michael.panoff
Established
I'm curious about your vuescan settings / workflow for B&W. Anyone care to share. Please indicate what scanner you are using.
In particular I'm curious if you scan to RAW and do level / curve adjustments in Photoshop. Is the gamma supposed to be way out of wack with the RAW scans? Personally I'm finding the film profiles do a pretty good job of isolating the histogram, ends up being much less work too.
How about Long Exposure and multi-pass? I have a Nikon Coolscan V, and assume that the scanner already does an internal multi-scan.. any sense in having the software do it?
Do you scan color negative or B&W negative?
In particular I'm curious if you scan to RAW and do level / curve adjustments in Photoshop. Is the gamma supposed to be way out of wack with the RAW scans? Personally I'm finding the film profiles do a pretty good job of isolating the histogram, ends up being much less work too.
How about Long Exposure and multi-pass? I have a Nikon Coolscan V, and assume that the scanner already does an internal multi-scan.. any sense in having the software do it?
Do you scan color negative or B&W negative?
kaiyen
local man of mystery
Michael,
I really got to get this up on the wiki that I once claimed I would do...
the RAW files from Vuescan are not intended to be edited directly in PS. The idea is that you scan once, then can output with different settings over and over from the one RAW file. So you use Vuescan to work with RAW Vuescan files.
The scanner won't "auto-multi-pass." If you're using vuescan, you need to tell it to do it. AFAIK, anyway. You can tell if it's doing a multi-pass - does it pass over it more than once?
I only use long exposure with dense negatives (overdeveloper) and slides. I haven't found that it does a huge difference, but it hasnt' hurt.
allan
I really got to get this up on the wiki that I once claimed I would do...
the RAW files from Vuescan are not intended to be edited directly in PS. The idea is that you scan once, then can output with different settings over and over from the one RAW file. So you use Vuescan to work with RAW Vuescan files.
The scanner won't "auto-multi-pass." If you're using vuescan, you need to tell it to do it. AFAIK, anyway. You can tell if it's doing a multi-pass - does it pass over it more than once?
I only use long exposure with dense negatives (overdeveloper) and slides. I haven't found that it does a huge difference, but it hasnt' hurt.
allan
michael.panoff
Established
Allan,
Yes, when I tell vuescan to multi-pass, it physically scans the negative however many times indicated. On the other hand, I think I read something on photo.net that the scanner CCD hardware does an internal multi-scanning.. after digging around the Nikon site I can't find any mention of such a thing. I guess there seems to be some confusion as to whether a scanner can support multi-scanning, in other words having the ability to exactly reposition the CCD. The Nikon hasn't had issues wth this.
So what are your thoughts on multi-scanning? Is it worth it? How many passes do you typically do?
How about the histogram adjustments, do you adjust the black and white points in vuescan? or photoshop?
Yes, when I tell vuescan to multi-pass, it physically scans the negative however many times indicated. On the other hand, I think I read something on photo.net that the scanner CCD hardware does an internal multi-scanning.. after digging around the Nikon site I can't find any mention of such a thing. I guess there seems to be some confusion as to whether a scanner can support multi-scanning, in other words having the ability to exactly reposition the CCD. The Nikon hasn't had issues wth this.
So what are your thoughts on multi-scanning? Is it worth it? How many passes do you typically do?
How about the histogram adjustments, do you adjust the black and white points in vuescan? or photoshop?
Last edited:
vicmortelmans
Well-known
michael.panoff said:
Hi,
I do b&w scanning of complete rolls using Vuescan on a Microtek Filmscan 3600. I limit myself to unattended options. After some experiments, I found that most of the auto-functions don't improve the result enough to account for the extra scanning time they take.
This is a summary of what I use (by heart):
- manual exposure ('2' suits me fine, but that value may be scanner-dependent, I guess)
- fixed cropping (I use 35mm slide, since it's a bit smaller than 35mm negative, so It doesn't cause black borders; I don't mind to loose a small edge of the picture)
- fixed focus (I use to manually focus once before starting the roll)
- single pass (not only because it's faster, but I didn't see any quality gain in multiple pass)
- no post-processing what-so-ever
- 16 bit grayscale raw output
I'm doing some low-level programmed post-processing myself, but if you're not in to that sort of stuff, the best approach would be to load the raw file into photoshop (my version of the GIMP doesn't do 16 bit---a real pity!) and use the curves tool.
