Scanner calibrated for slides, improvement?

[Usagi]

Hi,

I have used faust's IT8-targets for calibrating scanners now for about 3 years. But for my latest scanner I was a bit lazy and it took months until I went throught the calibration process.

What I found was that actually calibration is more or less nonsense.

My latest test attempts with Epson V700 and Minolta Scan Dual IV has shown that there's not much difference between calibrated scanning workflow and by using automatic options.. Both way you end up to the situation where you still need a lot of work in photoshop until the scan looks somewhat like the original color slide.

I am seriously wondering that why? Should the suppose of IT8 targets is to give easy way to get subject scanned as well as it is in the reality?

Any examples of scans without profile and with profile?
I really would like to see them and hear the workflow how the calibration is done to get reasonable results.


When I look now the tons of unmodified master scans of slides I had scanned during years with profiles, I realize that many of them has a bad magenta color cast. Especially Provia 100f and Velvia 100f scans.

It is often said that the best way for calibrating is to scan target as an RAW scan without any modifications for example using Vuescan.
Then after profile is created, do all scans as RAW and then apply profile to them afterwards.
Yes, that gives reasonable results, but the dark tones tends to 'tear' when scanned to RAW file and applying profile in photoshop.

This is true with Vuescan and Minolta's scanning software.

One possible problem is that these consumer level scanners like Epson V700 and Minolta Scan Dual IV just cannot resolve densier part of the slide well enough and that causes problems with the dark tones.

Eventhough Epson claims DMAX of 4.0 for V700, in reality it doesn't go above 2.70. Everything that has 2.70 or higher density is practically black. No way to get details out from the shadows.


Here's a comparison of Provia 400f with a different scanning methods.


And here's the test with Elitechrome 200 ED-3 (one of the worst films to scan). The description is included with the first picture.


Thanks,
Jukka

 

[mrred]

The only thing I have found to make life easier is to use ColorNeg and ColorPos. http://www.c-f-systems.com/Plug-ins.html I just do a raw scan (vuescan pro or Silverfast Ai) and process this with PS. It does most things perfectly. When off, the adjustments are easy and swift inside the plugin.

I have an IT8 target and would have better spent my money on the above.

 

[pellicle]

Hi Jukka

first, I am not any sort of expert ... (so why am I commenting? Perhaps somone will after me?)

Eventhough Epson claims DMAX of 4.0 for V700, in reality it doesn't go above 2.70. Everything that has 2.70 or higher density is practically black. No way to get details out from the shadows.

did you get much more out with the other scanner? I don't really get much more out of dark areas of slides ... its one of the reasons I use negative mostly now ...

Here's a comparison of Provia 400f with a different scanning methods.

seems a little dark and without vibrant colours to really be sure what was happening to the whole range ...

And here's the test with Elitechrome 200 ED-3 (one of the worst films to scan). The description is included with the first picture.

seems like more is happening here

to my understanding the reason for target colour calibration is for applications where you are totally sure you want the colour of the scanned article to totally match the colour present in the slide. Assuming you are photographing in photo-flood illumination or other certain colour temperature light the advertising crowd tend to get all "thinggy" about the colour of the packet of cereal not matching the dye selected for printing the brochure. When the shit hits the "we're not paying" fan then if you ain't sure of your data (as a seperator or whatever) the its your fault.

 

[Peter De Smidt]

I haven't used Vuescan or Epson scan in a long time. With my scanner, A Screen Cezanne, icc profiles made with Mr. Faust's targets do a great job. I use Lprof to make the profiles. Shadow detail can be an issue. The HutchColor scanning guide has a procedure that might help. See: http://www.hutchcolor.com/PDF/Scanning_Guide.pdf

 

[Usagi]

Some additional details: I have created the profiles using lprof software. It shows that there's more or less inaccurancy in the test scans that I made using target but not much.

If I scan the faust target and apply profile, it's fine. But somehow the actual slides aren't. Some hint's about magenta colour cast (especially in Provia) I found at here.

One thing that come to my mind is grain/noise. I haven't use any kind of blur or softening with scanned targets before profiling. Perhaps this could help?


Or perhaps mrred is right and it's just waste of money to invest to the IT8 targets.

 

[Peter De Smidt]

I scan the target at 1200 spi. I then bring the scan into Photoshop, making sure that no color conversions take place. I get rid of dust spots using the cloning tool and save the file. I then run Lprof to make the profile.

It's very important that your color management is set up correctly. Make sure to have Photoshop ask permission before any conversion takes place.

Make the scan with the exact settings used to make the profile. Make sure to turn off any kind of automatic adjustment. Save the file. It should have no profile attached. Now import it into Photoshop. Photoshop will have a dialoque pop up asking how to interpret the file's color. Assign the scanner/film icc profile and convert to your working space, usually Adobe 98, Ektaspace, or ProPhoto.

Assign and convert have specific meanings, and it's very important to do the right one.

It is possible that your scanner is not consistent enough for a profile to work well. There are things you can do to help. For example, I always warm my scanner up for 30 minutes before scanning, as is recommended in my manual. It's possible that your light source or sensor is not uniform over the whole area of the scanning bed. Try to scan in the same location.

 

[rnwhalley]

Usagi, I had the same problem using VueScan and a Minolta 5400, especially noticeable with Velvia slides. I found the solution with my set up was to set the "Media" on the "Input" tab to "Image" rather than "Slide". If you then go to the "Color" tab you will find you only need to set the ICC profile for the scanner rather than setting it for both scanner and film when "Slide" was selected. This made a major difference with my scanner.