Luminous Landscape Forum
Raw & Post Processing, Printing => Digital Image Processing => Topic started by: Lisa Nikodym on August 31, 2005, 12:24:33 pm
-
I use VueScan's feature that allows you to effectively build a profile for each film roll by scanning an unexposed bit of the negative first. It works *pretty* well (certainly far better than VueScan's canned film profiles, which are wretched, IMO). There's a page in the VueScan manual somewhere on how to do it.
If you don't do that, then, I agree, color correction with film scans is really frustrating. (One of the reasons I enjoy digital so much compared with film...)
Lisa
-
Ray, I'm doing an extensive amount of work on this issue, because I have a large legacy of colour negatives that I am converting to inkjet prints using the Minolta DSE 5400 - first model. I've been through everything you describe, and in fact I'm in the process of developing a write-up on a workflow that addresses colour negatives. The issues wouldn't be that dissimilar for positives, except that positives are much easier to colour balance, because there are IT8 targets for profiling positive films that don't exist for negative films.
My write-up will take some time to complete, because as your own experience indicates it is a bit involved. However, there is light at the end of the tunnel - I have produced A3 output that I truly think passes muster, and I shall soon be submitting for "peer review" to get a valued independent opinion. To anticipate some of my content, here are a few observations:
(1) There is no question in my mind that Silverfast Ai Studio is the best software option of the three you mention. If you will be doing a fair amount of this work, the outlay is worthwhile. It comes closest to giving you believable scans and has the most control points for achieving that objective. Better than Minolta DiMage Scan and VueScan by several country miles.
(2) That much said, it is quirky, with voluminous but poor quality documentation, which puts quite an onus on the user to figure it out. Their tech support people I must say are responsive where they can be. Taz Tally's Silverfast book provides value-added on the Silverfast manual, but it too leaves things to be desired. All said and done the learning curve is worthwhile if you will be doing alot of film scanning. The absolute first thing to beware of with Silverfast is that there must be ZERO blank space between the film and the film carrier for the image you are scanning, and the scan cropping frame MUST ABSOLUTELY fit within the actual image, otherwise you get garbage.
(3) Don't use the scanning software to get more than a roughly believable scan to start working with in Photoshop. None of the three software options holds a candle to Photoshop, and if you are working in 16 bit it just isn't worth the time messing around too much adjusting colour balance and luminosity in the scan software, except as indicated here. By trial and error select the film type that comes closest to believable results without further adjustments. Silverfast allows some fine-tuning of this preset. Use it to improve the preset and keep that recipe. Then for each image do a couple of basic adjustments for mid-range exposure and grey balance (using the grey "pipette"). These measures alone will provide a decent enough result to reduce the severity of further luminosity and colour balance adjustments that will be needed in Photoshop. I have not found a way to automate my way to "perfection" when working with negatives. It is an image by image process and takes time.
Silverfast does accept ICC profiles, so Jonathan's suggestion to make an ICC profile from shooting the GM Color Checker should not be dismissed out of hand - but it does require being able to replicate the film type and vintage effects and "appropriate" lighting conditions - so useful under some conditions, but not others - like everything.
(4) Don't use Silverfast for grain removal or sharpening; they vastly increase scan time and there are much better (more controllable) post-scanning solutions for those two requirements - e.g. Neat Image to clean-up the grain (auto-profiling each image and then previewing the result) and PK Sharpener Pro to recover any lost acutance. Both of these need to be used with care, which will be part of my eventual write-up.
(5) Use ICE at the scanning stage if there are black spots reflecting degradation from long-term storage. It even works with Kodachrome nothwithstanding that Minolta, for example, says it doesn't.
Hope this helps.
-
Dmerger, I forgot to mention - I don't know what size prints you are making from your Velvia scans, but unless you crop heavily or make full-frame prints much larger than A3, you are probably generating excessive file size and scanning time going for 5400 PPI on every scan.
Mark,
Now here's an interesting workflow issue. The Dimage 5400ll defaults to 1350 dpi every time one starts a new preview. I find it annoying because other software, including Vuescan, remembers the previous setting. Why does Dimage Scan have this default? I can only presume it's because most users find 1350 dpi scans sufficient, or maybe it's because I overlooked the instructions that tell me how to set 5400 as the default resolution. (Quite likely :D ).
Whatever, since I'm not in a commercial climate producing one-off jobs for a client, I see little reason to scan at less than full optical resolution. Too much resolution is rarley a disadvantage, whereas the reverse is frequently the case, or can become the case if one later wants to make larger prints.
Since SilverFast has a steep learning curve, as you mention, then perhaps I should persevere with this software. I've noticed artifacts using SilverFast's ICE with Kodachromes, but none so far with Dimage Scan. But I do occasionally get a more pleasing color balance with SilverFast, using fairly automatic and simple adjustments.
