Post by: saiguy on November 22, 2017, 08:41:35 pm
What thoughts have you smart persons on this?
Post by: Alskoj on November 22, 2017, 10:33:59 pm
I think I have read several times to scan B&W at 600 ppi. I always scan them in RGB mode. Read thru Mark Segal's SF8 book again but did not see any definitive mention on this.Are you talking about a flat bed scanner (for prints) or drum scanner (for film)?
What thoughts have you smart persons on this?
Post by: TonyW on November 23, 2017, 05:36:49 am
I think I have read several times to scan B&W at 600 ppi. I always scan them in RGB mode. Read thru Mark Segal's SF8 book again but did not see any definitive mention on this.If you are scanning a good clean original B&W print then 600 should in most cases be quite adequate - it may also be that if the original contains a lot of detail that your scanner may be able to record this using a higher res. E.g. 1200 SPI. You may want to experiment and see where your scanner limits are.
What thoughts have you smart persons on this?
On the other hand if you are trying to restore an old damaged and stained print then scanning in RGB may offer benefits e.g. removing stains by using colour adjustments rather than cloning and patching.
Post by: Mark D Segal on November 23, 2017, 07:32:50 am
I think I have read several times to scan B&W at 600 ppi. I always scan them in RGB mode. Read thru Mark Segal's SF8 book again but did not see any definitive mention on this.
What thoughts have you smart persons on this?
Chapter Five of my book covers this in considerable detail, and the same principles explained there apply whether scanning film or paper. The appropriate resolution depends on the relationship between the relative dimensions of the original versus largest scanned output you will ever want to retain and the number of pixels per inch you need for the output purpose (e.g. print or web viewing). It will be finally limited by the optical resolution of the scanner. Some people recommend scanning at the maximum "optical" resolution of the scanner (means resolution that isn't boosted by resampling in the software) so that you retain the maximum amount of highest quality information available without needing to rescan at some future date. Other people, me included, recommend scanning at a resolution that you think will meet your largest anticipated requirements. This may be less than the maximum optical resolution of the scanner. The purpose of this approach is to keep file sizes more closely aligned with need, but if you don't care about file size, go big.
Beware that if you are scanning a large original (for example flatbed scanning of a print), there are several things to think about before making a final choice of resolution: (1) what's the effective resolution of the original media you are scanning? Chances are it's no more than about 200~300 PPI equivalent, and no matter how many more PPI you scan it at, the process will not resolve more than it starts with (leaving aside deconvolution technologies). (2) Pixel dimensions grow very quickly because you are starting with a lot of inches, compared say with a 35mm slide or negative. You'll want to scan at a resolution which reproduces in the final output size at least the effective resolution of the original, but being mindful of file size bloat beyond what's needed for the purpose. I've provided formulae in the book for working all this out quite easily.
Post by: saiguy on November 23, 2017, 09:09:31 am
Thanks Mark, Will read chapter 5 again.
Happy Thanksgiving
Post by: Mark D Segal on November 23, 2017, 10:24:35 am
Happy Thanksgiving
And to you too.
Post by: digitaldog on November 23, 2017, 11:59:22 am
I think I have read several times to scan B&W at 600 ppi. I always scan them in RGB mode. Read thru Mark Segal's SF8 book again but did not see any definitive mention on this.B&W, color; makes no difference. Scan at the highest optimal resolution if possible. Then you'll have all the data possible for all needs from that one scan. Now if you know for a fact you will only use less data once and never again, OK but never is a long time.
What thoughts have you smart persons on this?
Scan once, use many. So maybe color since you can convert to B&W but can't go the other way!
Post by: Chris Kern on November 23, 2017, 05:07:53 pm
Scan at the highest optimal resolution if possible. Then you'll have all the data possible for all needs from that one scan.
A while back, I scanned a 7.5x10-inch print of my great-grandfather's home that was made in 1895. The scan resolution was 1200 dpi. The original had been exposed to light over the intervening 100+ years, and was a bit faded. I suspect that cost me some detail in the shadows, but the sharpness amazed me. A 19x26-inch enlargement that I made looks fine at any reasonable viewing distance—and remarkably good even using the classic photographer's test for acuity: i.e., at the end of my nose.
Post by: saiguy on November 23, 2017, 06:17:19 pm
Mark Segal in his SF8 book talks about the 2 different approaches, max scanner res., and right sizing. I have been using right size which for me from 35mm slides is an out put for 5x7 inch print at 300 ppi. This is for a general purpose archive. If a large print is needed I would just rescan.
Digitaldog, Scanning B&W in RGB allows access to more PS tools. It's particularly useful if there are photo restoration issues. If I print I use the Advanced B&W mode and do not convert.
Maybe the 600 ppi is a pre press norm. Seems to have no support here.
Thanks for all replies,
Post by: digitaldog on November 23, 2017, 06:28:17 pm
Digitaldog, Scanning B&W in RGB allows access to more PS tools.Kind of doubt that if the original itself is in B&W, otherwise I agree. Scan color IF you wish to convert to B&W or better, have both options. Kind of like scanning big; you don't need to rescan (assuming the scanner is of sufficient quality for your needs) plus you have no guarantee that in the future, that film or print will be around or in the same condition as the first time you scanned it. So my stance, as someone who's reviewed, sold, used and provided scans in a 'service bureau' (last century) is scan once, highest resolution and be done with it.
600PPI, 300PPI, folks all over the net believe there's some standard value to always use; not so. Like the old urban legend all displays are 72PPI (nope) or that all scans for all prints should be 300PPI (or the printer produces or must have 300DPI), nope.
Post by: TonyW on November 23, 2017, 06:48:45 pm
I mentioned scanning for restoration of less than pristine prints and if you do this type of work do not underestimate the value of scanning at higher sampling rates e.g. 1200. Doing so may assist with bringing detail in even the damaged areas to aid in the restoration process.
Finally as a digital rule of thumb if you require to end up with ‘X’ amount of data it is not a bad idea to start out with 2x ‘X’ to give some headroom to what happens in post prior to final output size
Post by: saiguy on November 23, 2017, 06:57:27 pm
I am starting a project to scan nearly 9000 Kodachrome 35mm slides. Maybe a dozen or 2 will later get rescanned. Won't know of such selects till the project is done and can be viewed. All are from 1977 to 2004 and shot in India. Of course all of these are color. My friend does have many historic B&W that will be scanned later.