1. Drag the left bottom point of the curve to the top and move it to the right until it aligns with the start of your histogram
2. drag the right top point of the curve to the bottom and move it to the left until if aligns with the end of your histogram
The result is now a positive, and dynamic range of the input data is mapped to the full dynamic range of the output. But the contrast is totally wrong. That's because you're working with linear pixel data, while the actual 'brightness information' it's representing, is logarithmic.
3. click somewhere in the middle of the curve to create an extra point and drag that down. The line becomes a curve. By moving it up and down and right and left, and looking at the preview, the optimum result will be easily found.
This way, you don't need gamma at all!
Groeten,
Vic
A related question:
It may be nothing, but often when performing this procedure, I noticed that the histogram itself shows at least on part of it's range, a curved downward slope. When I apply the steps 1-2-3 as described above, the curve's optimum position tends to approach the histogram's curve very close! Can this be explained, or is it just coincidence? If so, this might be a useful property which can be used to automate post-processing!
vicmortelmans
Well-known
kaiyen said:
OK, that's a useful Vuescan feature, but I don't see any reason why you wouldn't be able to do your own post-processing on the raw Vuescan files. They should contain the pixel data as received from the CCD and noting more. So certainly nothing Vuescan-specific in it...
kaiyen said:
I also tried this on my overdeveloped negatives, but the problem is that it blows out the highlights (= image shadows)... I didn't succeed in getting a decent scan from the overdeveloped negatives. Either the shadows loose all detail, or the highlights are cluttered with noise. So for me, it did make a difference, but the quality I gain on the one end, is lost on the other.
Groeten,
Vic
michael.panoff
Established
Thanks for the feedback Vic.
interesting.. so the RAW historgram is supposed to have values shoved all the over to one side, at least for B&W negative film? When I first saw this, I for sure thought something was wrong.
I'm getting pretty good results letting vuescan handle isolating the historgram.. but on some tests tonight I did notice that vuescan wasn't necessarily that much better than Nikon Scan. It looked like vuescan extracted slightly more detail, but the auto-exposure setting on the Nikon software did a pretty good job handling the histogram.
I will have to experiment some more on multi-pass vs single pass.
interesting.. so the RAW historgram is supposed to have values shoved all the over to one side, at least for B&W negative film? When I first saw this, I for sure thought something was wrong.
I'm getting pretty good results letting vuescan handle isolating the historgram.. but on some tests tonight I did notice that vuescan wasn't necessarily that much better than Nikon Scan. It looked like vuescan extracted slightly more detail, but the auto-exposure setting on the Nikon software did a pretty good job handling the histogram.
I will have to experiment some more on multi-pass vs single pass.
hth
Well-known
I am currently working on getting the hang of this.
I have a Minolta Scan Multi pro attached to a Mac and have used Vuescan, though lately I have tried also the supplied software.
I output on an Epson 2100 with MIS UT inks. I had a dreadful time geting the greyscale output any good with PS Elements, so I obtained a 30 day evaluation of CS2 and it solved all my problems. I use a densitometer to give QTR feedback and it comes out fine now.
I have just started to experiment with real output after getting the printer right. It did not look all that good as on the monitor (which is color calibrated, though I did that a couple of years ago).
With a B&W photo after using curves to make it look good, I noticed that slightly shadowed asian skin tones was at 50% which comes out too dark for my taste (this was from an 8-bit Vuescan). I rescan with supplied software, using 16 bit this time and and it came down a bit. I adjusted it down with curves to around 40% (slightly shadowed areas) and around 20% (in sunlit areas) and it printed much better.
I often find that I like the look I get after doing auto levels (sometimes this worsen it, but most of times it is good), then making adjustments with curves, especially making a point in the lower left range and dragging it slightly veritcally to create a curve, looking at the picture at the same time.
At the moment I am not happy that it looks so good on the monitor (skin at around 50%) and it comes out so bad. I do not really understand what the problem is, anyone knows?