-
Scanning at too much resolution can have several disadvantages - processing time, huge file sizes, downsampling for smaller prints - to name the ones that come immediately to mind. The appropriate way to work this is to visualize the extent of cropping you will do, estimate the largest size you are likely to make the print, then decide on the required INPUT PPI as a function of OUTPUT PPI in the range I discussed yesterday. It really is pointless scanning at excessive resolution. I would never set my default at 5400 PPI. I don't know how to change default settings on DSE 5400-II because I use the original 5400 model and have not used Minolta's software for some time now.
Mark,
This might be good advice for some and a recommended work flow for someone who is still using film and needs to make prints quickly, but for archival purposes I see no point in scanning at less than full optical resolution. The time that matters is really about the same, and that's the time it takes finding and selecting the slides for scanning, cleaning them with isopropanol, dusting and blowing, possibly removing them from the slide mount if they haven't been centred properly or if the slide mount is simply cropping stuff that's important to the composition, carefully setting them in the film holder and making the adjustments to each individual frame in the preview before scanning.
Who wants to go through that more than once ??? . The fact that a batch of 6 negatives takes half an hour to scan at 8x instead of 10 minutes is not an issue. I can be doing something else in the meantime, like writing this post .
A few years ago, a 230MB file would have been painfully slow to process in Photoshop. Not anymore. My fastest computer has 6GB of DDR2 RAM and the next fastest, 4GB. Some of the slides I am now scanning were first scanned by Kodak at 2000 dpi about 11 years ago. It took a full 2 minutes to open an 18MB PhotoCD file in Photoshop 4. It now takes me 3 or 4 seconds to open a 230MB scan of the same image. That's progress .
I'll experiment a bit with Neat Image and PK Sharpener (which I haven't got yet). If there really are better ways of removing grain than GEM or SilverFast's GANE, then I should be using them.
-
Agree - if you guys are making prints in the range of 3 feet long dimension then it certainly makes sense to scan at 5400 PPI - ideally a bit more, but that exceeds the brains of the machine!
RAW scanning - I never got into that with Vuescan so I can't comment on it, but funny you should ask, I just finished a sequence of emails with Lasersoft Imaging and I now have a clear fix on how to set-up Silverfast for doing it. I shall try it. For negatives of course it will produce a negative image; for slides it will produce a positive. There are four adjustments involved. Please let me know if you want them. I'll send you all the gory detail with no warranties attached, because I just got the info confirmed and haven't tried it yet.
The better image quality (colour balance and luminosity) I see with Silverfast happens as a result of PRESCAN adjustments - not post-scan. When you get a better quality scan into Photoshop, you are already ahead of the game. As I said, I don't look for perfection at the scanning stage - just a very good jumping-off point for Photoshop. I leave the fine-tuning to Photoshop, which does it better. It saves ALOT of tweaking time to get a good quality scan into Photoshop and Silverfast like it or not just does this better - from my experience, and I am doing this day-in, day-out.
-
No it's not a dog - dog's are generally friendly to us humans.
-
A bit rich as a solution, but I guess it's too much to expect "one size fits all" in this domain.
Mark,
The expense of SilverFast dwarfs the cost of the other programs. Nikon Scan and Dimage Scan are free sofware that ship with the scanner and Vuescan is exceptionally good value. Because the same version works with most scanners, I can consider it as free software for my Elite 5400ll because I'd already bought the software for my Nikon scanner. There seems to be just a one-off, lifetime payment for Vuescan which includes all future upgrades. There's almost no reason for not owning it if you have a lot of film to scan.
I find it strange that you have a mismatch between scanned image and preview. The scanner is made first and presumably LaserSoft then write the software to suit the scanner. With out a perfect, or at least very close match between the scanned result and the preview, all these very wonderful color adjustment tools of Silverfast are basically useless.
Whilst I think it unlikely that you've got the wrong settings in CMS, because you are more familiar with this software than I am; just in case, I'll give you mine :) .
Color Management[/u]
Internal - ICM
Monitor - ICM
Output - RGB
Profiles for ICM[/u]
Input - Konica Minolta .... 5400ll
Internal - ProPhoto RGB
Grey - none
Output/Printer - greyed out
Rendering Intent - Rel Col
Embedded ICC Profiles[/u]
Embed ICC Profile - box ticked
Profile to embed - ProPhoto RGB
One other point, and you are probably already aware of this, it is possible to get oneself in a situation where the preview looks just right but the scanned result is way out. There seems to be some combination of adjustments, or perhaps a hangover from a previous set of adjustments that haven't been reset, that requires a fresh prescan. I haven't identified the cause, but the fact that it happened once in an obvious way implies that it could have happened more than once in a less obvious way that I didn't notice because I'm not in the habit of comparing each and every scan side by side with the preview.
