Velvia Slide Scanning Help on Dimage 5400

[foxbatkan]

So I'm scanning some photographs on IS50 Velvia slide film. The scanner is the

DiMage 5400. I've just started using slide film and some of my other stuff went

fine, but I'm having troubles getting it to look as good as the slide does!

 

The settings are colour positive, autoexpose, autofocus, 16 bit and I used to

store the raw, but push the gamma to about 2.2 and use either Ektaspace or

Ethervizion's created profile for the Velvia50 on 5400.

 

The top left is the raw scan screenshot. The bottom left is the preview of

pushing gamma to 2.2. The top right is the actual result of pushing gamma to

2.2. Looking at the slide in plain light looks much better than all three of the

above, with a less defined vignetting.

 

What am I doing wrong, i.e what term describes what I'm experiencing? It's a

sunset behind clouds, so I half expected the sky to be well exposed and the sand

a little under, but it looks fantastic on the slide and disappointing when

scanned. I understand it should be a bit flat when raw scanned, but I can't tell

what path to take to fix it.

 

Thanks much!

Kannan<div></div>

[Robin Smith]

Kannan,

 

It is difficult to see what the problem is without seeing the original to be honest. The thing to remember is that a slide and a screen image has a greater contrast range than a print, so you will have to decide exactly what is important to you - with most slides like this the highlights are usually the thing to concentrate on. You could indeed use the shadows tool or use layers in PS to equalize the contrast - but I find you can only do this a little before it begins to rob the image of the very impact that you probably liked in the original slide.

 

One point to bear in mind is that the observer does not know what the original looks like - they can see judge what you show. This can be an advantage.

 

My suggestion is to use a curves adjustment layer in PS and see what you get.

 

Scanned images do often show vignetting - you can lighten the corners slightly in PS which may help. Often a little vignetting is a good thing and attractive to the observer (taken in moderation).

[mendel_leisk]

Kannan, I'm a little confused. Are you:

 

1. Running the scanner with Minolta Scan Utility (MSU)?

 

2. Outputting what MSU describes as "16bit linear"?

 

3. Opening that image in Photoshop (or other?), starting a levels dialogue and adjusting the mid slider to raise the the brightness to normal pc viewing gamma of 2.2?

 

Assuming the above is the case:

 

1. Why not just ouput a 16 bit file, and save yourself the grief of adjusting the 16bit linear? Or conversely, try using that file as Vuescan Raw File?

 

2. Looks like the sun is blowing out. Maybe leave auto exposure on, but move the master slider in the exposure Tab down a bit?

[richard_s.]

Kannan,<br>

<br>

16bit linear files of the Dimage 5400 are raw data without balanced

white-point. Simply applying a gamma correction of 2.2 will introduce

severe and difficult to remove colour casts. <br>

<br>

I recommend to use a appropriate colour profile for colour conversion:<br>

<br>

1. Assign, embed a  fitting colour profile to the untagged

16bit linear scan, Minolta supplies the profile MLTF5400p.icc

“Minolta DiMAGE Scan Elite 5400 (Posi Linear)”. Do

not  use the “Convert to” option.<br>

<br>

You image will now look very different on the monitor, but the data of

your image has not changed! The image editing program just interprets

it in different way.<br>

Your image might look much better, but is still not internal

white-balanced. So it is better to convert it to a white-point balanced

colour space, before making any large colour or contrast adjustments.<br>

<br>

2. I recommend to convert it to a wide gamut colour-space, like

“Ekta Space PS 5, J. Holmes.icm”, to preserve the

full range of colours. <br>

<br>

3. When finished, you might convert it to a colour-space suitable for

your output device like “SRGB” for web-display.<br>

<br>

I did just that with your sample image. Of course colour-space

conversion on 8bit JPEG files delivers not the best result.<br>

<br>

You might notice, he Minolta supplied generic colour  profile

MLTF5400p.icc “Minolta DiMAGE Scan Elite 5400 (Posi

Linear)” introduces a slight magenta cast.<br>

<br>

Therefore, I have profiled my scanner for several sorts of slide film,

with better results (I hope so). I used my “Velvia

50” profile on your image without a magenta colour cast.<br><div></div>

[foxbatkan]

Robin - I've actually never seen this projected/printed anywhere, I'm just looking at the slide in plain light, and it looked wonderful - now I'm trying to reproduce it on screen/digitally to share.

 

My problem was that I don't know how to technically explain what seems wrong, which would've been the first step to fixing it, so I just uploaded screenshots.

 

===========================

 

Mendel - Yes, yes and yes. I'd assumed that 16 bit linear is a wider gamut and I'd be better off with more raw data? Is it really that big a difference?

 

===========================

 

Richard,

 

Previously, my method was: to use karl martin/ethervizion's profiles found here, he posted a link to them in a previous thread:

http://www.ethervizion.com/lost_found/

 

I'd scan 16-bit, adjust gamma to 2.2, and write the raw files to DVD. Then, I'd use the profile and convert to JPEG to share with others.

 

Unfortunately, the ethervizion profile didn't work in this case. The other confusing thing is for me, photoshop's preview of a gamma change was different from the actual result of a gamma change, which threw me off.

 

I also assumed that this problem was a result of scanner exposure, and got frustrated when it didn't do me much.

 

I've never, ever dealt with the Minolta's given profile but I'll give it a go - those last two images look just like my slide does held up to plain light, so thanks!

 

But to settle it for all, is plain old 16 bit good enough? It's what I do with film, and it's what I've done previously with slide film. I want to settle down on a workflow and minimize how much I want to torture my film and scanner, and archive things on one shot. I tried 16 bit linear mostly by accident on this run.

 

Thanks a lot, your help is invaluable and can't really be found anywhere else!

[richard_s.]

<i>But to settle it for

all, is plain old 16 bit good enough?</i><br>

<br>

Definitely, 16bit is as good as possible. Most DSLR use only 12 bit to read out the CCD, which means significant less dynamic range than the Dimage 5400.<br>

<br>

The Dimage 5400 is excellent for the scanning of difficult, dense slides with deep blacks, like Velvia or Kodachrome.

<br>

[foxbatkan]

Thanks Richard. I guess what I've learned now is that I shouldn't get lazy and subscribe to a one-colour profile-fits all mentality. I plan to autoexpose and store the raw 16 bit scans, and I guess I'll have to tinker with various profiles when ready to print/display. I'd never heard about this posilinear profile, it works like a charm fixing a lot of earlier mistakes too!

[richard_s.]

<i>I'd never heard about this posilinear profile, it works like a

charm fixing a lot of earlier mistakes too!</i><br>

<br>

It is a pity, that Minolta has not a better manual, description of the

workflow. Minolta tried to make it “idiot-proof”,

as a result you will not be informed of the true capabilities of this

scanner, but other brands are not better in this regard.