Thatspec
Established
So... after I've done my scan (Vuescan/ls-5000/Tmax100), the saved .tif pops up (OSX 10.6.8 preview) as it should. What I can't figure out is... what is this picture that is now in the preview pane of viewscan?! It is markedly better than the saved .tif (when viewed at the same size) albeit significantly smaller (becomes pixelated when you zoom in). I've yet to get a saved .tiff to look as good in LR and have taken to just taking screen shots of the preview pane. They're small but they look great. Have used all the settings Chris uses; http://chriscrawfordphoto.com/technical/scanning.php, I just want a full size version of the preview pane :bang:
These are 100% crops, which is better to your eye?
These are 100% crops, which is better to your eye?
cabbiinc
Slightly Irregular
A few things I don't care for in that tutorial. Before I tell you please accept this as just my opinion and I'm not claiming to be an expert.
- Media, he tells you to always scan in greyscale. I'd say do that first, and if that doesn't work try RGB. The reason why can be found on Hamrick's website:
To state one way is correct and the other is incorrect without first clarifying what you're scanning in the first place can be right some times and wrong others.
- 16 bit grey, again if that works for you great, if not then maybe you should try 48 bit rgb. To let the program convert to grayscale then take it into post production and try to tweak the levels seams a bit like shooting in a digital camera in jpeg then trying to do extensive levels adjustments. Why would you when you could just get the whole spectrum to work with? I do agree that 8 bit is usually not worth it unless you like exactly what's coming out of the scanner and you plan to just leave it at that.
-Preview Resolution, Auto should work for any dedicated film scanner. It calculates the preview to be roughly 4mp. On a flatbed 4mp is rather low quality and doesn't give enough information to calculate what the in scanner adjustments are going to be, and all adjustments are made with the preview scan. So if your "exposure" in the scanner is drastically different from the preview increasing that dpi can usually fix that. If that has no effect then put it back to Auto.
- Auto Focus, I so no reason to not set focus to Scan. After all, you're not capturing the scan from the preview so sharpness usually isn't an issue. It just makes the scanner wait to focus on the preview when it's going to also focus on the scan. Focusing for just the scan is a time saver in my opinion, unless you're choosing what frames to scan based on the sharpness of the preview.
- Fine Mode, my scanner doesn't have this. I'd try with it on or off. You can read what it does at http://www.hamrick.com/vuescan/html/vuesc28.htm#inputfinemode
- Multisampling, don't bother. I agree, while I've seen a slight difference with it on or off it's minimal to say the least even on the most difficult slides.
- Multiexposure, he doesn't mention this. I occasionally use this when the dynamic range of the scene may not be captured entirely by the scanner. This is not something I have set all of the time and is used more like a hail marry pass situations ("this scan just won't work with anything else, it's too dark, try multiexposure").
- Filters, I agree with what's said.
- White and Black points. Setting them both to 0% is something that used to zero out the program. Some time ago that changed, and now setting white point to 1 is the same as what setting it to zero used to be. It's also what the program defaults too.
- Leave Curves alone, I usually apply a slight increase to Curve Low from 25% (default) to 28-30%. This adds just a touch of contrast.
- Output Color Space. Use Adobe RGB if you're using Photoshop to touch things up. It's like feeding your car with the correct octane gas. While your car seems to run ok with other octanes, it's designed to use a certain octane and is maximized at that.
- Film Type, Ed Hamrick (the author of Vuescan) said (somewhere that I can't verify or find at the moment) that he only uses film profiles provided by the manufacturer of the specified film. This film profile changes the way the program perceives the colors coming out of the scanner. http://www.hamrick.com/vuescan/html/vuesc24.htm#topic21 If you look up the various film types you'll see that not only are the 3 films that Chris mentions is there but also a lot of variations on those films based on the density of the negative. So how are you supposed to know which one to use? I'm sure there's a way for Vuescan to tell you, it has an option under Prefs|Enable Density Display, but I'm not exactly sure how it's used. But even if you did know the density of your Kodak film will setting one of these profiles override other adjustments that can compensate for other things like the age of the light source within your scanner? IDK.
