what ICE does to your (my) image

[Pherdinand]

I was wondering since a while, why lately my (even the 6x9 color slide!) images look weird in the detail of the colors, when i scan them on the epson v700.
So today i did some tests: different resolutions, different color spaces, different color control settings...(all in the epson software).
At the very end i also checked the difference between ICE (on Quality)and no-ICE. And then it hit me.

Look at the details in the grassy background or in the lips compared to the second (ice off) scan.

(scan at 1600dpi, 16bit, minimal levels adjustment, fuji sensia400)

Anybody else noticed this?

[mabelsound]

I have never used ICE...I just heal out dust and scratches in Photoshop. This way I get intimately familiar with the image and don't have to worry about any possible degredation, whether it's actually there or not. And it is not really all that time consuming.

[aad]

ICE is ok if you want to scan something not too critical and you think it's really full of dirt, but I think it's useless. And yes, it does remove detail. Plus it adds a lot more time to a scan.

[chrishayton]

yeh I found the same. My scans are razor sharp without ice (coolscan) and mushy with ICE on so I just heal any dust

[bmattock]

I never noticed a color change, but as the others have said, ICE can cause images to be less crisp. I generally do not use it. Notable exception - on very damaged negatives, such as the 30 year old 110 negs I am currently scanning. There are just too many scratches to deal with manually. I am willing to trade some slight sharpness for the magic of good automated scratch removal.

[Pherdinand]

I do it manually about 50-60% of the times, just because 50-60% of my shots are on trad. black and white film where ICE really doesnt work.
It can really get boring after 2-3 rolls scanned and manually "healed".

But i didn't know that the ICE messes up color film too. At least it's not me and not my scanner, then

[sevo]

ICE can take off fine details - but not as much as that on a scan at 1600dpi. I've had similar things happen, but they always could be traced to processing problems - there indeed was structure (dirt, unfixed silver or whatever) visible in the IR scan.

[gdi]

There are different "flavors" of Ice - some better than others (I think it works closely with the hardware, so that may be a factor). I use a Coolscan 9000 and ICE scans usually don't show much softening (at least compared to my old CS V).

[morback]

Any automatic dust & scratch removal will "deform" your picture. All it does is average out high contrast areas. It looks either blurred or smudged. I would not use any of those ICE and other gimmicks.

Provided you have the right tools (Wacom), clone stamp at 10px maximum is the way to go. It'll keep grain structure and if you keep picking new source spots in between each stroke you will never see a cleaning pattern.

If you like the image you will not mind the extra work. And with practice you get better & faster.

[gdi]

Originally Posted by morback

Any automatic dust & scratch removal will "deform" your picture. All it does is average out high contrast areas. It looks either blurred or smudged. I would not use any of those ICE and other gimmicks.

Provided you have the right tools (Wacom), clone stamp at 10px maximum is the way to go. It'll keep grain structure and if you keep picking new source spots in between each stroke you will never see a cleaning pattern.

If you like the image you will not mind the extra work. And with practice you get better & faster.

I don't think it that simple - Ice doesn't work on high contrast areas only, but on a special defect channel captured by the scanner using a special IR source (at least Nikon uses the IR method according to what I have read).

Theoretically, it could repair those defects without softening the rest of the image. I know may scans end up just as sharp with Ice than without , though I sometimes get a funky "correction" from the software.

I like it.

[Pherdinand]

ice is based on a second scan in infrared. It's not the usual "dust and scratch removal" blurring tool.
Normally it works well without destroying sharpness, but i just noticed that it makes funky color patches in dark areas. Somehow the magenta channel gets amplified.

[hanskerensky]

I have the following sympton with my Nikon Coolscan LS-40 ED (IV) :

When i turn ICE off to get the sharpest possible image i noticed that the Nikon exaggerated the corn-structure in the Fuji Velvia ISO50 slidefilm.
You can see that effect here clearly in the sky :
http://www.flickr.com/photos/29504544@N08/3689267963/

The image looked much better with ICE turned on :
http://www.flickr.com/photos/29504544@N08/3689243327/

[aldobonnard]

ICE on Minolta DImage 5400 (1st version ONLY !!!) works brilliantly without noticeable loss of sharpness. Makes the scans longer but anyway I'm using 4 times multi-exposure which is already long.
Even works on Kodachrome with the last version of ICE despite what the notice says. Not recommended, but works :-)

[pmu]

Originally Posted by hanskerensky

I have the following sympton with my Nikon Coolscan LS-40 ED (IV) :

When i turn ICE off to get the sharpest possible image i noticed that the Nikon exaggerated the corn-structure in the Fuji Velvia ISO50 slidefilm.
You can see that effect here clearly in the sky :
http://www.flickr.com/photos/29504544@N08/3689267963/

The image looked much better with ICE turned on :
http://www.flickr.com/photos/29504544@N08/3689243327/

The problem with ICE is in the Nikon software. Nikon's ICE does not work nearly as good as Vuescan. I am scanning all my color films with vuescan and ICE and no issues whatsoever even when pixel peeping.

