Getting on with Microtek F1 (M1)

[Matus Kalisky]

Hello,

just a few days ago I received a refurbished Microtek F1 (I am in Germany) and now I slowly learn how to use it. As it came along with a SilverFast Ai Studio 6.5 (which I upgraded to 6.6) I am using this software for scanning.

I also calibrated the scanner with IT8 target (Kodak Ektachrome)

I am not ready yet to offer some valuable review (I will be more than happy to do that later) - quite opposite - I would like to ask question or two (so this thread may grow with time)

Question 1:
- I am scanning a 4x5 Kodak E100G slide that looks very reasonable to start with. The preview of the scan at default scan setting looks OK too, but the final scan (48 bit color) has a very strong magenta cast - but only in shadows! The middle tones and the high values look very reasonable!

Do you have any idea what could be the reason and how to fix this? Please see below for a example:

Whole 4x5 image

Crop 1:

Crop 2:

 

[Marco B]

If you look at 800% or so, you will probably discover that the magenta is actually a mixture of reddish and sometimes a bit of green pixels, mixed with the proper dark color pixels. Since you are scanning slide, the darkest parts will be very dense, and thus cause electronic "noise" pixels, which in my experience, generally turn up as these reddisch pixels...

Not much you can do about it, except getting a less "noisy" scanner with higher true DMax (which will be difficult as well at reasonable cost, you are quickly ending up in the Imacon / Drumscanning range).

By the way, the overall scan looks a bit to bleuish or cool in my eyes on my (calibrated) monitor, but lets wait what others with better monitors have to say.

 

[Matus Kalisky]

I did not come to think of that. Just to see what comes out, bellow is 100% crop from 2400 dpi scan. Yes, the scan is slightly blue, however such a cast would be easy to remove. I did not adjust the scan before posting.

I understood that it should be possible with this scanner to make a scan with higher exposure (and blowing out the highlights). Doing such a thing could yield better shadow quality. I could merge two scans later. Now I need how to figure out how to do that. I guess just a simple multi scan would not solve much here.

However - I have seen some scans from Guy Tal with this scanner and he did not seem to have these issues ...

Unadjusted:

The same crop with levels pushed to 2.0 to see the noise in the scan:
Seems like your theory is correct.

 

[pellicle]

Hi

feel silly suggesting this as I don't have one .. but ....

try examining something like a stouffer stepwedge in the darker scales. apply curves to the resulting scans to see if the responce of R G and B is even. Split channels and look for differing noise ... I find that with some scanners you need to roll off the lower end to avoid noise with curves thus sort of obviating the extra step or so they get in claimed Dmax

For instance I found quite horrid results in my LS-4000 which woke me up to the reality of things a little (after having been dreaming of one for ages). I carefully cleaned the optics and paths in this scanner and these are the after scans.

 

[jd callow]

Pellicle is correct, but you don't need a stouffer wedge. You only need to look at each individual channel of your current scan to see if there is more noise in one channel over another. The best place to look is in an area where there is empty shadow.

 

[Matus Kalisky]

From the images (see below) it seems the what has been proposed is correct. The fully black area (first crop) the black is black, but the tire that is just on the edge of fully black (better seen in the adjusted crop) has a magenta cast. Closer look to histogram reveal sharp peaks in low values (around 6 - 8) mostly for red and some for blue, but very little for green.

One can get some reasonable cure using the color sampling tool for shadows in the Levels window, but it takes a lot of clicking to get the right one. Of course I would like to preserve as much of the shadow detail as possible.

I would like to take a scan with different brightness of the bulb - but I do not seem to be able to find this - it should be in the SilverFast in the Options->Special but it is just not there ... I hope to have an answer from SilverFast soon.

unadjusted crop - 100% @ 2400 spi:

The same crop with strongly pushed levels:

Leves of the whole image before any adjustments - you can see the peaks:

 

[pellicle]

Hi

seems that there is quite a bit of noise in that segment. Here are the channels split and cycled through (please see attachment).

At home (but not attached) I applied a little roll off on the toe of this to the RGB and additionally a wee bit more to he green channel. I took care to straighten the green slope again to be linear after about the value of around 30, this substantially cleaned up the image

ok ... since the attached gif seems not to work, here it is from my server

 

[Matus Kalisky]

OK, so I understand that you use the levels adjustment to suppress the lowest values. This approach would give smoother black parts of the scan. But this does not really solves the problem. Adjusting each color level separately usually leads to color shifts, though I would have to try this one first.

I am still looking for the exposure (hardware) adjustment in the SilverFast. The only one I found up to now is software based and this has no effect on the noise and cast problems. Hmm ...

