Silverfast 6.5.x
Filed in Scanning - March 11, 2007I’ve just updated my copy of Silverfast for my Nikon 8000 scanner. The upgrade to 6.5 was a paid for upgrade. The incentive was the introduction of multi-exposure scanning.
I’m mostly scanning 35m mono negs at the moment, so a key part of what I want from a scanner is to extract as much of the information that is on the neg as possible.
When I did comparative tests a couple of years ago, Silverfast did the job better than Nikon Scan by some margin. This post isn’t about that comparison, however, but about whether, if you are already using Silverfast, you might want the upgrade.
Once you get past optical resolution, and mechanical accuracy, I’ve found noise and shadows to be the two biggest areas to concentrate on (and, of course, noise and shadows interact). Over time, I’ve used fewer and fewer of the editing features of Silverfast (such as dust removal, noise reduction, or complex curves) preferring instead to let Silverfast do what it is best at – which is drive the scanner. Typically, I apply the bundled film-specific curve, add a gentle shadow opening curve of my own, switch everything else off and scan into my ‘colour’ space of the moment (which is currently the Gray Lab space which comes with QTR – I know some people recommend scanning into RGB space and then using only one of the channels. I discarded this methodology when I started using Silverfast. However, time and knowledge moves on, and I’m thinking of revisiting this.)
The piece of functionality that I do rely on though is the multi-sampling. Usually I run this at 4x. The point of multi-sampling is that each scan moment is repeated and the software compares readings. Differences in readings indicate noise, and some form of averaging or selection goes on. The resulting scans are very low noise (which is different from ungrainy. Grain aliasing is still a problem).
Multi-exposure is aimed at a different problem. Multi exposure (2x or 4x) takes complete rescans of the negative at different exposures to see if there is detail in the shadows that the normal exposure scan didn’t pick up. This isn’t applying a curve for the shadows like I do anyway, but actually running the scanner at a different exposure setting (as I understand it). The idea is that negatives have a bigger dynamic range than scanners, and that hitherto scanners have been set to optimise the highlights.
After a few of the usual Silverfast grr moments (no help files, obscure icons etc), I ran a number of tests. Multi-exposure scanning unfortunately requires multi-sampling to be turned off (or at least it does on my scanner where the film is being moved around inside the machine). I suppose the thought of mechanically aligning the scanner with enough accuracy for both a multi-sample and a multi-scan pass was too much.
I found that the 4x multi-exposure scans were unusable. Not marginally either. This must be a bug.
The 2x multi-exposure scans were good. In the negatives that I tested with, there was some slight upside. With a negative with a U shaped histogram I could imagine that the improvement would be more dramatic. There was a big downside, however, and that was any area that was fully black became very very noisy.
I tried switching the in-built noise reduction filter on, knowing that I wouldn’t want to use this, but wanting to see the results. This reduced the noise in the black areas at the cost of removing detail throughout the entire negative. Even then, it didn’t completely remove the noise. (The in-built noise filter works like USM in reverse, and is a software filter, not a change to the way that the scanner works. I don’t normally use it, finding it clumsy in comparison to Noise Ninja post scan).
There is no doubt that the noisy areas would have to be repaired manually (they are black, after all, so painting black in wouldn’t be so bad if absolutely necessary). I did try doing two scans – one with multi-sampling and one with multi-exposure – with the idea of merging them in Photoshop and picking the best bits from each. However, the registration wasn’t perfect.
I’m not particularly pleased with the results. This strikes me as an underdeveloped product update. It has also reawakened my general displeasure with the user interface. Results from other scanners may be different. I would recommend using one of the forums to find somebody with an identical set up (and media to scan) to yours before you pay for an upgrade.
Good to know!
Thank you very much for being multi-exposure beta-tester ;)
But, do I understand correctly, that SF is still not-Intel (Rosetta) version?
March 11, 2007 @ 4:09 pm
Rudolf,
Can’t answer the Intel question, sorry. I’m using a G5 for photo processing.
March 11, 2007 @ 4:32 pm
Colin,
I have the Nikon 8000 for my 6×6/6×7 negs, Tri-X from the old days, now Kodak Color Neg, still mono prints:- ) Anyhow, the thing I found with the N-8000 scanner is the compressing of the shadows (folks think it about 30), which was resolved with the Nikon 9000 scanner. Nikon did not provide a download to fix the 8000 firmware, and I am too pressed to sell the 8000 and buy the 9000. The question; has SilverScan recovered the Nikon 8000 shadow compression?
You can open up the images to a full tonal range with PS levels layer, but even with the 14 bit capture, the resulting historgram looks toothy. grrrrrr
Best regards,
Doug
March 11, 2007 @ 7:21 pm
Doug,
I can’t definitively answer your question.