Otherwise I will expeiment further with some other photos, adjustments and try different lighting for viewing the print to see if I can get the hang of it.
/Håkan
I have a Minolta Scan Multi pro attached to a Mac and have used Vuescan, though lately I have tried also the supplied software.
I output on an Epson 2100 with MIS UT inks. I had a dreadful time geting the greyscale output any good with PS Elements, so I obtained a 30 day evaluation of CS2 and it solved all my problems. I use a densitometer to give QTR feedback and it comes out fine now.
I have just started to experiment with real output after getting the printer right. It did not look all that good as on the monitor (which is color calibrated, though I did that a couple of years ago).
With a B&W photo after using curves to make it look good, I noticed that slightly shadowed asian skin tones was at 50% which comes out too dark for my taste (this was from an 8-bit Vuescan). I rescan with supplied software, using 16 bit this time and and it came down a bit. I adjusted it down with curves to around 40% (slightly shadowed areas) and around 20% (in sunlit areas) and it printed much better.
I often find that I like the look I get after doing auto levels (sometimes this worsen it, but most of times it is good), then making adjustments with curves, especially making a point in the lower left range and dragging it slightly veritcally to create a curve, looking at the picture at the same time.
At the moment I am not happy that it looks so good on the monitor (skin at around 50%) and it comes out so bad. I do not really understand what the problem is, anyone knows?
Otherwise I will expeiment further with some other photos, adjustments and try different lighting for viewing the print to see if I can get the hang of it.
/Håkan
michael.panoff
Established
Are you doing everything in the same profile.. sRGB? If you are on a PC, did you run the Adobe Gamma program in your control panel? On the mac go to system preferences -> monitor -> color -> calibrate. FYI, I use a 2.2 gamma, even on the mac.
I haven't done any printing recently, since I've moved.. but last I remember, my monitor calibration was off just a bit, and my prints looked washed out.
I haven't done any printing recently, since I've moved.. but last I remember, my monitor calibration was off just a bit, and my prints looked washed out.
michael.panoff
Established
and in another random tangent.. I found that using a dedicated noise removal tool, like Neat Image, does a superior job at grain removal, compared to scanning software.
kaiyen
local man of mystery
warning - I haven't read all the way through the thread yet. Just responding to a direct response to my...response. (tired this morning)
I have not had an issue with my IV in terms of softere scans with multi-pass, indicating that it cannot position the CCD correctly. The film is held in place just fine becuase I use the feeder, not the holder.
I do 2 passes. I did a test with 2, 8, and 16 and could not find a difference between them with most of my negatives. If you process them right, you probably won't need the extra passes.
I have vuescan do auto-levels, and I set the profile to tmax-400 ci: 55 as a starting point. I adjust the CI to get the entire histogram on there, and I double check that the auto level black/white points are right. Even if it looks like you're right at the edge of clipping in VS, when you take it into PS you'll notice that you actually have more room on either end than you would've thought. In other words, VS is pretty conservative when it comes to setting black and white points with auto-levels.
allan
michael.panoff said:
I have not had an issue with my IV in terms of softere scans with multi-pass, indicating that it cannot position the CCD correctly. The film is held in place just fine becuase I use the feeder, not the holder.
I do 2 passes. I did a test with 2, 8, and 16 and could not find a difference between them with most of my negatives. If you process them right, you probably won't need the extra passes.
I have vuescan do auto-levels, and I set the profile to tmax-400 ci: 55 as a starting point. I adjust the CI to get the entire histogram on there, and I double check that the auto level black/white points are right. Even if it looks like you're right at the edge of clipping in VS, when you take it into PS you'll notice that you actually have more room on either end than you would've thought. In other words, VS is pretty conservative when it comes to setting black and white points with auto-levels.
allan
kaiyen
local man of mystery
vicmortelmans said:
Yeah, it's a weird use of the word "RAW." But that is indeed what it is intended for. And trying to get a full tonal range from the file in PS is pretty hard from my experiments. I scan to TIFF and RAW all the time, and go back to the RAW file as needed, in VS.
Not sure what that means. highlights = image shadows?