-
Ray, OK, I've done it now - scanned a negative RAW with all the settings correct, got a wierd wine-looking negative image, reversed the a and b curves in LAB and that failed to convert it properly, so undid that and went to Invert in Photoshop, which got it into a wishy-washy positive mode, then went back to LAB and tried to improve it, which was futile. Conclusion: three cheers for scanner software that converts and adjusts beforehand!
My next test will be to try the settings combo you recommended that other day.
-
Having just acquired a Dimage Scan Elite 5400ll I've been busy comparing scans using different software; namely Vuescan, Silverfast and the software that came with the scanner.
It's a right mess, I have to say. The results are all over the place. I'm reaching the conclusion that one particular scanning software will do a better job than another only in respect to a particular film type.
For example, Vuescan does a better job than Dimage Scan with regard to Ektachrome and Kodachrome because profiles of those films are available in the Vuescan sofware. In Dimage Scan there's no choice of film profiles.
Silverfast has no choice of slide profiles and (therefore?) Silverfast produces worse results than Vuescan in this respect.
When it comes to scanning negatives, Dimage Scan is best with Fuji Reala (for example. I haven't tried every film type), even though there is not a Reala choice. Vusescan is not as good, even thought there is[/u] a Reala choice. SilverFast is worse still and there's no Reala choice.
Total chaos as far as I'm concerned. To get the best scan from a particular slide or negative, I feel I have to create a separate folder for each image and scan it about 6 times with different software and settings.
Remember, these are old slides and negatives. I certainly can't remember the exact color scheme of the scene and I'm relying upon the software to give me a pleasing, natural and convincing result.
-
I don't think I want to go to that trouble, especially considering that most of the stuff I'm scanning will have faded to varying degrees. I was sort of hoping that the premier scanning software, SilverFast, would have included a greater variety of film profiles. That there's no Fuji Reala listed is surprising considering how expensive the software is.
The degree of control in Silverfast is quite phenomenal but I haven't the inclination or time to spend fiddling around with minute color adjustments which often don't get me as far as a single click on Photoshop's 'auto color'.
-
Ray-
The purchased version of Siverfast comes with an IT-8 target for your particular scanner. You then create an icc profile for the scanner, not for each individual film type. (Although you certainly can do this if you wish!)
I have gotten some great results using Ektachrome VS, E200 & Fuji chromes (this photo is E100VS: http://www.kezman.info/lisbonne/lisbonne.html (http://www.kezman.info/lisbonne/lisbonne.html) ), but you are correct -- there is a large difference between the 2, especially for B&W. I shoot a lot of Ilford FP4 & PanF & Vuescan just falls down there. The TMax profiles though are quite good!
The other thing to do is to scan flat -- no cc whatsoever & then do the adjustments in Photoshop. With some images, this is the only way to get good results.
I hope this helps & happy scanning!!
-Jim
-
Thanks Lisa for that suggestion. When I come across an unexposed frame, the tail end or beginning of a strip of film which I didn't bother to cut off, I'll give that a try.
Frere Jacques, you're right. LaserSoft have stuck Reala right at the bottom of the list of ISO settings, almost as an afterthought it seems. I'm still undecided though, about buying SilverFast. The before and after shots I've seen after calibration with an IT8 target are far too subtle to make much difference with the stuff I'm scanning. I'm still searching for a positive and consistent improvement that I had expected SilverFast would provide. In fact, with grainy and scratchy film, Dimage Scan's combination of ICE and GEM at 100% seems to do a better job than SilverFast's GANE and ICE or SDR.
But I'm still experimenting. Haven't tried B&W yet.
-
Ray,
Just to confirm Mark's suggestion not to try and perfect your image in the scanning software; it's so much easier in Photoshop, providing you get a good basic scan to start with.
Although I do not have any experience with the Minolta, I have been down the same route as Mark with a Nikon Coolscan 4000. I eliminated Silverfast on the basis that the software was excessively expensive for the results compared to Nikon Scan and Vuescan. After extensive testing I've settled on Vuescan, having created a profile in Vuescan with an IT8 slide from Wolf Faust for the films I tend to use. Very simple operation to create the profile and it does make a subtle difference to the scan. Of course, this information is of no use to you if you do not have IT8 slides for the films that you are scanning.
However, even without the profile, I found that the basic scan was good enough from Vuescan to provide a good base to continue in Photoshop. Having said all of that, there are some slides that respond better to Nikonscan than Vuescan and I have no idea why that should be.
I agree with Mark - forget grain reduction and sharpening until you get the image into Photoshop. The tools available are just so much better.