I just used Generic. There's too many other variables that make this very slight changes too to worry about.
- Printed Size, this just sets what the "dpi" value will be saved at, not at what dpi it is scanned at though. So if you open the file in a photo editor like PS CS2 and you have the ruler set to show inches the scans will always import at .9 x 1.3 inches (or there about) for a 35mm scan. If you set the Printed Size to Fixed DPI at 300 the exact same scan with the exact same number of pixels will come out to 5.7 x 8.8 inches (35mm frame, 2000 scanner dpi). Again this is the same number of pixels, you've just stated a different dpi to measure it, which still doesn't change anything because you've got the same pixels and they even display at 72 dpi (or whatever you have your editor set too or your display monitor displays). i.e. this setting is absolutely non-critical. The only real time to worry about this would be if you have a printer that insists the dpi setting makes a difference and rather than try to explain how it really doesn't you can just default Vuescan to give them what they think they want.
- File Type: It's a personal preference. If I like what's coming straight out of the scanner I'll use Jpeg. If I think it needs lots of work then I'll use RAW (which is an unprocessed tiff). If you want to have an unprocessed tiff that's inverted (a neg is now a positive image) then checking RAW Save Film does that. When I use raw, much of the setting discussed here are discarded because I use RAW Output with Scan, if I wanted to include all of these adjustments I'd use RAW Output with Save. If you want to go with Tiff's and not RAW files and that works for you I'm going to say more power to you. Use what works and go with it. Some people like a RAW DNG, but that was more of an afterthought in Vuescan and it wasn't really designed for it. Still some people like it.
- Tiff Compression, None. I agree and for the same reasons Chris states.
- Tiff File Type. If 16 bit Gray works for you then go for it. If you need more then increase to 48 bit RGB.
One more thing Chris didn't mention that I do:
- Crop Tab, Buffer %. Setting that to 15% or higher keeps the edge of the image from affecting the exposure of the scan. If you have a lens that vignettes and you're getting blown centers then your vignette is being included into too much of the calculation, the buffer % is like how much you want to spot meter (95% being a 5% spot meter but that's probably far too aggressive).
Again, take any of this with a grain of salt. I've tried to tell why I do what I do so that you can make your own mind up on what you want to try.
- Media, he tells you to always scan in greyscale. I'd say do that first, and if that doesn't work try RGB. The reason why can be found on Hamrick's website:
http://www.hamrick.com/vuescan/html/vuesc21.htm#topic18
To state one way is correct and the other is incorrect without first clarifying what you're scanning in the first place can be right some times and wrong others.
- 16 bit grey, again if that works for you great, if not then maybe you should try 48 bit rgb. To let the program convert to grayscale then take it into post production and try to tweak the levels seams a bit like shooting in a digital camera in jpeg then trying to do extensive levels adjustments. Why would you when you could just get the whole spectrum to work with? I do agree that 8 bit is usually not worth it unless you like exactly what's coming out of the scanner and you plan to just leave it at that.
-Preview Resolution, Auto should work for any dedicated film scanner. It calculates the preview to be roughly 4mp. On a flatbed 4mp is rather low quality and doesn't give enough information to calculate what the in scanner adjustments are going to be, and all adjustments are made with the preview scan. So if your "exposure" in the scanner is drastically different from the preview increasing that dpi can usually fix that. If that has no effect then put it back to Auto.
- Auto Focus, I so no reason to not set focus to Scan. After all, you're not capturing the scan from the preview so sharpness usually isn't an issue. It just makes the scanner wait to focus on the preview when it's going to also focus on the scan. Focusing for just the scan is a time saver in my opinion, unless you're choosing what frames to scan based on the sharpness of the preview.