When I've done long pass, it has retained the shadow detail as is, and helped salvage the highlights. So instead of total noise I get just blown out highlights
. I guess that's just a difference in how our scanners work.allan
kaiyen
local man of mystery
michael.panoff said:
NS does okay. But sometimes you'll notice that it clips the histogram a bit more tightly, and sometimes a lot more tightly. I don't want to lose anything at all prior to doing work in PS, so that matters a lot to me. I also believe that NS does some processing of its own - notably sharpening - that you can't disable. I think.
allan
Last edited:
kaiyen
local man of mystery
hth said:
comparing a 16 bit file with an 8 bit one is kind of tough...
This is standard procedure for me in PS. Double check levels. Dodge and burn (though usually barely at all since my development is pretty dialed in), then an S-Curve. I do more creative stuff from there.
Years ia WAAAY too long between calibrations. I calibratae every week. Or at least I try to.
allan
hth
Well-known
I use a Mac. The monitor was profiled 3 years ago, so I remade the calibration now using an eye-on. The new profile makes for a more neutral greyscale, intensity has not changed though. This is a profile where I selected gamma 1.8.
Grey scale documents are assigned to greyscale with dot gain 20% in CS2, I switched now to grey gamma 1.8 (but I have not printed anything yet).
The printer uses a custom profile with a greyscale output that looks good and matches density tables quite nicely.
If I click on a slightly shadowed face with curves open, I get around 40% if I point to it. Judging from output on monitor and printer, I would say I trust the printer to be the one with more accurate output.
There are so many profiles here. The document has one, the display has one and the printer one. I can convert to using another profile for the picture, but that does not change it visually.
I do not think I have to use the same profile. The monitor profile should convert this to something that looks right on the monitor, while the printer should do the same for the printer. Neither should need to be the same as the profile used for the picture.
I can adjust down brightness on the monitor, it is at the highest setting at the moment. The monitor calibration seems to make the color balance right, but does it do the same for brightness? It does display various grey/black pages (in additon to the color ones) during calibration... Why is it so off? I do not understand.
/Håkan
Grey scale documents are assigned to greyscale with dot gain 20% in CS2, I switched now to grey gamma 1.8 (but I have not printed anything yet).
The printer uses a custom profile with a greyscale output that looks good and matches density tables quite nicely.
If I click on a slightly shadowed face with curves open, I get around 40% if I point to it. Judging from output on monitor and printer, I would say I trust the printer to be the one with more accurate output.
There are so many profiles here. The document has one, the display has one and the printer one. I can convert to using another profile for the picture, but that does not change it visually.
I do not think I have to use the same profile. The monitor profile should convert this to something that looks right on the monitor, while the printer should do the same for the printer. Neither should need to be the same as the profile used for the picture.
I can adjust down brightness on the monitor, it is at the highest setting at the moment. The monitor calibration seems to make the color balance right, but does it do the same for brightness? It does display various grey/black pages (in additon to the color ones) during calibration... Why is it so off? I do not understand.
/Håkan
michael.panoff said:
hth
Well-known
kaiyen said:
Yes, I happened to have the 8 bit file as I could not use 16 bit in any good way with Elements. As I am trying CS2 now it was an obvious first thing to try. It did improve things a little, but did not solve the problem with monitor output. See my other post above.
This CS2 is quite expensive (which is why used Elements), but it solved all my printing problems immediately, so I guess I at some point have to find a way to afford it. (Maybe I should take the money from my small M8 fund as part payment..
)CS2 is 1.5 years now BTW, is there any new version on the horizon? Elements 4 has a nice things that are not in CS2. I think I would hate having to pay for an upgrade a couple of months later..
kaiyen said:
I have mostly use a "half S-curve" so far, will probably do more on that when I can trust what I see more...
kaiyen said:
Yes, I realize that, I will definitely do it more often from now.
/Håkan
kaiyen
local man of mystery
hth said:
as there should be. the file's data stays the same, the profiles translate from file to the color gamut of each output device (monitor vs. printer, etc).
It _should_, yes. My Eye-One makes me turn the brightness to 100, then dial it down to the right level. Then the contrast to 100, then again dial down. That's calibration. I do the profiling after that.
