Minolta 5400 Mark 1 Vuescan vs. Silverfast vs. Dimage scan

Im wanting to know if there is a big difference between in the quality of scans im gonna get using the different available options for software and then converting using NLP. I currently use the original dimage scan software on a windows computer and scan as a positive and then transfer and convert on my mac. Figured vuescan and silverfast might be an easier option to forego the needing to transfer and dealing with old software. Will I see much of any difference using vuescan and or silverfast in combination with negative lab pro?

If you are still able to use it, the original Minolta software was perfectly good at getting a decent scan out of the Dimage.

I haven’t done a recent comparison of the two, but I went for Vuescan years ago as I needed something which worked on newer OS and it’s always been excellent.

After drying dimage (in a VM on Windows XP 32 bit), Silverfast 6 (in a VM on Windows XP 32 bit), Silverfast 8 (good in modern OS) or Vuesca (good in modern OS)…

… I discovered that - for my Scanelite Mark1 at least - the only programs managing to get a good scan are dimage and Silverfast6.

Silverfast8 doesn’t use Digital Ice, but their own IR-assisted iSRD. And it creates artifacts (never had issues with my previous scanner with it). Maybe my IR channel is different to what Silverfast expects, but I get weird lineskips and blotches, areas where the iSRD goes nuts while nothing was there. And no, I’m not scanning silver-based negs or Kodachrome.
Silverfas6 doesn’t has this issue.

Vuescan has issues with the autofocus. If you pick a spot to autofocus, it does it’s thing, but way faster than way Dimage and Silverfast are able to do.
I think it does something wrong in this, because when I tweak - painfully by trail and error - the focus number in Vuefast I manage to get it just as sharp as Dimage and Silverfast… but Vuescan’s autofocus just doesn’t pinpoint that number (while the other two work fine!!).

So, Dimage ‘scan as positive’ (and I tweak the R, G and B gain to let the filmstrip of C41 be aligned and quite high, but quite some safe margin to clipping). For positive (slides) I just scan with the gain equal and at 1.0. I have a Fuji it8 target that seems to fill that range, so I can color-calibrate it to that.

Only drawback of Dimage is that it doesn’t do ‘batch’ scanning.

In Silverfast6 - if you get around all the bugs and quirks of the very old software - you can queue up an entire filmholder and press start and go away. But - officially - it is not available anymore.

The 16bit linear tiff files from a ‘scan as positive’, I even managed to through in Vuescan and then ‘scan raw from file’. Through this trick you can scan ‘from raw to raw’. If you set your output raw as DNG, you can create neatly Vuescan-created DNG files. Through exiftool you might have to fix ‘camera maker / camera model’ for NegativeLabPro to pick up the correct profiles, but I managed to get this working, and the results were OK.

Don’t use NLP anymore, so I don’t the converting from .tif to .dng handy from head. But I still scan with Silverfast6 and my Dimage5400 :).

(I did some tests with my Sony A7m2 with Sony 50mm 2.8 macro lens with NLP vs the same negs with my Dimage 5400 converted through NLP… the Minolta won hands down (in the colors, the resolution the film is oten the limiting factor), since then I never tried to DSLR-scan again).

Ive been using dimage scan for a little bit now but I have had trouble with windows xp virtual machines. The scanner will get recognized the first time i load up the machine but after that any time i restart it will show the scanner is there, and the dimage scan software will launch, but it just hangs there indefinitely. have you had any issues with vms and dimage scan?

I use vmplayer workstation 15 player with windows XP sp2 32bit. That way the official driver still works.

In vmware you can set an USB device as ‘pass through’. So the moment I flip the switch on my Dimage 5400 and vmware is running, it shows up as a new connected USB device in the VM, not on my ‘real’ machine. Never got it working through firewire.