 

[pellicle]

Matus

you'll notice that I only adjusted the lower levels, then returned the scan to linear quickly. I find that colour shifts are less obtrusive in the shadows where blue will be too high anyway in outdoor situations ... then there is the reality that in the shadows there will be all manner of reflected colour temperatures anyway mucking it all up.

did you notice any colour shift in my returned sample? It had a white area too

 

[Matus Kalisky]

OK, I played a bit more and did the scan with 4 following settings:

- no noise reduction of any kind
- gamma of 2.0
- 1200 dpi (to make the handling of files in PS faster)

A) 48 bit, auto adjustments (color, curves, etc..)

B) HDR 48 bit (no color, levels or curves adjustments possible with HDR setting)

C) MultiExposure (ME) - with ato color, curves and leves. In this setting the scanner takes two scans with different exposure settings and merges them

D) HDR + ME (again, no curves, leves or color adjustments)

after the scans were made all scans had "global" levels adjusted to cut away empty regions (very different in the 4 cases - (A) needed basically no level adjustment) and then some color balance adjustment. With these very basic processing I was able to get rather reasonable result with (B) and (D), and not that good with (C), the worst being the (A). Indeed - one can play more and I believe that using techniques mentioned before would allow cleaner work. I will be trying them once I master a bit more the scanner itself.

From the noise point I got the best looking shadows with the mode (D). (A) displayed ugly dark areas - separated from the rest of the image.

Concerning the shadow detail the (D) yielded the best results. While the amount of shadow detail captured by (D) when compared to (A) was not better by a large margin, it was blending into detail less areas in a much more smoother way. Caveat here - I was looking in a out of focus areas. I will have a look at this issue with some other slide.

So - I consider this issue more less solved. I still hope to get the answer from SilverFast as there do not seem to be many users of the F1/M1 around.

Thank you for your help - I will keep you updated about this scanner and its capabilities.

 

[Matus Kalisky]

One more issue: - the scans have around 30 pixels wide (with 2400 dpi scan resolution) a light green strip. Always the same place and always the same color. I get with both SilverFast and with ScanWizard, so I exclude a software problem for now.

The strip is inly in the green channel - the red and blue seem to be fine ...

It can be hard to spot in a color scan, but in the BW it can be very obvious (see bellow).

To try to solve the problem I unmounted the top lid and the glass bed - to be able to access the interior of the scanner - and with a photo-brush wiped the part I though could have something to do with scanning. I also (holding the scanner on the side) used a canned air to blow the interior of the scanner a bit. But no change ...

I am contacting Microtek - hope for some reasonable answer. I have still a few days left on my return warranty ...

The strip looks like this (Bw scanned as color positive and inverted):

 

[pellicle]

that's definately an obstruction. It could be on the scan head (or optic) or it could be on the 'calibration area' (assuming that the Microtek has that)

it its warranty I would most definately send that back

 

[donbga]

I have still a few days left on my return warranty ...

That's exactly what I would do. No sense in trying to fix a problem like this.

 

[Matus Kalisky]

I should probably also make clean that apart from return warranty I also have a full warranty (from Microtek - 2 years I guess). So - the scanner goes on Monday back and will be repaired. I keep my fingers crossed ...

 

[OldBikerPete]

I have the M1 in Australia and have just noticed this discussion. Although I scan negatives rather than transparencies, I too have found that the multiple passes mode in the Microtek software goes a long way towards removing noise.
I don't like the Silverfast methods of color correction, so I don't use that software, only Microtek's and (recently) Vuescan.

 

[Matus Kalisky]

Hello Pete,

I have yet to test the Microtek software in detail as I used it only briefly. I run SilverFast right now. I found that useing the SF with the "Multi exposure" (it has symbol of small sun with number 2 in bottom right corner when activated does a very good work on the shadow noise and detail. I did not try the multi scan (up to 16 scans) yet as the would be way too slow.

Could you maybe mention a few details how do you set up the Microtek software (say for positive scans)? It would be great to have a starting point.

Also - do you use the AF of you have the focus fixed. Up to now I have used only AF, but it seems to work fine. Even the scans of 4x5 slides where the film was not lying flat in the holder did not have any traces of out-of-focus parts.

Anyhow - I will be (most probably) sending the scanner for a repair on Monday (because of the band problem). I actually wrote to Microtek and got the following answer:

... it could be dust at a mirror inside the carriage. You remove the Cover from
Carriage and look for the dust.
But be careful, it is very easy to scratch mirrors. For cleaning, you can
remove Carriage cover and Lamp-Holder, but no other parts!
Otherwise, you can send scanner for repair, but dust is no warranty. ...