That's why I plan to use a 5x7 general use size. I can get good letter size prints from them on my 3880 printer. Comparing a native 5x7 print to a letter size of the same file. my eyes can't see degradation.
Post by: digitaldog on November 23, 2017, 07:00:09 pm
Digitaldog, Katrin Eismann says to scan B&W in color mode and gives ample reasons why in her Photoshop Restoration & Retouching book.I've know her for decades and respect her enormously. I don't see the point. But then, if the original is B&W, depending on the scanning software, it is possible that the results setting it to grayscale could be suboptimal; the scanner IS scanning in color; alway! What it does with that trilinear color data to get grayscale can differ. But scanning a color image; you've got a lot of color data for the conversion so I'd never scan in anything but color IF I wanted B&W.
Post by: saiguy on November 23, 2017, 07:16:51 pm
Can you comment on my general purpose size for the 9k slides. They will use them to put together presentations, LR Collections, maybe publish to SmugMug. They might get a photo printer for letter size max prints.
Post by: CeeVee on November 23, 2017, 08:29:54 pm
Sent from my Pixel XL using Tapatalk
Post by: Chris Kern on November 23, 2017, 10:11:47 pm
nice house too.
The photograph probably was made just a couple of years after the house was built. The backstory is here. (http://www.chriskern.net/history/oldWoodsfieldHouse.html) Some time after I wrote that essay in 2009, my sister located the original print while we were closing up my parents' home after my father's death. And shortly after that, a distant cousin sent me a newspaper clipping that confirmed the photo was made in 1895; it also authoritatively identified the individuals in the tableau—almost all of them relatives of mine, although I previously had never heard most of the names.
Post by: TonyW on November 24, 2017, 05:35:27 am
TonyW, When I get to my friends B&W I will be using settings similar to your suggestions. There may be only 100 of them and would require what you are saying.9k slides is quite a job and the requirements for storage quite high. Other than printing image size requirement probably quite modest and you may want to be selective in how you scan bearing in mind purpose - only a decision that you can make.
Can you comment on my general purpose size for the 9k slides. They will use them to put together presentations, LR Collections, maybe publish to SmugMug. They might get a photo printer for letter size max prints.
FWIW just a few thoughts:
I would be looking at getting the optimum data from the image and using a Nikon scanner that get close to their claimed 4000 spi I would be tempted to go for max. But that may not necessarily be the best for your project?
My assumptions
Your print requirements max. Letter size 8.5" x 11" @ 300 PPI = 2,550 x 3,300 = 8.4 MP
35mm film = 36 x 24 mm = 1.42" x 0.94"
1. Scan for the maximum size you are going to print (in this case assuming print ppi required 300) so in the case of letter size (8.5"x11") you are going to need a scan comprising of 2550 x 3300 pixels. The pixel image size being 8.42 megapixels
Or
2. Scan at your scanners maximum optical resolution (believed to be 4000 spi for Nikon film scanner) which will create 3780 x 5669 pixels. The pixel image size being approx 21.43 megapixels approx 2.5x the size of the lesser scan. If you print Letter size from this you will be sending 515 ppi to the printer (close to the highest Canon printers require of 600 ppi). Will it make a difference? You may want to experiment and find out for your particular images
Which will be better for purpose depends on what remedial work you need to undertake and if the original contains the detail to make scanning at a higher (optimum) resolution worthwhile.
If you need to keep file size down and you are going to require remedial work on the image then IMO it may be worth while to scan at the optimal resolution, work on the image then resample to finished size and discard the larger working file
Post by: kers on November 24, 2017, 07:50:16 am
... a distant cousin sent me a newspaper clipping that confirmed the photo was made in 1895 ...
a beautiful slice of time !
One of the main attractions of photographs..
Post by: saiguy on November 24, 2017, 08:47:31 am
The Essay is great. Anyone scanning old photos should read it.
In Mark Segal's SF8 book he examines the Right Size & Max Size debate. Using an old 35mm Kodachrome slide BTW. I'll quote him, hoping that is OK to do; "My conclusion, based on this test and other work, is that for most intents these aren’t choices over which to lose much sleep".
The only remedial work expected is to address any cleanup that SF8 tools and Neat Image don't handle sufficiently.
Will go with your advise and use an out put size for letter. That will be 7.5x11 inch I believe. If they get a printer, likely will be a Canon, so will use 300 ppi. File size will be about 60 MB.
thank you again
Post by: Mark D Segal on November 24, 2017, 09:56:22 am
Digitaldog, Katrin Eismann says to scan B&W in color mode and gives ample reasons why in her Photoshop Restoration & Retouching book.
I am starting a project to scan nearly 9000 Kodachrome 35mm slides. Maybe a dozen or 2 will later get rescanned. Won't know of such selects till the project is done and can be viewed. All are from 1977 to 2004 and shot in India. Of course all of these are color. My friend does have many historic B&W that will be scanned later.
That's why I plan to use a 5x7 general use size. I can get good letter size prints from them on my 3880 printer. Comparing a native 5x7 print to a letter size of the same file. my eyes can't see degradation.
I think it may time to do a review of fundamentals here.
Let's start with the "optical resolution of the scanner" - the maximum the hardware can resolve without resampling the data. We throw this expression around, but it doesn't necessarily mean what the scanner specs say it is. For example, you can buy a scanner that rates itself at 6000 PPI (or DPI, same thing in this context) but when you test it with a resolution target you find that the EFFECTIVE maximum resolution may top-out at 3000 (page 125 of my SilverFast book provides some sobering information on several scanners in this regard). The scanner manufacturer isn't necessarily lying - what they call 6000 may well describe how the sensor is designed, but other factors such as lens quality, the lighting system, the flatness of field, and others can impact the EFFECTIVE outcome of scanning. Those factors will not vary as a function of the resolution you select. So if the scanner can't really deliver any more than say an EFFECTIVE 3000, it's pointless packing more unusable information into the file.