When I switched from Nikon Scan to Silverfast, the shadows got better, but I’ve always felt that there was a little more in there to be got out. This Silverfast update sounded like it was going to address that, but (with further experience this afternoon) it is terribly buggy and not at all well explained (meaning, if I’m accidentally doing something wrong I’d never find out from anything Lasersoft / Silverfast has written).
However, the 8000 is the best scanner that I’ve had, so I’ve not much to compare it with. Scans from the 8000 were a revelation when I first started using it. Also, I’ve tended to concentrate on the highlights and top quarter tones when evaluating scans because that generally is the area which interests me. I’m often using the darker tones as backdrop and use Lightzone or Photoshop to drop detail in those areas, so I might not notice what was missing. It is only recently that I’ve begun to wonder what should be there.
My experience of looking at histograms within the Silverfast interface is that usually any free space in the histogram is at the bottom end. This might suggest shadow compression. However, I’ve had plenty of examples where the histogram touches both ends, so that’s not definitive either.
I’ve very little experience of scanning colour neg. All the comments above related to mono – usually HP5+
March 11, 2007 @ 8:34 pm
Double exposure to see through shadows had been offered by VueScan for years.
The thing is, it was never meant to be used with negatives, where shadows are almost transparent, requiring no exposure boost in the scanner, that’s why it is called a negative. With negative film, it is highlights that could get some improvements from a second, boosted, exposure. But only for grossly over-developed ones. I am not familiar with the way it is implemented in SilverFast, but no properly developed negative should be a problem for any film scanner on the dynamic range side.
Slides, of course, are an entirely different matter.
March 11, 2007 @ 10:33 pm
Stephane,
I’ve never been a VueScan user. When I tried it a couple of months back it wouldn’t drive my Epson scanner at all. I sort of lost heart.
The Silverfast publicity for multi-exposure is indeed about slides, however, the implementation (I checked with support) includes negatives and grayscale.
As the multiple scans happen you get to see the thumbnail scan preview (like a progress bar if you’ve not used Silverfast). The first exposure looks much like any other scan going through. Subsequent exposures only show detail in the deepest shadow areas, so yes, the exposures are geared to negative film.
So, Silverfast is doing what it claims to do. Scanning for normal and then scanning again for shadows at whatever the opposite of “boost” is.
That’s why it is picking up noise in the fully black areas. There is no signal there at all (or rather just film base).
Whether it should be needed or not is irrelevant as it doesn’t work :-)
I was thinking of scanning as a positive and then inverting in PS to see if that made any difference.
March 11, 2007 @ 11:36 pm
Silverfast seems to have the problem that it is essentially feature complete, and has been for quite some time. Most, if not all updates are gimmicky and unintegrated. For example, the “Studio” update introduced some improvements but actually as side-effects, probably unintentional (e.g simultaneous view of input and output histograms). The white balance slider was useful, but since it is not calibrated in any way, unfinished. They just pile on more and more half-baked ideas, desperately looking for revenue. For the multi-exposure, for example, surely this is sort of “HDR-lite” (not that Silverfast HDR has ever been less than opaque).
I wouldn’t use any else to drive my scanner, but I’m afraid the company ran out of ideas a long, long time ago, has never managed to produce a good UI, and now, probably, its too late.
Actually, thinking about it, if I was in Silverfast product management, I’d be working on a totally redesigned, definitive B&W scanning system. This must be the only growth area left in film scanning.
March 12, 2007 @ 9:33 am
David,
What an excellent summary. You may find this being quoted…..too good to be left in a comment.
March 12, 2007 @ 10:41 am
Hi, I have been monitoring this post with great interest for a couple of days. There are quite a few aspects mentioned that we are looking into and I thought as productmanager of the discussed software I’d leave an official comment to let you know that we are listening and appreciate the mostly fair criticism. At first I am very confident that we won’t run out of ideas for innovations like Multiexposure for a long long time and it is positively not too late for us to improve those issues that need improvement whether it’s in terms of functionality or design. In regards to our UI there have always been the folks that like it very much and the folks that don’t. Overall it’s not that bad since even the people that don’t like it, would never run their scanner with anything else. I don’t agree with the Multiexposure and “HDR-lite” comparison because HDR is a general term for “high dynamic range”. The desired effect is the same, a high dynamic range digital image tone mapped together by multiple (bracketed) exposures of an original whether it is in a camera or a scanner. Our feature increases the effective dynamic range of old and new scanner hardware and there is little half baked about that. There is a test report available here which was initiated to evaluate the dynamic ranges achieved by using this feature and to compare the result to a single pass scan. To determine the dynamic ranges the method described in ISO 21550 was used.