Clive
-
I also use the original Minolta 5400, but so far I've just used the Minolta software and scanned Fuji Velvia 50. I use ICE, exposure compensation and manual focus. I use exposure compensation primarily to get as bright a scan as possible without clipping, and sometimes to do some minor color correction. I always scan at 5400 ppi and 16 bit (sometimes linear). I also use Neat Image (I've made my own profile that I use for all my Velvia scans) and PhotoKit Sharpener. I make all other adjustments with Photoshop CS. I'm not interested in batch processing. I scan all my slides unmounted.
I've read a lot about SilverFast and VueScan, but I'm still not clear if they would provide any real benefits for my workflow. I've always assumed that I would get the best results by using scanning software just to make hardware adjustments with all post scan adjustments done with Photoshop, Neat Image and PK Sharpener.
Does SilverFast or VueScan provide more hardware control than the Minolta software? What can VueScan or SilverFast do for my work flow (no batch processing) that the Minolta software cannot?
The one area of improvement that I'd like is to be able to get more dynamic range, specificaly brighter scans without clipping. I assume that I'm hardware limited and that neither VueScan nor SilverFast will be an improvement in this aspect. Am I correct?
-
Dmerger, I forgot to mention - I don't know what size prints you are making from your Velvia scans, but unless you crop heavily or make full-frame prints much larger than A3, you are probably generating excessive file size and scanning time going for 5400 PPI on every scan. Tests that I conducted with Harald Johnson are reported on pages 83 to 85 of his book "Mastering Digital Printing Second Edition". Harald and I concluded: "......there's a wide latitude of scanner settings to produce similar quality results" and ".........you don't need as much resolution as you once thought". More specifically, and this is corroborated elsewhere in the literature, anything in the range of 240 to 360 PPI (depending on image size) will deliver very high photo quality output.
For example, looking at the long dimension of an A3 print with a one inch border each side, one is scanning for 14.5 inches output. Say at 300 PPI, total pixels is 4350 (300*14.5). The long side of a 35mm film frame is 1.41 inches. Therefore if using the full frame, one would scan at 3085 INPUT PPI (4350/1.41). A far cry from 5400. One gets closer to 5400 of course the larger the final print and the more cropped the original frame.
I have played with manual and auto focus on that scanner and I am convinced manual focus (at least in my hands and eyes!) adds nothing to image sharpness. The auto-focus of the DSE 5400 is uncannily accurate in the sharpness of detail it reveals, including film grain unfortunately!
-
Clive[/quote]
I agree with Mark - forget grain reduction and sharpening until you get the image into Photoshop. The tools available are just so much better.
Well, grain reduction in Vuescan is not as effective as GEM in Dimage Scan and Nikon Scan. Are you implying that Photoshop can do a better job than GEM or ICE?
I find if I want the smoothest result with a grainy, scratchy film with minimum loss of fine detail, then ICE and GEM at 100% plus sharpening at 100% in the Dimage Scan software produces very acceptable results. The sharpening is halo free and at least partially restores the slight loss of resolution without bringing back the grain.
However, given the time, I expect I could do a better job in Neat Image regarding grain and noise removal. But it seems a fact of life that all grain and noise removal techniques result in some loss of resolution. It's a matter of chosing the better of two evils, I guess.
-
Ray,
Scanning at too much resolution can have several disadvantages - processing time, huge file sizes, downsampling for smaller prints - to name the ones that come immediately to mind. The appropriate way to work this is to visualize the extent of cropping you will do, estimate the largest size you are likely to make the print, then decide on the required INPUT PPI as a function of OUTPUT PPI in the range I discussed yesterday. It really is pointless scanning at excessive resolution. I would never set my default at 5400 PPI. I don't know how to change default settings on DSE 5400-II because I use the original 5400 model and have not used Minolta's software for some time now.
If you find Minolta DiMage Scan's version of ICE works better than that of Silverfast for Kodachromes , and if the Kodachromes are really dirty then perhaps use DiMage Scan, but in principle ICE should work in about the same way for both scanning suites, so perhaps there are some settings in Silverfast that it doesn't like. I have not yet tried scanning Kodachrome with Silverfast so I haven't seen the problem or had to work on a solution.
Once you've used Silverfast intensively for a week or two, and discover what procedure and generic settings work best for you, it does become quite easy and even somewhat routine thereafter. For each image, I seldom need to tweak much more than the grey balancer and mid-range luminosity to get a very acceptable scan for further work in Photoshop. Now that I bought it, pestered and lectured their tech support to near exhaustion and have a good enough handle on how to use it, I can't think of any value-added going back to VueScan or DiMage Scan.
Grain reduction and sharpening at the scanning stage is workable of course but in my view sub-optimal. It adds enormously to scan time and the tools simply don't provide the kind of control and reversibility that you have with software such as Neat Image and PK Sharpener (no visible halos at any print size you are likely to make). When you talk about choosing the better of two evils between grain reduction and resolution - this is exactly the point - the control that you have over these processes with superior software really does minimize the trade-off - in fact often eliminates it totally. ICE is a different story - it is aimed at removing sludge, crud and other impacts of emulsion deterioration; while this can be replicated in Photoshop it is more simply and effectively done with ICE at scanning.