Now let's look at the "storage is cheap" argument. While that may be true, it's useful to consider how much storage you'll need as a function of file size. Firstly, you'll want to back-up the files you create. So right there, double the storage you start with. Then there is processing. If you are processing them "non-destructively" in Lightroom, metadata adds very little to file size, but if you are doing so in Photoshop, layers easily double the file size, if not much more, and assuming you wish to preserve those layers (the "non-destructive" aspect) you can at least double the file size again, along with its associated back-up. So what may start life as a 60MB storage requirement, for example, can easily end-up as at least a 240MB storage requirement by the time you finish - i.e. 4x what you started with. Cheap X 4 may no longer be so cheap. Add to this the consideration that larger files can take longer to retrieve, edit and save than smaller files and you end-up perhaps consuming more time than necessary for your purposes.
9000 slides is a lot of scanning - a daunting exercise in fact. So let's look at this with a bit of arithmetic. I'm assuming these are 24 * 36mm slides, so the maximum dimension is 1.5 inches (and I'm assuming no cropping here - cropping changes the calculations). The maximum print size you EVER IN THIS LIFETIME intend to make is letter size, with say a one inch border on each side, so the maximum dimension of the photo itself is 9 inches. You want to print it in a Canon printer at a "native" resolution of the Canon driver of 300 PPI. So you need 9 x 300 pixels, or 2700 total output pixels on the maximum dimension. To pack 2700 output pixels into a 1.5 inch input, you'll need 2700/1.5 = 1800 input PPI setting for the scan resolution. Or seen in a perhaps more intuitive obverse manner, if you scan that 1.5 inches at 1800 pixels per inch, you obtain 1.5 x 1800 = 2700 total pixels, which when divided by your 9 inch output (the print) gives you the 300 PPI the printer driver likes.
Now let's see what file size that creates using the formula explained on page 124 of my book (which delivers the same results that scanner software such as SilverFast reports for the same settings): the result is 27.8 MB, assuming 16 bit depth for the scanning, and a 6*9 inch photo (say on an 8.5 x 11 inch sheet, but only the photo dimensions count). Now let's say for some reason or other you decide you want to provide for an 11 * 14 inch print (or given a 1 inch border, 9 * 12 inch printed photo on an 11* 14 sheet) even though you may never ever in your lifetime make one. Applying the same formula with all the same inputs except for the change in photo dimensions, you'll have a file size of 55.6 MB - double what you need. And, by the way, you'll need 3600 output pixels on the maximum photo dimension to achieve this result, which means you need to scan at 2400 input PPI instead of the 1800 needed for the 6*9 inch photo, so scanning will take longer.
Now that we have all that straightened away, let's look at the size of this job - as I mentioned above, daunting. You would want to automate as much as possible. For this I would recommend that you look carefully at SilverFast's batch scanning capabilities.
You will want to get the scans in reasonably good shape at the scan stage to minimize time spent in post-scan editing. How much editing to do at the scan versus the post-scan stage is a mixture of technical reality, taste and judgment, all explained in my book - too much of a long story to put in a forum post, but if I had to put the most basic advice into a nutshell for you, whatever balance you wish to strike about what edits to do where, at the least, make sure when you scan you are sending to file an image that is relatively open (no excessive contrast) without blocked-up highlights and shadows, i.e. minimizing highlight and shadow clipping, and make sure your scanner is properly colour-managed (decent profiles etc.). Once the image file is created and you need to do any further editing and printing, I would recommend using Lightroom. The edits are all meta-data, non-destructive, preserved for future editing, and the workflow from editing to printing within that one application is seamless. Lightroom also has very good tools for sharpening and noise control, which work well with moderate film grain as well.
On the colour versus B&W scanning, I think we are all agreed that if you are scanning colour originals, scan them in colour mode and if you wish to convert them to B&W, do that afterward. Lightroom is great for this. This allows you to conserve the maximum amount of control possible over the eventual tonality of the B&W renditions.
Post by: TonyW on November 24, 2017, 10:16:57 am
....In Mark Segal's SF8 book he examines the Right Size & Max Size debate. Using an old 35mm Kodachrome slide BTW. I'll quote him, hoping that is OK to do; "My conclusion, based on this test and other work, is that for most intents these aren’t choices over which to lose much sleep".I have not read Mark's book but I do know him to be a knowledgeable and experienced photographer/author and therefore I am sure the conclusion he reaches is based on sound practice backed up with a good deal of practical testing. Even without the benefits of reading the rest of his text, therefore leading me to make some guesses about the thrust of the article suggests that his conclusions would be very sound.
The only remedial work expected is to address any cleanup that SF8 tools and Neat Image don't handle sufficiently.
For old photo restoration sometimes we get presented with very poor data and the expectation may be to make huge improvements (think of CSI Miami, Blade Runner or even better trying to polish a turd) sometimes much more than can reasonably be expected. Faced with such issues capturing the image at the scanners optimum spi can help, enabling post processing work to produce accurate masks that can help enormously in the clean up (attached is a real example of such a mask from a damaged print - white areas show what needs repairing. Sorry I cannot post the original and finished versions).
Similarly with B&W prints you may find that there is staining, silvering and general damage resulting in colour cast over the print. I view and treat this type of image as colour and find that this can offer advantages in post work even if it ends up back to B&W
Post by: Mark D Segal on November 24, 2017, 10:27:28 am
Post by: TonyW on November 24, 2017, 11:46:47 am
Thanks Tony. Let me suggest you'll get more bang from the buck with "turd polishing" by making sure to scan in 16-bit depth (not 8) rather than packing-in un-needed resolution; that said, it will remain useful to ensure having enough output pixels to meet the "native" resolution of the printer driver.Hi Mark, yes I agree scanning at 16 bit is a given for me. But I would also suggest that in the scenario I highlighted there is no un-needed resolution.
To recap in the case of the OP's print scans there is an unknown here that IMO needs testing. That is the actual limit of resolution before nothing useful is gained.
So for a print that may merit a hero treatment it is worth investigating the point of diminishing returns i.e. small area scan at 300, 600, 1200 ppi etc. stopping at the point where there is no noticeable/usable improvement.
I do not know the Epson 11000 XL but the review rate it quite highly even with its fairly modest optical resolution of 2400 ppi it seems that it can actually achieve 2170, pretty good when you look at other Epson models v600 to v800 achieving around 1500 - 2600 ppi approx. respectively and their claimed optical resolution much higher at around 6400 ppi.