The problem is I do not exactly understand which parts he had in mind. I did remove the the to lid, the top scanning glass and the whole top plastic cover before to fix a small problem with the On/Off button, but I do not really know which part are the "Carriage cover" and the "Lamp-Holder". Would you maybe have a better idea?

thanks

 

[OldBikerPete]

Hello Pete,

<snip>.....

Could you maybe mention a few details how do you set up the Microtek software (say for positive scans)? It would be great to have a starting point.

Also - do you use the AF of you have the focus fixed. Up to now I have used only AF, but it seems to work fine. Even the scans of 4x5 slides where the film was not lying flat in the holder did not have any traces of out-of-focus parts.

<snip>....

I have only ever scanned a few 35mm slides for a friend on the M1. I just set Scanwizard on automatic everything and they seemed to turn out OK.
I mostly scan 5x4 negatives (Portra 160VC) at 48bits per pixel and for them I sent the densities manually. I go into the densities dialog and individually set the max and min densities for each of the R, G and B channels to the respective Dmax and Dmin specified for the film. I uncheck the 'adjust color' boxes. This produces a washed-out looking scan as only the most horribly contrasty light conditions will use all of the range of the film. Next, in Photoshop, I add a 'levels' adjustment layer and click the 'auto levels' button and what results is fairly close to an ideal image. One then adds adjustment layers for curves, brightness/contrast, color balance and saturation to bring the scan up to perfect.
Focussing is always left at automatic and even if the M1 had spot and scratch removal, I would turn it off as I have found that I much prefer Photoshop's healing brush.

The problem is I do not exactly understand which parts he had in mind. I did remove the the to lid, the top scanning glass and the whole top plastic cover before to fix a small problem with the On/Off button, but I do not really know which part are the "Carriage cover" and the "Lamp-Holder". Would you maybe have a better idea?

thanks

I've never even had the glass off, so I can be of no help there. Sorry!

Pete.

 

[Matus Kalisky]

Just a small update - I have sent the scanner for a repair - still waiting.

Still - just before I have sent the scanner out I was scanning a 4x5 slide (E100G) and when using the ME (Multi Exposure) mode (which does help to cool down the shadows) I realized that there is a problem in there:

If I do not use the ME feature (Just making a "normal" single pass scan) I get:

But turning the ME feature on makes strange artifacts around high contrast places:

Both above are 100% crops from a 2400 dpi scan:

I contacted SilverFast and will continue on this issue once the scanner is back. They do recognize this as a problem, which is good, but need more input which I can not provide right now.

As I wanted to make a print, I used the single pass scan, you can see the final (adjusted) picture HERE. I printed the image to 50x70cm and looks very nice (with some shadow noise I hope to improve)

 

[OldBikerPete]

I think you will find that those artifacts are caused by the software clipping at the lowest density parts of the transparency and then alternativley clipping or not clipping on the multiple passes. If you look at the histograms and the place at which the software has chosen to set the white point. I expect that you will find that the software has chosen a white point which will cause some small proportion of your highlights to be 'whiter than white' and the software has internally 'rolled over' to black - sometimes - at those places.
If I have not made this clear, please say so and I will try to elucidate further.

 

[Matus Kalisky]

After longer time I am back to scanning again - and that means new problems appear

Color cast after calibration:

I have just calibrated the scanner with the calibration slide (Kodak Ektachrome) under SilverFast. The procedure is automated and quite fast. However the results are less than convincing. After the calibration when I try to scan a positive Slide (Velvia 50) I get very strong magenta casts in the shadows. The only way to remedy that is to make some substantial adjustments on the blue curves during the preview of the scan. I will post samples tomorrow.

So I am wondering whether other M1/F1 users have experienced something similar. When (after calibration) I scan the calibration slide I do not really see the same, but indeed the neutral grey part of the calibration has green-blue cast which is way beyond my current miss-calibration of my monitor (new Spyder 3 Pro should arrive tomorrow, my old Eye-One 2 died recently)

Fast Update:
I have just checked the 4 scans of the calibration file I have obtained AFTER 4 different calibration runs - only the first one has correct colors. The rest has a cast - I guess I have to investigate this more. Maybe it is the Automatic Color & Levels button (looks like a picture of lens-aperture) that screws things up - I thought that it should only apply reasonable black/white point setting and apply the calibrated color profile. Maybe it is trying to do more than necessary.... Hmmmm

 

[Matus Kalisky]

OK, I git it.

Clicking on the "aperture symbol" (it has "C" inside as the scanner was calibrated) is asking the scanner to be smart. Not a good idea. I found out that after calibrating a straight scan does the best job (if the film was exposed reasonably well). So now my scans look reasonable.