-
Hi Ray - I see your point - the whole scanning process is indeed painful - one of the reasons I bought a used Canon 1Ds last year. What I'm doing with negs now is legacy stuff that I wish to seriously improve over what the commercial processors gave me. Within a wide range it's really a matter of comfort and personal requirements what scanning res to use - so if ultra-high res gives you that warm and fuzzy feeling, that's fine. I also do other stuff while the scanner is churning away, so nothing lost really. (For example, working my way through Dan Margulis' new book on working in LAB - a real "tour de force" in the overcrowded Photoshop literature.)
-
My goal is to get the best possible "RAW" scan from my scanner, without any post scan, non-hardware image adjustments, and then make all my image adjustments with Neat Image, PK Sharpener and Photoshop CS. So, what I'm really interested in is whether SilverFast or VueScan is capable of producing better "RAW" scans.
That's an admirable goal but I don't think it's possible. For example, if you are scanning film that's underexposed, in order to reduce noise in the shadows it's necessary to make an 'exposure compensation' adjustment in the scanning software. Dimage Scan allows one to do this and one then has to rescan the preview to see the effect. Nikon Scan has a similar feature they call 'analog gain'. It also requires a rescan to see the effect. The exposure or gain control changes the time period the film is exposed to the scanner's light.
Even in situations where there's a full histogram but overly dark shadows, you will find that moving the middle gamma slider to the left, in the preview's histogram, will result in less noise in those shadows than doing a straight scan and opening up the shadows in Photoshop.
I've also wondered if third party scanning software is able in some fundamental way to get more out of the scanner, increase its basic performance, but I haven't found this to be the case with any of the software and scanners I've tried so far. Alternative scanning software seems to me to be mostly about alternative color and contrast balance and different or more options to avoid later work in Photoshop. I imagine anyone who is a real whiz in Photoshop techniques could do an equally good job whatever scanning software they used (within reason :) ).
-
I guess I'm not making myself clear. Ray, exposure compensation adjustments are hardware adjustments. As I understand how it works, the adjustments shorten or lengthen the exposure time. I use those adjustments in the Minolta software. The result is a "RAW" scan in that no software only, post-scan image adjustments are made.
Mark, as I understand SilverFast and VueScan, as well as the Minolta software, all scaner adjustments (other than exposure compensation and ICE) are done by adjusting the data resulting from the physical "RAW" scan. Although such adjustemnts are selected pre-scan, they are merely post-scan data manipulation, which can also be done in Photoshop.
Mark, I'd be very interested in any info you received about RAW scanning with Silverfast.
Notwithstanding what I've said about wanting a "RAW" scan, I can see where there could be a real benefit in improved and/or custom profiles for slide film and my scanner. From what little I've read, however, it appears that custom scanner profiles are not very helpful.
-
Now, turning to the nitty gritty on RAW scans with Silverfast, here is the recipe, as confirmed by LaserSoft and so far UNtested by me:
(1) In the FRAME tab for SCAN TYPE select <48 bit HDR colour>.
(2) In the OPTIONS>GENERAL tab for Gamma Gradation leave the box before <for HDR Output> UNchecked.
(3) In the OPTIONS>CMS tab for <profiles for ICM Internal> select NONE.
(4) Also uncheck <Embedded ICC Profile>.
Mark,
I've tried the above with my trial copy of SilverFast. I need to make one more adjustment in Options/CMS, Input-Internal = None.
I can see no benefit in doing RAW scans. There's no more shadow detail, the colors and contrast are completely off, and the work in Photoshop to correct everything and make the image look as good as the scanning software adjustments I normally make, is either beyond my expertise or beyond my patience.
By the way, RAW scans of color negatives come out still as negatives, despite the fact that the preview is positive. Is this right, and if so, what's the purpose of that?
-
Secondly, I ran into a problem with Silverfast because it does not handle Prophoto colour space properly (the colour balance and luminosity in the scan window are not replicated as such in the scanned image file as shown in Photoshop when ProPhoto is used in both). In order to see whether it is a scanner problem or a software problem, one needs to start with the raw data and do tests.
Mark,
That is a worry. At the moment I'm using SilverFast on a cheap Sony 17" LCD on its lowest luminance setting, which I've calibrated with ColorEyes and the X-Rite DT94.
Working space is ProPhoto RGB and SilverFast settings embed that profile. I see no difference between the scanned image opened in PS with the embedded ProPhoto profile, and the SilverFast preview. That's how it should be and I'm happy with that, and also surprised that I can get this close a match with a run-of-the-mill LCD monitor.