So I would not expect to gain anything from a print at 2400 ppi but the fact that it can reach this level suggests a good acquisition system that may offer good things scanning reflective media.
Why would you want (and I advocate) to use a high resolution scan of a low resolution print?
To quote Ctein from his book on Digital Restoration:
"High resolution scans of low resolution prints can be useful when theres physical damage with clear edges. Scanning at a higher resolution spreads out real image detail over many more pixels while the edges of damaged areas remain pixel sharp. This makes it easier to use edge finding filters and similar tools in your image processing program to extract the boundaries of the damaged areas..."
Once these boundaries extracted into your mask (as in the last example) it can be fairly simple and quick to use tools to fill in the damaged information with good info from surrounding areas without affecting undamaged areas. My conclusion after handling quite a few of these nasties is that this is a useful tool to have in your bag of tricks
The same methods can be used with 35 mm negatives and transparancy and again the same scanning at optimum resolution which in the case of the Nikon LS 9000, the claimed scan resolution being 4000 ppi. Actual measured resolution indicates a very respectable 3900 ppi. So bearing in mind condition of slides may dictate going for optimum.
I do agree that the scanning of 9k slides a mammoth undertaking. Quick calc. suggesting IF you could scan continuously and that each scan took 2.5 minutes to complete AND change to the next slide that you would be busy for around 2 months working continuously 7 hours a day just to scan. Then add time spent editing, well..... :o
Post by: Alan Klein on November 24, 2017, 12:03:47 pm
Post by: Mark D Segal on November 24, 2017, 01:05:25 pm
Spotting dust off of 9000 slides!!! :o :-\ :-[
SilverFast iSRD or SRDx can make very fast work of that without individually spotting the slides.
Post by: Mark D Segal on November 24, 2017, 01:09:52 pm
Hi Mark, yes I agree scanning at 16 bit is a given for me. But I would also suggest that in the scenario I highlighted there is no un-needed resolution.
To recap in the case of the OP's print scans there is an unknown here that IMO needs testing. That is the actual limit of resolution before nothing useful is gained.
So for a print that may merit a hero treatment it is worth investigating the point of diminishing returns i.e. small area scan at 300, 600, 1200 ppi etc. stopping at the point where there is no noticeable/usable improvement.
I do not know the Epson 11000 XL but the review rate it quite highly even with its fairly modest optical resolution of 2400 ppi it seems that it can actually achieve 2170, pretty good when you look at other Epson models v600 to v800 achieving around 1500 - 2600 ppi approx. respectively and their claimed optical resolution much higher at around 6400 ppi.
So I would not expect to gain anything from a print at 2400 ppi but the fact that it can reach this level suggests a good acquisition system that may offer good things scanning reflective media.
Why would you want (and I advocate) to use a high resolution scan of a low resolution print?
To quote Ctein from his book on Digital Restoration:
"High resolution scans of low resolution prints can be useful when theres physical damage with clear edges. Scanning at a higher resolution spreads out real image detail over many more pixels while the edges of damaged areas remain pixel sharp. This makes it easier to use edge finding filters and similar tools in your image processing program to extract the boundaries of the damaged areas..."
Once these boundaries extracted into your mask (as in the last example) it can be fairly simple and quick to use tools to fill in the damaged information with good info from surrounding areas without affecting undamaged areas. My conclusion after handling quite a few of these nasties is that this is a useful tool to have in your bag of tricks
The same methods can be used with 35 mm negatives and transparancy and again the same scanning at optimum resolution which in the case of the Nikon LS 9000, the claimed scan resolution being 4000 ppi. Actual measured resolution indicates a very respectable 3900 ppi. So bearing in mind condition of slides may dictate going for optimum.
I do agree that the scanning of 9k slides a mammoth undertaking. Quick calc. suggesting IF you could scan continuously and that each scan took 2.5 minutes to complete AND change to the next slide that you would be busy for around 2 months working continuously 7 hours a day just to scan. Then add time spent editing, well..... :o
Ctein's advice is perhaps relevant to a situation the O/P may not have to deal with. Slides don't usually have sharp-edge damage. If higher resolution may be beneficial in certain special cases, so be it, but reserve the otherwise excess resolution for those situations.
For the generality of the O/P's task at hand I'll abide by the advice I provided in Post 15 on how to select an appropriate scanning resolution.
Post by: Alan Klein on November 24, 2017, 02:02:55 pm
SilverFast iSRD or SRDx can make very fast work of that without individually spotting the slides.ICE onEpsonscan slows scans down scout threefold
And it still leaves some shots that have to be spotted in post. I found it's faster withoutICE and just spot in post. But it all takes time regardless. Frankly, I can't imagine other scan programs spot better than Epsonscan. The infrared detection in the scanner physically works only so much regardless off the software.
Post by: Mark D Segal on November 24, 2017, 02:20:48 pm
ICE onEpsonscan slows scans down scout threefold
And it still leaves some shots that have to be spotted in post. I found it's faster withoutICE and just spot in post. But it all takes time regardless. Frankly, I can't imagine other scan programs spot better than Epsonscan. The infrared detection in the scanner physically works only so much regardless off the software.
SilverFast doesn't use ICE. The scanner provides an infra-red channel. The software does the rest to make use of it.
Instead of imagining what an application you may not know well enough does, try it so you'll have the facts from first-hand experience.
Post by: saiguy on November 24, 2017, 02:36:50 pm
File size bloat. This is why I'm keen on a small general utility out put. Back to thinking 5x7 again. They will probably put this on an average PC lap top. They know nothing about digital imaging. Could teach them how to print from LR.
Automation. Will batch scan 64bit HDRi. Long ago I used SF with ACC in batch mode. It sets a very good white point and usually no black point. So will open them one at a time in HDR. Will save to a new folder.
SF8-HDR work flow. The ACC sets a white point that I can rarely improve on. Have set Auto Prefs to give about a 242 level and check wether it is neutral. LR has intelligent exposure and I set this there. I manually set a neutral black point anywhere from mid 20's to low 30's. Increasing black in LR to taste. If I see color cast will try to mitigate with Neutral PIP, or Selective Color. Will take full advantage of iSDR if possible, or SRDx with due carefulness. Save to a new folder.