CS3 is supposedly coming out in December-ish. I think they are working on the Universal Binary version now for the Mac side, so they can release both at the same time.
allan
hth
Well-known
I use a cinema LCD. The Display preferences allow me to set the brightness. There are no controls on my LCD display. As far as I can see, my Eye-One Display (previous edition to what they sell today) ignores the amount of light emitted completely when it creates profiles.
I have googled around and from what little I have found it might be that this LCD is much brighter than a CRT. Someone suggested adjusting it down, quoting some luminance numbers I have no idea how to measure. Another suggested turning it down to about half output.
Turning brightness down to half does give a reasonable interpretation on screen of how the print will look like. However, it is not easy at all to compare a bright lit LCD with reflective paper. It is like comparing slides on a lightboard to a print, they do not really compare the same day. A slide picture simply comes to life and blows a print away. Having said that, the print I managed to get pleases me a lot, but it is not the same.
The LCD being backlit is simply much more forgiving as far as I can see.
Kind of frustrating, but I should have understood that this would be the case (as I have experience with slides on a lightboard before). I guess I have to get used to interpreting the screen by making lots of prints and learn how to judge and compare. Kind of a guessing game that with experience will be less of guessing... The percentage shown in Levels is a good help and I guess I should spend more time looking at histograms too.
Hmmm, CS3 around December. I wonder how I will be able to survive. Maybe I should look around for other cheaper alternatives. I wish gimp would be usable...
/Håkan
I have googled around and from what little I have found it might be that this LCD is much brighter than a CRT. Someone suggested adjusting it down, quoting some luminance numbers I have no idea how to measure. Another suggested turning it down to about half output.
Turning brightness down to half does give a reasonable interpretation on screen of how the print will look like. However, it is not easy at all to compare a bright lit LCD with reflective paper. It is like comparing slides on a lightboard to a print, they do not really compare the same day. A slide picture simply comes to life and blows a print away. Having said that, the print I managed to get pleases me a lot, but it is not the same.
The LCD being backlit is simply much more forgiving as far as I can see.
Kind of frustrating, but I should have understood that this would be the case (as I have experience with slides on a lightboard before). I guess I have to get used to interpreting the screen by making lots of prints and learn how to judge and compare. Kind of a guessing game that with experience will be less of guessing... The percentage shown in Levels is a good help and I guess I should spend more time looking at histograms too.
Hmmm, CS3 around December. I wonder how I will be able to survive. Maybe I should look around for other cheaper alternatives. I wish gimp would be usable...
/Håkan
kaiyen said:
Pherdinand
the snow must go on
Great thread.
I have a question: I downloaded the trial VUescan program. Options like "advanced" colour (colour management) are switched off. The program saves everything in...AppleRGB space!!! I mean, why??? It's really a weird colour space, i wish it would use stg nicer like adobe or maybe the scanner's basic working space, not this ...thing. OK, i know, i use a Mac so maybe the software "thinks" he has to pick apple rgb?
anyway. Is the basic "full" version doing the same? Is it only possible to change this in the "pro" version?
thanks
I have a question: I downloaded the trial VUescan program. Options like "advanced" colour (colour management) are switched off. The program saves everything in...AppleRGB space!!! I mean, why??? It's really a weird colour space, i wish it would use stg nicer like adobe or maybe the scanner's basic working space, not this ...thing. OK, i know, i use a Mac so maybe the software "thinks" he has to pick apple rgb?
anyway. Is the basic "full" version doing the same? Is it only possible to change this in the "pro" version?
thanks
Pherdinand
the snow must go on
By the way, multi-scanning question: As far as i understood, you can also select a single-pass multi-read option, when the ccd value is read uyp to 4 times without multiple scanning the image. The problem with multiple scanning image is, at least with flatbed scanners, that it is really difficult to scan exactly the same region many times. Maybe with film scanners where the physical movements are much smaller, it is piece of cake.
vicmortelmans
Well-known
kaiyen said:
Quite simple. Since you scan a negative, the highlights as perceived by the scanner are actually the shadows in your final image. By increasing the scanner exposure, detail in the image schadow area's will be lost.
Groeten,
Vic