I've been undecided about SilverFast for some time. My impressions so far are that Dimage Scan (for the 5400ll) produces better results with negatives. But I haven't tried all types of negatives in my archives yet. Vuescan seems to do the best job with B&W. SilverFast does a better job with Kodachrome and Ektachrome slides.
Since I have a lot of treasured Kodachromes and Ektachromes that go back 40 years, I've decided to buy the SilverFast unlock key. But I'm nevertheless puzzled that SilverFast is not doing a good job with B&W and color negatives. A friend has a Dimage Scan Dual lll, 2800 dpi. The software did rather poorly with color negatives. He's upgraded to the 5400ll on the strength of the superior handling of color negatives.
IF I have to use 3 different scanning software programs for 3 different types of film, then so be it.
-
Hi Ray, yes I am aware of those options, but I haven't activated them, because I live in "mortal fear" of things that say something like "Reset ALL" when "all" isn't fully defined and their challenged documentation doesn't help much. I've had in the back of my mind that there could be gremlins lurking in the deep dark corners of the software ( of which there are many) just waiting to skewer me for being over-zealous with the commands.
Anyhow, now that you have taken the risk of de-mystifying it, that's great, I'll be less fearsome. (Perhaps a bit irrational in the first place - after all, just a few more minutes of head-scratching to get things back on track in case all-#### breaks loose; but I just have so many of these legacy negatives I want to finish scanning, that I try not to get bogged down, while still mastering the thing enough to get the quality I want - which by and large I have.)
The ProPhoto dialogue I'm having with Lasersoft - and they have NOT yet been able to either replicate or resolve - is an attempt to put a bit of icing on the cake just in case I can eak-out a bit more in-gamut values that the Epson 4000 can handle. Of course you are absolutely correct that it makes nonsense of image adjustment in Silverfast if what I see in their prescan window doesn't end up looking the same in Photoshop - that was the basis of my original complaint to Lasersoft. I am open to the (perhaps remote) possibility that the problem may not be Silverfast - it could be model 1 of the 5400 scanner - that is why I got into the idea of producing RAW files and working them in LAB using ProPhoto space, to see what misses out on what. But I haven't tried it yet. The 5400-1 specs do not list ProPhoto as a supported working space, but it does support something called "wide gamut RGB" which may - or may not - be the same thing. Minolta tells me the scanner itself would not clip colours within the ProPhoto work space. Let me ask you - does the 5400 model 2 explicitly support ProPhoto?
I agree, in respect of its fundamental performance Silverfast is a VERY good piece of software, underneath all its quirkiness and poor quality documentation. I especially like the grey triangular "pipette" for achieving colour balance where there is something (or some things) in the image that definitely should be true grey. Also, its colour wheel is handy for some balancing work, and the mid-range exposure setting "L" is quite powerful at rescuing seemingly hopeless negatives. Sometimes I find this stuff really is good (if not essential in case of heavily under-exposed negatives) to use before getting the image into Photoshop, but I remain convinced there is nothing out there that fully competes with the scope, depth and flexibility of Photoshop's colour and luminosity adjustment tools.
-
I use the Dimage Scan software with the Minolta DS5400 (first version) and get excellent results for Fuji Provia 100F and Kodachrome 64. In the case of Fuji, I scan, then in PS apply a profile generated from a Q60 slide from ColorAid in Germany. Unfortunately I cannot find a Q60 slide for KR64, but I find that the scan is fairly close, and just needs some shift to the relative RGB brightness before the scan. I have Vuescan but don't like it. It's supposed to allow generation of a profile but I couldn't get it to work. I tested SilverFast, but I hate it. The interface is IMO a dog in terms of ergonomics.
I find that ICE works well with KR64, and the auto-focus is useless. I always manual focus and get sharper results. Very rarely I get 1 pixel wide coloured lines in a scan. Blooming is not an issue, unlike with several other scanners I owned.
Leif
-
Ray, interesting that you have not had the ProPhoto problem. Now, you are using the 5400-II and I am using the original 5400. Maybe that is the rub. Perhaps Minolta have improved something affecting this.
For my set-up Silverfast is by far the best at handlng colour negatives notwithstanding the ProPhoto issue. [I haven't used it for anything else so don't know the relative merits for other media.] I do know that in the 5400-II Minolta claims superior negative handling capability - probably some combination of both hardware and software.
Yes - in terms of what software to use - I think you have it right - as long as the moolah is available, buy whatever works best for each medium. A bit rich as a solution, but I guess it's too much to expect "one size fits all" in this domain.
-
Hi Ray,
Thanks for your settings. Next time I scan I shall compare them with mine, and if not the same, try them out to see what happens. I'll let you know.
If I have made any adjustments at all to an image, I always reset them before doing the next image. This is most easily handled by clicking on the colour balance, mid range exposure or histogram tools as appropriate and then clicking reset. It works.