PS. Open the images via Bridge. Check for additional clean up, Neat Image with masking if warranted. Export to new folders that will be imported in LR as the master files.
LR. Do all tone/color work, capture sharpening, DAM, etc. Export the project files, 16bit tif, ARGB, sharpen for scree or print ??? to their new high quality HD and bake-in all editing.
They will put LR6 perpetual on probably a new dedicated PC lap top, and import the project files with a logical file structure which I can do from the excellent notes he gave me. Was lucky to get LR6 DVD from B&H photo for them.
5x7 at 300 ppi should be about 20 MB. Letter size about 60 MB. I think 60 MB files should be snappy on a new PC.
Seems to me a reasonable approach. And would rescan selects for printing using appropriate settings.
Post by: Alan Klein on November 24, 2017, 02:53:28 pm
SilverFast doesn't use ICE. The scanner provides an infra-red channel. The software does the rest to make use of it.Epsonscan ICE uses infrared to remove spots etc. That slows down the scan process a lot. What does Silverfast use and how does it work?
Instead of imagining what an application you may not know well enough does, try it so you'll have the facts from first-hand experience.
Post by: TonyW on November 24, 2017, 03:09:00 pm
.... If higher resolution may be beneficial in certain special cases, so be it, but reserve the otherwise excess resolution for those situations.I do not think we are in any general disagreement and broadly appear to be singing from the same hymn sheet albeit you in C major me in b flat major i.e maybe you thinking in best case scenario and me imagining worst case scenario where all images are poor ;) ;D
For the generality of the O/P's task at hand I'll abide by the advice I provided in Post 15 on how to select an appropriate scanning resolution.
As I have not revisited Silverfast for a good number of years (settling on Epsons own and Vuescan as being perfectly adequate for my needs and less costly) I am curious about the licencing and costs. Forgetting just for the moment the capabilities of the software and what it may bring to the table (I know it is a damn fine application!) in a particular case like the OP with two different means of scanning Epson and Nikon it is not clear to me if two seperate applications at Euro 336 required - it looks like it does making this pretty expensive compared to other options.
Post by: Mark D Segal on November 24, 2017, 06:03:41 pm
Mark. You are soo generous with your time and knowledge. I have read your book at least 2 times cover to cover, and reread sections many times. I follow your advise as best I can. Everything I know about SilverFast has come from you and Taz Tally. Have read your web site articles many times also.
File size bloat. This is why I'm keen on a small general utility out put. Back to thinking 5x7 again. They will probably put this on an average PC lap top. They know nothing about digital imaging. Could teach them how to print from LR.
Automation. Will batch scan 64bit HDRi. Long ago I used SF with ACC in batch mode. It sets a very good white point and usually no black point. So will open them one at a time in HDR. Will save to a new folder.
SF8-HDR work flow. The ACC sets a white point that I can rarely improve on. Have set Auto Prefs to give about a 242 level and check wether it is neutral. LR has intelligent exposure and I set this there. I manually set a neutral black point anywhere from mid 20's to low 30's. Increasing black in LR to taste. If I see color cast will try to mitigate with Neutral PIP, or Selective Color. Will take full advantage of iSDR if possible, or SRDx with due carefulness. Save to a new folder.
PS. Open the images via Bridge. Check for additional clean up, Neat Image with masking if warranted. Export to new folders that will be imported in LR as the master files.
LR. Do all tone/color work, capture sharpening, DAM, etc. Export the project files, 16bit tif, ARGB, sharpen for scree or print ??? to their new high quality HD and bake-in all editing.
They will put LR6 perpetual on probably a new dedicated PC lap top, and import the project files with a logical file structure which I can do from the excellent notes he gave me. Was lucky to get LR6 DVD from B&H photo for them.
5x7 at 300 ppi should be about 20 MB. Letter size about 60 MB. I think 60 MB files should be snappy on a new PC.
Seems to me a reasonable approach. And would rescan selects for printing using appropriate settings.
I think you are using too many applications. A combination of SF8 nd LR should be more than adequate to do everything you want to do and will save on time and workflow complexity.
If you are doing most of the scanning for viewing the output on a laptop display (not printing), you can indeed use 5*7 linear dimensions, but your output resolution need not exceed 100 PPI, which will reduce file size dramatically. Also after making all the image adjustments in 16 bit, you could convert them to 8 bit JPEGs to make them even smaller in size and more responsive for this kind of viewing. The downside is that you would need to rescan for anything they want printed.
Post by: Mark D Segal on November 24, 2017, 06:05:00 pm
Epsonscan ICE uses infrared to remove spots etc. That slows down the scan process a lot. What does Silverfast use and how does it work?
SilverFast uses their own iSRD and SRDx. Ample material about this on their website and my previous articles.
Post by: Mark D Segal on November 24, 2017, 06:07:38 pm
I do not think we are in any general disagreement and broadly appear to be singing from the same hymn sheet albeit you in C major me in b flat major i.e maybe you thinking in best case scenario and me imagining worst case scenario where all images are poor ;) ;D
As I have not revisited Silverfast for a good number of years (settling on Epsons own and Vuescan as being perfectly adequate for my needs and less costly) I am curious about the licencing and costs. Forgetting just for the moment the capabilities of the software and what it may bring to the table (I know it is a damn fine application!) in a particular case like the OP with two different means of scanning Epson and Nikon it is not clear to me if two seperate applications at Euro 336 required - it looks like it does making this pretty expensive compared to other options.
SilverFast and Vuescan have completely different marketing approaches and for the most part Vuescan is cheaper. I believe the OP has SilverFast already, why I did not hesitate to recommend using it.
Post by: saiguy on November 24, 2017, 07:13:50 pm
Lively discussions here. Yes, I have a fully implemented color managed WF. Have Kodachrome IT8 target from LSI. Use EIZO CS240 for judging color/tone adjustments. Although I only use the MBP Retina for SF. Doesn't play well on the EIZO, and work more by the numbers in SF anyway. My project slides are all the .9 x 1.4 variety.
