Minolta 5400 excessive noise with Silverfast AI

[robbrown99]

I've found yet another problem with Silverfast AI and the Minolta 5400. I've just started to scan some slides with dark areas (some night shots and one bright scene with a black pillar in the foreground). On inspection of the dark areas I notice that there is excessive noise (banding running in the direction of the scan - that is longest side of the frame). I then thought that maybe this was due to not using the multi-sample scan option, so I turned this on and ran a few experiments, the results of which are very puzzling and alarming for this software (all tests at max res 7200x4800, 48->24bit, ICE, autofocus, no curves edited):

1. 16x multisample, autobrightness on. WORST noise
2. 4x multisample, autobrightness on. Slightly less noise
3. 2x multisample, autobrightness off LEAST noise, but still VERY evident
4. 2x multisample, autobrightness on. Less noise than 2, but more than 3.

I then tried the following test:

5. Scan at 7200x4800, 4x multisample, autobrightness off, ICE in the Minolta software that came with the scanner: RESULT: NO NOISE EVIDENT WHATSOEVER!!!!!!!!!!!!!!!

So, can someone from Silverfast please explain to me why their supposedly BETTER software actaully performs worse than the original and free software? Can someone investigate this ASAP as I am a little distressed that 'professional' scanning software is producing images with so much electronically induced noise banding that they are totally unusable for my clients?

There must be a problem with the way in which Silverfast AI operates the minolta 5400. I am making a big assumption, but I reckon it could be the way in which the software operates the scanning motor that drives the carrier. This must be fixed and it is very obvious to me that the scanner is not at fault since my last test (5) proved that a noise-free scan of the same tranny can be achieved with the free Minolta Software.

Yet again Silverfast fails me - somebody please sort this out ASAP!

[robbrown99]

By the way the Minolta software produces a much smoother sound when scanning - Silverfast has a more stepping, jerking sound. Not sure if it is to do with the motor control, but this makes me think it is.

All previous tests (1-4) conducted in the standalone Silverfast AI version.

New test:

6: 4x multisample with Silverfast AI, same settings as test 2:
RESULT: Noise bands in exactly the same positions as Test 2!


I seem to be spending more time testing this buggy software than I do actually scanning or taking pictures! Wouldn't really mind if it was freeware, but it isn't!

And can someone from Silverfast issue me with another ticket to add to the other one that still has not been resolved: 2005022210000329 - yes that one is two months old and counting!

Someone please fix these problems before I go insane! By the way I recommended Silverfast on a well known forum - I'm seriously thinking about rectifying that due to the rather serious functionality problems I've had!

[robbrown99]

Samples from above test:

Whole Picture:

http://www.brownphotographic.com/temp/wholepic.jpg

TEST 1: 16x multi in Silverfast

http://www.brownphotographic.com/temp/16x.jpg

TEST 2: 4x multi in Silverfast

http://www.brownphotographic.com/temp/4x.jpg

TEST 3: 2x multi in Silverfast, no autobrightness

http://www.brownphotographic.com/temp/2x.jpg

TEST 4: 2x multi in Silverfast, autobrightness

http://www.brownphotographic.com/temp/2x_auto.jpg

TEST 5: 4x multi in Minolta Software

http://www.brownphotographic.com/temp/4x_minolta.jpg


I then applied a high-pass custom filter in photoshop to more clearly define the noise. The kernel I used was:

0 -1 0
-1 7 -1
0 -1 0

I used to use something like this when analysing satellite images for noise in the sensors (moving parts in the scanning setups cause the same kind of banding). The results of the high pass filters are below:

TEST 1: 16x multi in Silverfast

http://www.brownphotographic.com/temp/16x_high_pass.jpg

TEST 2: 4x multi in Silverfast

http://www.brownphotographic.com/temp/4x_high_pass.jpg

TEST 3: 2x multi in Silverfast, no autobrightness

http://www.brownphotographic.com/temp/2x_high_pass.jpg

TEST 4: 2x multi in Silverfast, autobrightness

http://www.brownphotographic.com/temp/2 ... h_pass.jpg

TEST 5: 4x multi in Minolta Software

http://www.brownphotographic.com/temp/4 ... h_pass.jpg

Note the distinct lack of ANY noise in the minolta software scan. Hard evidence that suggests the Silverfast software driving the Minolta 5400 is causing this error. I have not in any way manipulated the source images, so this is real data. Surely other 5400 owners must see the same thing? This is not just a 'bit' of noise I'm talking about! The results above should speak for themselves.

Good luck LSI!

[munchello]

I am astonished that no one has replied to these posts. I have noticed this particular kind of noise problem when multi-sampling w SF6.x on my 5400, as well as with my old Epson flatbed and with my Microtek 4000tf, and with my old Nikon LS-2000!

Maybe this problem was addressed privately through Lasersoft Imaging Support, but it sure would be reassuring for you folks at Lasersoft to at least let us know you are working on the problem.

You have the best calibrated color accuracy according to most review/tests I have read. Time to address the noise problem.

Thanks.

[XIII]

I have also notice exactly the same problem with my 5400 and don't seem to have the same problem with v**s**n (which I primarily use for negatives and Silverfast for slides).

Any support would be appreciated.

Cheers
STEVE