggest in the landscape photo?<br>

<br>

I should of used some fill in flash on the second photo, but I don't really like people to be aware that I am photographing them, I guess I just need to be more confident.<br>

<br><hr><br>

<b>Thanks again everybody.</b>

[jtk]

Two very basic strategies for use of meters in cameras or reflected reading with a hand held meter... unfortunately they both require thinking and learning:

 

1) If you have a typical caucasian skintone, hold your hand in light like your main subject (eg the light on someone you'll photograph). Meter the back of your hand. Then add an additional half stop of exposure because your hand is a half stop (or maybe a full stop) brighter than middle grey. This is an alternative to running up to your subject and metering her directly. D'oh.

 

2) Learn what middle density looks like, find it in the image, meter that.

[jbq]

You seem to be getting reasonably accurate colors and a full dynamic range when scanning, which is actually very good. The alternative (having the scanner software clip highlights and shadows, along with inaccurate colors) isn't pretty.

[digitalirony]

Thanks John,<br>

<br>

<blockquote>"Learn what middle density looks like, find it in the image, meter that"</blockquote>is exactly what I am hoping to learn. I'll have to buy or borrow a book on it, because to date everything I have read online hasn't helped me. That isn't to say that the information is poor, just that I have not yet managed to comprehend it.<br>

<br><hr><br>

Thanks, John-Baptiste,<br>

<br>

VueScan defaults to a 0.1% cliping value and based on what I read and was advised I changed this to 0%. I am very pleased with the results of the scans once they have been corrected, I was just worried that there was something that I was missing that would have produced a better result at the scanning stage.<br>

<br>

Now that I know that the results are what is to be expected, I will just get on with scanning my negatives and not worry about the process any more.<br>

<br>

I will eventually start scanning at 48bits (16 bits per channel) which will make the effects of the level and curve adjustments less noticable (although I tried it on one frame that I scanned twice, at 24 bits and also 48 bits and to be honest I didn't notice a difference), but for now my hardware limits the practicability of doing so.<br>

<br>

With regard to the accuracy of the colours, I have tried to stay away from making any colour shifts in processing as I don't have any hardware colour calibration tools and I put more faith in the scanner than my eyes and monitor.<br>

[emre]

There lies the mistake: you should leave the clipping value at 0.1%, which is a very conservative figure anyway. It is okay to have a few pixels at the extremes! In exchange, you will get less banding once you have performed color correction.

[digitalirony]

Thanks again Emre,<br>

<br>

I'll give it a try. It is hard to know what is best sometimes as there are many conflicting opinions out there.<br>

<br><hr><br>

Hi John,<br>

<br>

I found a <a hreff="http://www.cambridgeincolour.com/tutorials/camera-metering.htm" target="_blank">website about metering</a> that is a bit more helpful than most (at least for me) regarding metering and identifying middle grey.<br>

<br>With a bit of luck and some practice I should eventually learn how to determine what to choose as the subject for partial metering.<br>

<br><hr><br>

Cheers Everyone,<br>

<br>

Jon

[steven_clark]

Another thing I find often helps is setting a lower brightness when scanning negs, something like .65 or so. This gives more contrast so you get a better idea of what the image might actually look like. Of course I'm saving raw as well as processed TIFF so I can actually go back and fix it if the software processing settings were too agressive.

[ellis_vener_photography]

First try dragging the white and black end points to just barely beyond the ends of the

"mountain range' in the prescan histogram in the stage, then adjust the midpoint until the

overall brightness looks right. This should solve most of your problems. If Vuescan won't

let yo udo this consider other scanning software like SilverFast Ai6

[roger_smith4]

In Vuescan, rather than dragging the white point slider, because if you do this for one scan, the setting will apply to ALL the scans you do in a batch, in the color tab set white point % to .01 or .1 and see the difference this makes. If there aren't any highlights you're worried about losing, you can even set this higher (.3, .5) which will give you final scans that require less tweaking.

I was setting white point to 0.0% but it took too much time to fix in Photoshop compared to clipping at .1%. Black point is fine at 0%.

[digitalirony]

Thanks Steven, Ellis & Roger,

 

The wealth of knowledge I've gained in one day is amazing. I will try all of the tips everyone has been so kind in sharing with me and once I have formed an informed opinion of my own, I will post a follow up, complete with examples (although this may be a few weeks away).

 

Many thanks,

 

Jon

[jsbc]

Depending on the Vuescan version and your editing program, the color space could be off - you can test the different gamut.