Looked at some recent 35mm slides I prepared for print. HDR showed I used 838 zoom, 360 ppi. Gives out put size of 4046 x 2700, or 7.5 x 11.24 inch, 35.5 MB. Pretty sure this MB is after shedding the IR channel. Some I out put for 8x10 with with similar numbers.
Appreciate your advise about too many apps. My experience says iSRD doesn't catch everything, possibly even more so with Kodachrome. I am new to Neat Image, but so far I think it preserves details better than LR for noise reduction. May not even be an issue at all, or only on some. You have said Kodachrome has the least film grain of any film you have scanned.
I want to deliver good files, albeit smallish ones. Their lap top viewing will be the same more or less, not calibrated likely, and most web browsers will give a worst look with ARGB. Can not control that. But at least they can get decent small prints, unlike from 100 ppi 8bit jpeg files.
Any selects for printing would be rescanned. I might do the printing myself on an EP 3880. One problem is that I am near Chicago, they are in St. Louis, 5.5 hour drive.
Post by: Mark D Segal on November 24, 2017, 07:34:03 pm
Mark.
Lively discussions here. Yes, I have a fully implemented color managed WF. Have Kodachrome IT8 target from LSI. Use EIZO CS240 for judging color/tone adjustments. Although I only use the MBP Retina for SF. Doesn't play well on the EIZO, and work more by the numbers in SF anyway. My project slides are all the .9 x 1.4 variety.
Looked at some recent 35mm slides I prepared for print. HDR showed I used 838 zoom, 360 ppi. Gives out put size of 4046 x 2700, or 7.5 x 11.24 inch, 35.5 MB. Pretty sure this MB is after shedding the IR channel. Some I out put for 8x10 with with similar numbers.
Appreciate your advise about too many apps. My experience says iSRD doesn't catch everything, possibly even more so with Kodachrome. I am new to Neat Image, but so far I think it preserves details better than LR for noise reduction. May not even be an issue at all, or only on some. You have said Kodachrome has the least film grain of any film you have scanned.
I want to deliver good files, albeit smallish ones. Their lap top viewing will be the same more or less, not calibrated likely, and most web browsers will give a worst look with ARGB. Can not control that. But at least they can get decent small prints, unlike from 100 ppi 8bit jpeg files.
Any selects for printing would be rescanned. I might do the printing myself on an EP 3880. One problem is that I am near Chicago, they are in St. Louis, 5.5 hour drive.
iSRD works well with some versions of Kodachrome and not so well with others. For those they provide SRDx, the new tool that handles all Kodachromes as well as B&W, which neither iSRD nor ICE can deal with. How well iSRD works in general does depend on your settings for it, relative to the debris on the media you are cleaning up. It's a refined piece pf software allowing you to strike the balance between image detail and automatic debris clean-up. There are some very aggressive settings that can destroy detail unless you are a bit careful with it. But unlike ICE, it has a lot of flexibility/controls to get the targeting right.
Neat Image is a very good application for reducing the appearance of film grain, but LR's noise reduction is a very close second, and Topaz deNoise is also good. All of them can do a reasonably good job, bearing in mind that none of them are for film grain - they are all meant to deal with digital noise. Unlike those, SilverFast's grain reduction tool is designed for mitigating film grain and does a good job, but here again you want to be careful not to bake very aggressive settings into the scan, because it can impair detail.
Post by: saiguy on November 24, 2017, 08:45:42 pm
I wish I knew how to extract quotes from a previous post like you power users. Kevin should have a place for that on "how to use this forum". If it exists can you point to it.
You mention SF's grain reduction tool. Is that other than iSRD/SRDx? If so which tool is it? You say Kodachrome is the most grain free, so hopefully I don't need it anyway. My approach has been to use iSRD conservatively, and nearly avoid SRD/SRDx. That's why I am quite willing to go to PS to have a closer look and fine tune clean up there. I have also noticed it is less effective on scratches. And if there are mold issues, which won't be the case in my current project as they have been well stored, mold seems to be "in" the film and not "on" it. My understanding is that all debris is denser than film, and that is how IR can identify it. SRDx with pen masking sky areas is definitely what I will use, for iSDR also, if it is clearly called for.
Thanks for your continued attention to this thread.
Post by: Alan Klein on November 25, 2017, 01:09:36 am
Post by: Mark D Segal on November 25, 2017, 08:11:15 am
With flat bed scanner like the Epson, ICE infrared spot removal doesn't work well on Kodachrome and BW film because of the way the film is manufactured. It works OK on Ektachrome slides and color negatives. But it requires two scans which slows down the whole scanning process. It may effect the sharpness as well. ICE works better with dedicate scanners like the Nikon.
iSRD doesn't require two scans so it doesn't slow anything down. It also works well on some Kodachrome versions, less well on others, for which the alternative SRDx is a good solution. iSRD does not affect sharpness because its removal algorithm isn't a sweep of the whole image - it's like a sophisticated clone or healing operation focused only on the debris that is revealed by the IR channel and earmarked for removal automatically in the software according to user settings. Tools such as iSRD and ICE can only work with scanners that actually have an IR channel.
Post by: Mark D Segal on November 25, 2017, 08:37:24 am
Mark,
I wish I knew how to extract quotes from a previous post like you power users. Kevin should have a place for that on "how to use this forum". If it exists can you point to it.
You mention SF's grain reduction tool. Is that other than iSRD/SRDx? If so which tool is it? You say Kodachrome is the most grain free, so hopefully I don't need it anyway. My approach has been to use iSRD conservatively, and nearly avoid SRD/SRDx. That's why I am quite willing to go to PS to have a closer look and fine tune clean up there. I have also noticed it is less effective on scratches. And if there are mold issues, which won't be the case in my current project as they have been well stored, mold seems to be "in" the film and not "on" it. My understanding is that all debris is denser than film, and that is how IR can identify it. SRDx with pen masking sky areas is definitely what I will use, for iSDR also, if it is clearly called for.
Thanks for your continued attention to this thread.
Use the "Quote" button on top of the post for extracting the whole post to be quoted. Not sure how to do it extracting parts of a post. Those who do should chime in.
Grain reduction is a separate tool from iSRD/SRDx. When you put the software into Transparency mode you will see a GANE tool in the vertical toolbar between AACO and ME. It's the GANE tool for grain mitigation.