Something I definitly see in the future for SilverFast is what you described as a designated Black&White scanning system. That is a potential market along with a couple more areas. It would be too much stating that we are working on this but I can say that we are doing our homework to come up with something very much alike what you suggested (no definite release date yet). But enough “auguries” for now. Let me just thank you for this informative post and the comments, the feedback and experiences shared as they help us improve SilverFast. All doors, mailboxes and ears are open for ideas and concepts you would like to see realized in SilverFast.
March 14, 2007 @ 2:06 pm
Jan-Willem,
Thanks for contributing. I’ll pull together a few ideas and write some more.
March 14, 2007 @ 3:48 pm
@Rudolf
At the moment SilverFast runs in not-Intel. We are working under the hood of SilverFast to make it a Universal Binary. Our roots are in Mac development so naturally we will bring a native SilverFast to the Mac. From a technical point of view the user won’t recognize much of a difference because a native UB SilverFast won’t make the scanner run faster. So don’t expect much of an performance increase in 90% of your scanning due to the fact that the hardware will still be the same.
BTW we will stille be compatible with native Photoshop CS3 when it comes out at the end of march because the wonderful PhotoshopSDK provides a project which allows us to do that.
March 14, 2007 @ 5:55 pm
Just a few comments:
- I don’t see any problem with running Silverfast under Rosetta… UB is not a priority
- My point about multi-exposure and HDR (and it should have been clear I was referring to “Silverfast HDR”, the product) is that the end result is essentiallly the same. In theory. And any number of ISO standard reports won’t help if the feature is as buggy as Colin says it is.
- I don’t want to be negative about Silverfast. I even tried hard to write a positive review about Silverfast DCPro/VLT, which, lets face it, was not easy. I rely on Silverfast. But I’m afraid I still see complacency and a clear reluctance to REALLY listen to reviewers (professional ones included) and customers. The criticisms here are not new. They could have been applied to versions at least 5 years old. So where’s the progress ?
March 15, 2007 @ 2:20 pm
I am just trying the new multi exposure on my nikon 9000. It’s a shame when I turn multi exposure on it won’t allow multi sampling. After a couple of scans, I really don’t see a huge improvement that is worth paying the upgrade for. The example they show on the website looks impressive. If I could achieve that I would gladly pay.
June 6, 2007 @ 2:37 pm
Andrew,
I agree. It seems like this was a case of needing to add a feature to generate sales than any well worked and needed functionality.
There is probably a specific set of circumstances for which this is useful, but for the average user, no.
June 6, 2007 @ 3:28 pm
Yes, I would have to agree Colin. I don’t see any major improvements to warrant paying for the upgrade. I really think it should have been free to existing owners.
I am finding using HDR and multisampling at 8x or more provides me with a cleaner, crisper and better scan than the multiexposure feature. From this I can manipulate the shadows better in HDR and PS after. I only scan my best images so I don’t care how long it takes.
One thing I find good for Silverfast is NegaFix for scanning negatives. Colin do you know if the profiles for NegaFix have been added/updated/improved in 6.5?
Cheers
June 8, 2007 @ 3:10 am
One more thing, what is to stop you doing two or three scans, for shadows, midtones and highlights, then combining in PS after? Does this work?
I have never tried it, but wonder if any does and had success.
June 8, 2007 @ 3:18 am
Andrew,
The answer to your last question is registration. In my scanner (Nikon 8000), and I suspect any scanner where the negative moves, you don’t get perfect registration when trying to superimpose two scans on top of each other.
As to NegaFix, I’ve been meaning to go back to the site to find out the answer to that. The Delta and HP5 profiles seem much closer together than they used to be, but I haven’t checked this out in earnest, and I could be being fooled by one of those sticky settings issues that SilverFast is prone to.
June 8, 2007 @ 8:14 am
Hi!
I just read your thoughts about wet scanning and I think you are right about the hassle!
Anyhow I have plans to do an article for my web site to reflect around wet scanning.
There will also be a D.I.Y wet scanning equipment part.
I made some initial experiments during my Canon 8800F review, if you are interested to read it just go here:http://www.stockholmviews.com/canon_8800f_review/8800fpage1.html
regards/ Stefan
December 23, 2007 @ 5:49 pm
As said in another thread, I do like the multi exposure feature and it works fine for me (LS 8000). I suggest retesting it with an up-to-date version of Silverfast.
The early multi exposure versions may have had some bugs or may have been underdone.
Silverfast has won an EDP for the multi exposure last year, that’s why I took notice about it. I think, it’s a fully developed tool.
August 21, 2009 @ 11:19 am