It’s an old beast though :). When the VM is started, I need to turn it on. I have to remove the filmholder from the device, then start a scan app (dimage / silverfast6). Then it inits the scanner, and it starts making noises and things start moving. It can take a good 30 seconds or so for the app to continue starting. Then when the app is started, you insert a filmholder. In Silverfast you’ll see that it recognizes which of the filmholders you inserted and it’ll switch to that mode. Configure it how you want and try scanning.

If you don’t use the scanner for a while, it goes to some sort of sleep mode and the app looses connection. You need to remove the filmholder, close the app, restart the app, and when it’s started insert the filmholder again.

Silverfast6 has quirks to. If I modify the gain settings (in the settings somewhere), they become active for the normal scan mode. But if you use the ‘batch mode’, the selected gain settings only get applied after quitting and restarting the app. And this requires removing and inserting the filmholder again :P.
Also enabling IR-cleaning seems to limit the speed to 5400 dpi. As in, if I disable IR-cleaning and go down in DPI, the scans go faster. But if I enable IR-cleaning, it doesn’t matter which DPI I choose, it always goes slow. As if the IR channel has to be scanned at 5400 DPI or something.

Sounds annoying - and it is - but once you know this stuff it doesn’t really matter. If I scan a roll of film I’m setting up the gain/exposure this way, and finally if I’m happy I make all the settings how I want them (DPI, Digital Ice, diffuser enabled or not, focus point I always put somewhere between the center and the outer edge it always focuses on the grain, etc…). In Silverfast I can ‘save’ these settings.

So then when I’m ready to go, you go in the batch mode, there is a little icon that does a filmholder-preview (special button in the batch window, not in the normal window) and there you add images from the filmholder to the batch. When you do, it asks which settings to apply. Pick the settings you just saved too, do not click on ‘automatic adjust stuff’, give a filename to each of the 6 images and click ‘start’ and it will do 6 at a time.

Doesn’t make the process that much faster, but I’m more free to do some other stuff around. It’s much easier during a workday for me to put in a new strip of film every 30 minutes then having to start the next frame every 5 minutes in Dimage :).

i would like to briefly chip in my experience.

A lot of you seem to mention old windows installs/ virtual machines to dimage scan(which in my opinion does a great job btw).

There seem to be two common misconceptions about this software on the web and in this forum

1.) “it doesn’t run on windows 10” I can personally attest, that it runs perfectly and that all you need to do is modify the driver for the scanner. I completed this process in <30min following this guide: Link The headline mentions a different scanner, but don’t be deterred, dimage scan elite 5400 is covered as well. How trusted sources for drivers are disabled differs slightly for windows 10, but a quick google search should settle that.

2.) it doesn’t batch scan:
You can enable autofocus on scan, enable manual exposure, set the same resolution, exposure and maximum crop for all frames and then index scan. Once index scanned you can select all frames(ctrl+a or ctrl+left click on frame) and hit scan. Works for me and delivers consistently good results on e converted with NLP. (As a tip I discovered by chance: setting the white balance to NONE in NLP seems to deliver the best results when coming from the dimage software)
I do get a lot of digital noise in my sigle pass scans, but for the time being I tame it using Lightroom, where visible. I think some experimentation wih the exposure might be in order, but I don’t have time for that right now and for batch jobs(non-critical personal archiving) the results are remarkably good.

On a related note, I only switched to this software because I am batch scanning colour negatives and the ICE isn’t supported by Vuescan(FWIW on account of licensing issues I believe). Vuescan is great for BnW and in case you are not going to use ICE.
My experiments with vuescan indicate, that the scanning time is linearly proportional to the RGB gain on the exposure which i therefor always set to 1. I think some experimentation with respect to noise would be in line here as well. There is no visible clipping of the histogram(it does sit in the lower half though), but maybe noise could be reduced by using more optimal signal strength.

Lastly, please spread the info regarding the original software. i find the idea that people are purchasing silverfast just for ICE with minolta scanners frustrating. I think the software is overpriced to the point of extortion.