You don't need to go to PS for manually eliminating debris. Lightroom's spot healing/cloning does a fine job of this if you are using Lightroom for other purposes, (such as creating the web galleries, making prints, general photo organization and editing).
iSRD can be very effective with scratches, depending on the tool settings you select.
Post by: Slobodan Blagojevic on November 25, 2017, 08:49:33 am
Use the "Quote" button on top of the post for extracting the whole post to be quoted. Not sure how to do it extracting parts of a post...
You quote all, then delete the unnecessary.
Post by: Mark D Segal on November 25, 2017, 09:02:08 am
Post by: saiguy on November 25, 2017, 09:49:26 am
All of these place a long string of "computer speak." above and below the actual post.
Yes I could delete the useless strings. Even when I high light the text I get the same result.
Post by: TonyW on November 25, 2017, 11:17:07 am
Quote
Slobodan. In the thread view there is a quote button at the far top right. In the reply text box view there is a quote button above the smily guy with sun glasses. In the Topic Summary there is an Insert Quote to the right of Posted by:
All of these place a long string of "computer speak." above and below the actual post.
Yes I could delete the useless strings. Even when I high light the text I get the same result.
Highlight the bits you want between the beginning of the quote "[quote author***]" and end of the quote "[/quote]"
To end up with something like this:
...All of these place a long string of "computer speak." above and below the actual post...
The message string will look similar to this (replace quite with quote):
[quite author=saiguy link=topic=121772.msg1013078#msg1013078 date=1511621366]...All of these place a long string of "computer speak." above and below the actual post...[/quite]
Post by: Slobodan Blagojevic on November 25, 2017, 11:50:09 am
Slobodan. In the thread view there is a quote button at the far top right. In the reply text box view there is a quote button above the smily guy with sun glasses. In the Topic Summary there is an Insert Quote to the right of Posted by:
All of these place a long string of "computer speak." above and below the actual post...
Let me illustrate the three different uses of "quote":
#1 "In the thread view there is a quote button at the far top right"
Clicking that button simply quotes the whole text. Then you can delete unnecessary strings, as I did above.
#2 "In the reply text box view there is a quote button above the smily guy with sun glasses"
I use that one when I am quoting something else, other than a forum member's post, e.g., a quote from a newspaper, or a favorite motto (this one is from a very smart guy)
Quote
When everybody thinks the same... nobody thinks.
#3 "In the Topic Summary there is an Insert Quote to the right of Posted by"
This one is useful if you want to quote somebody else's sentence from the same thread, for instance:
Beautiful....why didn't I think of that?!?! :-)
The "computer speak" you talk about is only visible while you are composing a reply. To see the final version before posting, you click on the "preview" button.
Post by: Alskoj on November 25, 2017, 11:52:31 am
Thanks Alskoj, No drum scans. Have Nikon LS 9000 and Epson Expression 11000 XL.I wish you much success! Happy Thanksgiving!
Thanks Mark, Will read chapter 5 again.
Happy Thanksgiving
Post by: saiguy on November 25, 2017, 12:15:06 pm
Thank you Slobodan; 3 ways to skin a cat.
Post by: saiguy on November 25, 2017, 12:18:23 pm
The "computer speak" you talk about is only visible while you are composing a reply.
Post by: saiguy on November 25, 2017, 12:22:21 pm
This one is useful if you want to quote somebody else's sentence from the same thread, for instance:
Guess I'm too dense to follow well written instructions
Post by: Slobodan Blagojevic on November 25, 2017, 12:30:33 pm
Post by: saiguy on November 25, 2017, 12:32:49 pm
Post by: saiguy on November 25, 2017, 12:38:04 pm
Ok, let me illustrate:
another try
Post by: Slobodan Blagojevic on November 25, 2017, 12:43:19 pm
Slobodan. 3 tries 3 failures. Used method 1. I get a quote box above the reply text entry. . Same text is in the reply text box and can be edited, shortened. I do not know how to exit from there and comment on the quote. The quote itself does not stick.
Let's try step by step:
Step 1 is just pressing the quote button from method 1.
Step 2 and 3 show you CAN edit at that stage: step 2 highlights a sentence, which is then deleted in step 3.
Now, one has to remember that we are in a "quick" reply mode, therefore not all editing options are available. For that, we need to click on the Preview button and then have the range of options in the form of icons.
Post by: saiguy on November 25, 2017, 12:44:14 pm
At least you thought me how to skin a cat one way. The 3rd method. Seems to be the best method as you can grab from posts even days earlier.
Not very intuitive though
Post by: saiguy on November 25, 2017, 12:53:18 pm
Step 1 is just pressing the quote button from method 1.
Step 1 attempt,
Post by: Slobodan Blagojevic on November 25, 2017, 12:55:43 pm
Step 1 is just pressing the quote button from method 1.
Step 1 attempt,
This would be the result of the proper "method 1, step 1." Click "Quote," click "Post." I am not sure what exactly you are doing.
Post by: Mark D Segal on November 25, 2017, 01:01:27 pm
This would be the result of the proper "method 1, step 1." Click "Quote," click "Post." I am not sure what exactly you are doing.
OK, here's your full quote just above by pressing the "Quote" button.
Then I want to make it selective by deleting the last sentence, so I am only quoting the first sentence. I copy-paste the whole quote from above, then go in there, highlight the sentence I don't want to keep, and click delete on my keyboard:
This would be the result of the proper "method 1, step 1.
This works.
Post by: saiguy on November 25, 2017, 01:25:27 pm
This would be the result of the proper "method 1, step 1." Click "Quote," click "Post." I am not sure what exactly you are doing.I hit the quote button. am now typing in the text box starting where the computer speak ends. Will what happens. Mark's copy/paste technique is some advanced thing.
Post by: saiguy on November 25, 2017, 01:29:22 pm
Thank you both for caring, sharing, and patience.
Post by: saiguy on November 25, 2017, 08:45:30 pm
Post by: CeeVee on December 16, 2017, 06:06:50 am
Sent from my Pixel XL using Tapatalk
Post by: saiguy on December 16, 2017, 05:56:51 pm
The B&W thing was really a question about scans from B&W prints or from off set print originals.