-
Mark,
The Dimage Scan software for the 5400 ll also does not have a ProPhoto option. This fact was one of the first differences I noticed when comparing Dimage Scan with both Vuescan and SilverFast. I don't think your mismatch between preview and scanned image, using the ProPhoto space, can be explained by this omission.
-
Mark,
Glad to be of assistance. I'm often accused of being impractical :D .
-
If the scan software accepts ICC profiles, you can shoot a Color Checker SG under appropriate lighting and make your own film profiles if the film type is still manufactured. If not, there isn't much you can do but sweat blood and grit your teeth.
-
Ray-
One other thing. Silverfast does have a Reala profile, it's just well hidden -- in NegaFix, choose Fuji, the Superia, then in the last box scroll all the way down & you will find Reala 100 listed.
-JMK
-
Clive,
Yes - if only using positive transparency film of a type for which there is a useful IT8 target, Vuescan should be fine at much lower cost. I must emphasize that my experience and recommendation is based largely on what works best for colour negative film. This is a very important distinction that fundamentally affects choice of software.
That much said, I just looked at Silverfast pricing and saw that while they charge 258 dollars for the version that supports the Minolta scanner, they charge 407 for the equivalent version supporting the Nikon Coolscan 4000. That ups the ante quite a bit! It's never been clear to me why these large price differences as a function of scanner model - maybe it reflects complexity of configuration and corresponding development effort, who knows.
Dmerger, again dealing with negatives, the luminosity range and colour balance of colour negative scans is much more controllable with Silverfast and producers brigher, richer results than I've been able to achieve with either Vuescan or with Minolta's native software.
There is a balancing act about how much adjusting to do in the scanning software and how much in Photoshop. Heavily underexposed images do need to be tweaked quite heavily at the scanning stage, otherwise Photoshop will not be able to rescue them to their full potential. For more normal exposures, the saw-off I discussed in my previous post works well.
In relative terms, I don't think we are hardware limited with the Minolta 5400 scanner - it has the highest specified DR and optical resolution of any desktop film scanner on the market. I think software issues are much more determinative than hardware constraints.
-
Well, grain reduction in Vuescan is not as effective as GEM in Dimage Scan and Nikon Scan. Are you implying that Photoshop can do a better job than GEM or ICE?
Not really. For grain I use Neat Image as a PS plug-in and sharpening is handled using Photokit Sharpener. I do use ICE in NikonScan, or the equivalent in VueScan.
Clive
-
so if ultra-high res gives you that warm and fuzzy feeling, that's fine.
Hey! It's the 24x36" prints from my Epson 7600 that give me the warm, fuzzy feeling :D .
-
Mark, for my purposes, I agree with Ray about scan resolution. I try to scan once at the highest resolution. My biggest prints have been 22 x 33.
I'm afraid that you may have missed my point about SilverFast and VueScan. I'm not suprised that you can get a better looking scan with SilverFast, but how much of the improvement is purely post scan image adjustments made with SilverFast software? I suspect that virtually all the improvements are just post scan manipulation and that I could do the same or better with Photoshop CS. Is my assumption incorrect?
My goal is to get the best possible "RAW" scan from my scanner, without any post scan, non-hardware image adjustments, and then make all my image adjustments with Neat Image, PK Sharpener and Photoshop CS. So, what I'm really interested in is whether SilverFast or VueScan is capable of producing better "RAW" scans.
-
dmerger, you are raising an interesting and important question about what adjustments happen pre-scan and what post scan but pre-Photoshop. The importance is that the key technical argument used by the proponents for adjusting as much as possible before scanning is that the scanned image then ends-up in Photoshop with no data loss and little do in Photoshop that would take its toll on the pixels - remembering that most image corrections destroy information. That argument loses alot of its mathematical appeal if in fact most of the calculations affect already scanned pixels whether it happens in the one piece of software or the other. One thing there is NO question about - heavily underexposed originals MUST be brightened up during scan exposure otherwise Photoshop cannot cope adequately.
As for custom scanner profiles for handling POSITIVES, from what I've read in the colour management literature, calibrating and profiling the scanner for the material that will be scanned is part of a properly colour-managed workflow and should help improve both predictability and process efficiency. Fraser, Grey, Hutchinson etc. all write about this.
Now, turning to the nitty gritty on RAW scans with Silverfast, here is the recipe, as confirmed by LaserSoft and so far UNtested by me:
(1) In the FRAME tab for SCAN TYPE select <48 bit HDR colour>.
(2) In the OPTIONS>GENERAL tab for Gamma Gradation leave the box before <for HDR Output> UNchecked.
(3) In the OPTIONS>CMS tab for <profiles for ICM Internal> select NONE.
(4) Also uncheck <Embedded ICC Profile>.