EDIT: I am guessing that also with Epson scanners things would be better if used Vuescan...

[kully]

The V700 is a special case - try again with ICE set to 'speed' rather than 'quality' - the scan will be better.

[dexdog]

Originally Posted by pmu

The problem with ICE is in the Nikon software. Nikon's ICE does not work nearly as good as Vuescan. I am scanning all my color films with vuescan and ICE and no issues whatsoever even when pixel peeping.

EDIT: I am guessing that also with Epson scanners things would be better if used Vuescan...

I have had the same experience on my Coolscan V. The scans made with Vuescan don't have the smudged look that the Nikon software produces- of course you have to pixel-peep to see the difference, but it is there

[hanskerensky]

Originally Posted by pmu

The problem with ICE is in the Nikon software. Nikon's ICE does not work nearly as good as Vuescan. I am scanning all my color films with vuescan and ICE and no issues whatsoever even when pixel peeping.

Thanks PMU,

Good point !

i will give Vuescan a try with the same photo as mentioned in my earlier posting.

[sojournerphoto]

Originally Posted by bmattock

I never noticed a color change, but as the others have said, ICE can cause images to be less crisp. I generally do not use it. Notable exception - on very damaged negatives, such as the 30 year old 110 negs I am currently scanning. There are just too many scratches to deal with manually. I am willing to trade some slight sharpness for the magic of good automated scratch removal.

Yep, on a really bad neg ice makes it useable. Most of my scanning is black and white anyway, but funnily enough the colour negs are always much worse for dust and scratches. Must be the labs...

Mike

[not_in_good_order]

I don't see any degradation from using ICE on my Coolscan 5000 with c-41 or e-6 film. When I first bought the scanner, I did a series of tests where I scanned an image with and without ICE and compared the results when viewed at 100%.

Kodachrome is a completely different story. Linear objects in an image, like say overhead electrical wires, will look broken up, for lack of a better term, when ICE is used. At normal viewing magnification, it usually isn't too noticeable, but it is noticeable enough for me to never want to use ICE with Kodachrome.

The only time I have used ICE with Kodachrome is when my girlfriend's mom gave me some of her old slides to scan and they were extremely scratched from being loose in a box for 25 to 30 years and she only needed small images for the web so any edge detail degradation wouldn't matter.

[amateriat]

Originally Posted by gdi

There are different "flavors" of Ice - some better than others (I think it works closely with the hardware, so that may be a factor). I use a Coolscan 9000 and ICE scans usually don't show much softening (at least compared to my old CS V).

Agreed. I've used ICE to good effect on my Minolta DS 5400 (I). Periodic a/b testing with several film types (including Portra and Fuji Pro 400/800) revealed no discernible artifacts in my experience. Then, too, my negs and slides are usually pretty clean, so ICE, ironically, probably isn't working all that hard from the get-go.


- Barrett

[Pherdinand]

Originally Posted by kully

The V700 is a special case - try again with ICE set to 'speed' rather than 'quality' - the scan will be better.

Are you serious??
I'll have to try this.

[delft]

Originally Posted by hanskerensky

I have the following sympton with my Nikon Coolscan LS-40 ED (IV) :

When i turn ICE off to get the sharpest possible image i noticed that the Nikon exaggerated the corn-structure in the Fuji Velvia ISO50 slidefilm.
You can see that effect here clearly in the sky :
http://www.flickr.com/photos/29504544@N08/3689267963/

The image looked much better with ICE turned on :
http://www.flickr.com/photos/29504544@N08/3689243327/

I've just got a déja lu: Maybe this explains the corn-structure:
http://www.luminous-landscape.com/re...i-pepper.shtml

greetings,

Dirk

[hlockwood]

Originally Posted by amateriat

Agreed. I've used ICE to good effect on my Minolta DS 5400 (I). Periodic a/b testing with several film types (including Portra and Fuji Pro 400/800) revealed no discernible artifacts in my experience. Then, too, my negs and slides are usually pretty clean, so ICE, ironically, probably isn't working all that hard from the get-go.


- Barrett

I use ICE, within NikonScan 4.02, routinely on Ilford XP2. This thread prompted me to do a comparison with and without ICE. I could find no difference whatsoever between the two scans at views up to 100%. These negs are also very clean and don't really need ICE. So if there is selective treatment only on the area of actual spots, then this test is not definitive. But I was unaware of such selectivity. Can someone knowledgeable confirm this selectivity?

Harry