Post by: CeeVee on December 21, 2017, 08:06:52 pm
CeeVee, Well this thread is getting old by now. I started the scan project. There are about 9k Kodachrome slides. Have batch raw scanned 1230 of them to SF8-HDR so far. Am using 5X7 out put scaling.Offset print originals?
The B&W thing was really a question about scans from B&W prints or from off set print originals.
I'd scan prints on a flatbed (we have Epson 750) and the chromes on a film scanner. I've used Nikon 9000 and Hasselblad X-5.
If you're really scanning copies off an offset press you'll be dealing with de-screening. Lots O' fun.
Sent from my Pixel XL using Tapatalk
Post by: Mark D Segal on December 21, 2017, 09:04:07 pm
.................
If you're really scanning copies off an offset press you'll be dealing with de-screening. Lots O' fun.
..............
Not really too much of a problem. SilverFast has a de-screening tool that can work pretty well with carefully chosen settings, the objective being to achieve the optimum possible balance between image detail and de-screening. (This is covered in Section 6.6 of my book.)
Post by: Garnick on December 22, 2017, 08:03:56 am
Not really too much of a problem. SilverFast has a de-screening tool that can work pretty well with carefully chosen settings, the objective being to achieve the optimum possible balance between image detail and de-screening. (This is covered in Section 6.6 of my book.)
Yes, the de-screening tool does work pretty well, with a bit of tweaking and patience. Another approach, and perhaps better in some cases, is the FFT Filter used in Photoshop or perhaps other apps. This is the "Frequency Separation" filter that can fix a number of issues, including screening. I've also used it to subdue the Moire effect and it does a great job. it can also be used when retouching certain skin problems with portrait work. I've seen it demonstrated, but haven't had need to use for that application myself. Google "FFT Filter" and you'll get all the information you need concerning this filter and how to download it as a plugin for Photoshop.
Another very useful application for this filter is when restoring old B&W prints on the ubiquitous pebble textured paper, which is almost impossible to fix otherwise. With a bit of finessing, the FFT Filter will do a great job in such situations. You will also find examples of that on Google.
Gary
Post by: saiguy on December 22, 2017, 12:39:09 pm
FFT Filter. Sounds interesting especially with "pebble paper". Will look into that.
thanks,
Post by: saiguy on December 22, 2017, 09:54:30 pm
Been looking at some videos on the FFT Filter. All of them a quite dated. Can the tool be gotten and used on current PS Submission Plan. Where would I look to get it?
Andy Anderson has a DVD on Infinite Skills where he shows to scan a paper textured photo 4 times rotating it 90 degrees each time and using a blending mode that works well. Photo Restoration With Photoshop. Has more to do with how the light bounces off the texture than the pattern itself. But this FFT looks much better.
Thanks for the "new discovery" Hope the plug in is still available.
Post by: TonyW on December 23, 2017, 05:31:44 am
You are aware that Silverfast uses their own version of FFT for descreening?
That filter does not seem to be freely available/supported any longer although it will work with the new versions of PS. It also had a drawback of file size limitations also if you are a Mac user you are out of luck
I would suggest that you have a look at a program called ImageJ available Windows and Mac. This is a stand-alone app only so once finished you will need to bring the result into PS
https://imagej.nih.gov/ij/download.html
Post by: Bart_van_der_Wolf on December 23, 2017, 08:27:41 am
The Fast Fourier Transform (FFT and IFFT) that most of us used was probably the one published by Alex Chirokov as a PS plugin. I believe it’s original photographic use in Forensics.
You are aware that Silverfast uses their own version of FFT for descreening?
That filter does not seem to be freely available/supported any longer although it will work with the new versions of PS. It also had a drawback of file size limitations also if you are a Mac user you are out of luck
I would suggest that you have a look at a program called ImageJ available Windows and Mac. This is a stand-alone app only so once finished you will need to bring the result into PS
https://imagej.nih.gov/ij/download.html
FWIW Affinity Photo also has built-in FFT denoise capabilities. Although it can work very very well, there can still be some issues (mostly due to input image quality) with some images.
They also have a tutorial that demonstrates the functionality:
https://vimeo.com/161180581 (https://vimeo.com/161180581)
Cheers,
Bart
Post by: TonyW on December 23, 2017, 10:47:18 am
FWIW Affinity Photo also has built-in FFT denoise capabilities. Although it can work very very well, there can still be some issues (mostly due to input image quality) with some images.That interesting Bart and looks to be a nice implementation of FFT. Showing the results of automatic application of IFFT in live mode looks pretty good too.
They also have a tutorial that demonstrates the functionality:
https://vimeo.com/161180581 (https://vimeo.com/161180581)
Cheers,
Bart
Post by: Garnick on December 23, 2017, 11:33:21 am
Garnick,
Been looking at some videos on the FFT Filter. All of them a quite dated. Can the tool be gotten and used on current PS Submission Plan. Where would I look to get it?
Andy Anderson has a DVD on Infinite Skills where he shows to scan a paper textured photo 4 times rotating it 90 degrees each time and using a blending mode that works well. Photo Restoration With Photoshop. Has more to do with how the light bounces off the texture than the pattern itself. But this FFT looks much better.
Thanks for the "new discovery" Hope the plug in is still available.
Are you on Mac or Windows? I can send the Mac version if you want to try it. It will work on almost any version of PS as a Plugin. This is the original download site - http://www.djjoofa.com/download, but they are no longer supporting FFT.
Post by: saiguy on December 24, 2017, 09:53:21 am
Thanks for the offer to send me Mac FFT. Will test a textured photo scan with SilverFast De-screening first.
Affinity Photo videos were very impressive.
Post by: Bart_van_der_Wolf on December 24, 2017, 10:41:52 am
Affinity Photo videos were very impressive.
Yes, their implementation of a well-known solution toward the removal of raster (or other repetitive) patterns, is quite nice (with adjustable preview). Most of the controls and functions are very well implemented, and it helps that AP works with floating point precision and makes use of GPU processing for many of its internal operations.
It also works on selections, so one could, for instance, select bright regions like sky and cloud, or shadow regions, or colors, or subjects, and optimize the "denoising" for those specific areas of interest. It also handles different raster frequencies or rotations by individual color channel, so one does have a reasonable chance to fill in the blanks (even in problem cases).
Cheers,
Bart