That ‘Sony film scanner’ beta driver is completely new to me. Looks more legit than I imagined :).

Problem with most xp drivers is that they are 32bit only, so on 64bit os you are out or luck.

But this is a beta driver for vista/7 when x64 windows was more a thing.

Anyway, for me, the fact that it is unsigned is a deal breaker. But that it’s only because I believe a) that will not be an option somewhere in the future and more importantly b) my vm setup is working fine and just as easy for me, and has the advantage I have my software for scanning easily ‘plug and play’ between multiple machines :).

But good info! Not widely known!!

About the ‘batch’ scan, I am really going to check this out. The only reason I use silverfast6 is for the batch scanning, otherwise the Minolta utility worked fine.

But I remember that I couldn’t find out how to batch scan in it, and also reading (in the manual?) that it couldn’t.

I always scan entire rolls with exact same settings. Scan as positive, linear 16bit, no auto exposure.
Scanning c41 I always take a bit of filmstrip to crank the gain high enough so a piece of filmstrip is (safely!) shy of clipping and as balanced as it can.

For e6 film I just use what I used for scanning my profile (color calibration target). Which is actually just gain at 1.0 :).
The scanner (elite5400 at least) has crazy dmax anyway so it doesn’t really matter that much.

Going to give the Minolta scan utility another try!

just out of curiosity: why do you do the linear tif scans?
Dimage scan offers the possibility to output tif files in AdobeRGB which has the gamma 2.2 required by NLP and should spare you the time/memory consuming(36exp@5400dpi~ 10GB) conversion of the tif-scan prep.

Though deviating from Nate’s guide, my tests seem to indicate that scanning positive to adobeRGB, white-balancing of the rebate and then setting the white balance in NLP to (NONE) yields the best results out of he box. In some cases the whitebalance needs tweaking afterwards, but most of the time and particularly for images taken with a flash the colours come out perfectly natural like this. I first noticed this with a series of flash images on Portra, but Kodak Gold actually looks gold and Fuji 400h(RIP) looks just the way it should.

Please share your experience, as I am very much in the process of experimentation and any input is welcome.

long time :). (And I just noticed all the horrible typos in my earlier posts… I shouldn’t be typing on a phone).

Because 1) I want to save the scans unprocessed. Just as saving RAW files from a digital camera, I want to save the raw data from the scanner (which is unprocessed). Also for properly color-calibrating the scanner, I want to start without any color-space conversions done to the scan.

And for proper conversion of negatives the data needs to be linear and start without any gamma adjustment. All other tools doing negative inversions require it to be in linear-gamma. Why NLP sort of requires it the other way around, I don’t know, but as I said: I don’t use it anymore.

I always opened the raw linear TIFs in Vuescan. It has a ‘scan from raw’ option where you can open a linear tif file and you can process it through vuescan again. When you set the output file format as Vuescan’s ‘DNG’ format, it creates DNG files that work in NLP with the Minolta profiles that NLP shipped. With some trial and error you can also do it from the commandline with exiftool and ‘dng_validate’ from the DNG SDK.

If I would want to use the tif function from NLP, I would just run all the files through imagemagick before hand, with a -evaluate pow 0.4547069271758436944937833037 to shift the gamma to the correct AdobeRGB 2.199 way. But as I said, it still feels wrong for me to do processing in gamma-corrected space.

I actually tried the DiMage scan utility again, and indeed. You can ‘multi select’ the pictures in the index-scan section, and hitting prescan / scan will then do multiple in a pass.

I do have a feeling the shadows are noisier though, but that might just be that I didn’t use the scanner in a while or the slide-film I was trying through DiMage (Which thanks to the pandemic has been sitting for a year on the shelf waiting to be developed).

I am wondering what happens when you set the focus point to different areas for different pictures, if the utility remembers that… but I often set the autofocus point to exactly the same spot on all the scans anyway (halfway between the outer edge and the center, to hopefully get everything within focus).