I'm curious to get your reaction on how it works and what it does. Let us compare notes.
-
Ray,
You got what is expected. As for why do it, there are several reasons:
Firstly, specific applications can ask for such data - for example, CF Systems for converting from negative to positive.
Secondly, I ran into a problem with Silverfast because it does not handle Prophoto colour space properly (the colour balance and luminosity in the scan window are not replicated as such in the scanned image file as shown in Photoshop when ProPhoto is used in both). In order to see whether it is a scanner problem or a software problem, one needs to start with the raw data and do tests.
Thirdly, if you want to test whether the scanning program's settings and adjustments are doing all they can for you, it is good to be able to start with the same raw data it starts with and see whether you can improve on it. If you can, aspects of their software are not worth paying for; if you can't, they are.
I haven't made the time to experiment with it yet - other things have priority just now, but when I do in the near future I shall most likely be making my adjustments in LAB rather than RGB or CMYK. This is simply because the easiest way to convert from negative to positive is to reverse the A & B curves, then colour balance can be further tweaked with those two curves leaving luminosity unaffected, while luminosity can then be adjusted on its own using the L curve. In this way one may be able to develop a recipe, or a "preset" generally useful for a film type, very much like the scanning software does. If it provides no value-added over the standard way of using Silverfast, so be it - back to the standard way. If I find I've improved image quality relative to the standard approach, well - value-added.
-
This is most easily handled by clicking on the colour balance, mid range exposure or histogram tools as appropriate and then clicking reset. It works.
Mark,
Are you aware that there's a 'reset' and 'reset all' option?
At the bottom of the 'Frame' window where you have Prescan, Scan, Quit and Options buttons, pressing ALT turns the Options button into a Reset button and pressing SHIFT turns the Options button into a 'Reset All' button.
I now 'reset all' with a single click before each new prescan. Resetting 'all' does not change the color management settings, fortunately :) .
The more I use SilverFast, the more impressed I am. Their color adjustment tools seem to be as good as Photoshop's curves, but easier to use.
-
Color Management[/u]
Internal - ICM
Monitor - ICM
Output - RGB
Profiles for ICM[/u]
Input - Konica Minolta .... 5400ll
Internal - ProPhoto RGB
Grey - none
Output/Printer - greyed out
Rendering Intent - Rel Col
Embedded ICC Profiles[/u]
Embed ICC Profile - box ticked
Profile to embed - ProPhoto RGB
Ray,
This recipe works. The Prophoto problem is solved. Thank you EVER SO MUCH. When working with negatives, the only key change I had to make, apart from (obviously) selecting the ProPhoto working space, is to select <ICM> instead of <none> in "Colour Management" for the "internal>monitor" window. With negatives "input>internal" is greyed out, and under "Profiles for ICM" "Input" is greyed out. So the whole problem was one simple choice for "Internal>Monitor" that I missed trying and Lasersoft after all this time and back and forth with me never thought of suggesting. I much appreciate your astuteness for discovering this.
Mark
-
Having just acquired a Dimage Scan Elite 5400ll I've been busy comparing scans using different software; namely Vuescan, Silverfast and the software that came with the scanner.
It's a right mess, I have to say. The results are all over the place. I'm reaching the conclusion that one particular scanning software will do a better job than another only in respect to a particular film type.
Ray: you really must buy some IT8/Q60 profile slides. www.coloraid.de do some. Then create profiles. With Dimage Scan, scan as per normal, then apply the profile in Photoshop. Vuescan and SilverFast lets you create a profile and then scan with the profile 'loaded'. If the profile is applied while scanning (lamp brightness adjusted) then you should get better results. Otherwise I see little advantage to either method.
I find it easy to get good scans from the Minolta 5400, using Dimage Scan, which require little post scan adjustment apart from application in PS of a generated profile. The adjustments I make tend to be to compensate for a blue cast present on the slide. In the case of KR64, I find that reciprocity failure is so common that no profile would help. Usually adjusting the lamp brightness (reduce blue) is important.
Leif
-
Leif,
I have the professional version of Vuescan and SilverFast which allow calibration of the scanner using an IT8 target. I've thought about ordering a target through the internet but I'm not convinced it will serve much purpose. My previews in the scanning software match the scanned result very closely; closely enough in fact. The conclusion I draw from this is my scanner (the Elite 5400ll) is already sufficiently well calibrated.
I don't see much point in creating profiles for different film types using IT8 targets because the film I'm scanning is mostly old stuff which has faded to some degree. I now use digital cameras. At present I'm scanning old Kodachromes and Ektachromes, twice, once using Vuescan and once using SilverFast. Sometimes the automatic settings in Vuescan, with their built-in Kodachrome and Ektachrome profiles, produce better results than several minutes of stuffing around with SilverFast. :)