- Fine Mode, my scanner doesn't have this. I'd try with it on or off. You can read what it does at http://www.hamrick.com/vuescan/html/vuesc28.htm#inputfinemode
- Multisampling, don't bother. I agree, while I've seen a slight difference with it on or off it's minimal to say the least even on the most difficult slides.
- Multiexposure, he doesn't mention this. I occasionally use this when the dynamic range of the scene may not be captured entirely by the scanner. This is not something I have set all of the time and is used more like a hail marry pass situations ("this scan just won't work with anything else, it's too dark, try multiexposure").
- Filters, I agree with what's said.
- White and Black points. Setting them both to 0% is something that used to zero out the program. Some time ago that changed, and now setting white point to 1 is the same as what setting it to zero used to be. It's also what the program defaults too.
- Leave Curves alone, I usually apply a slight increase to Curve Low from 25% (default) to 28-30%. This adds just a touch of contrast.
- Output Color Space. Use Adobe RGB if you're using Photoshop to touch things up. It's like feeding your car with the correct octane gas. While your car seems to run ok with other octanes, it's designed to use a certain octane and is maximized at that.
- Film Type, Ed Hamrick (the author of Vuescan) said (somewhere that I can't verify or find at the moment) that he only uses film profiles provided by the manufacturer of the specified film. This film profile changes the way the program perceives the colors coming out of the scanner. http://www.hamrick.com/vuescan/html/vuesc24.htm#topic21 If you look up the various film types you'll see that not only are the 3 films that Chris mentions is there but also a lot of variations on those films based on the density of the negative. So how are you supposed to know which one to use? I'm sure there's a way for Vuescan to tell you, it has an option under Prefs|Enable Density Display, but I'm not exactly sure how it's used. But even if you did know the density of your Kodak film will setting one of these profiles override other adjustments that can compensate for other things like the age of the light source within your scanner? IDK.
I just used Generic. There's too many other variables that make this very slight changes too to worry about.
- Printed Size, this just sets what the "dpi" value will be saved at, not at what dpi it is scanned at though. So if you open the file in a photo editor like PS CS2 and you have the ruler set to show inches the scans will always import at .9 x 1.3 inches (or there about) for a 35mm scan. If you set the Printed Size to Fixed DPI at 300 the exact same scan with the exact same number of pixels will come out to 5.7 x 8.8 inches (35mm frame, 2000 scanner dpi). Again this is the same number of pixels, you've just stated a different dpi to measure it, which still doesn't change anything because you've got the same pixels and they even display at 72 dpi (or whatever you have your editor set too or your display monitor displays). i.e. this setting is absolutely non-critical. The only real time to worry about this would be if you have a printer that insists the dpi setting makes a difference and rather than try to explain how it really doesn't you can just default Vuescan to give them what they think they want.
- File Type: It's a personal preference. If I like what's coming straight out of the scanner I'll use Jpeg. If I think it needs lots of work then I'll use RAW (which is an unprocessed tiff). If you want to have an unprocessed tiff that's inverted (a neg is now a positive image) then checking RAW Save Film does that. When I use raw, much of the setting discussed here are discarded because I use RAW Output with Scan, if I wanted to include all of these adjustments I'd use RAW Output with Save. If you want to go with Tiff's and not RAW files and that works for you I'm going to say more power to you. Use what works and go with it. Some people like a RAW DNG, but that was more of an afterthought in Vuescan and it wasn't really designed for it. Still some people like it.
- Tiff Compression, None. I agree and for the same reasons Chris states.
- Tiff File Type. If 16 bit Gray works for you then go for it. If you need more then increase to 48 bit RGB.
One more thing Chris didn't mention that I do:
- Crop Tab, Buffer %. Setting that to 15% or higher keeps the edge of the image from affecting the exposure of the scan. If you have a lens that vignettes and you're getting blown centers then your vignette is being included into too much of the calculation, the buffer % is like how much you want to spot meter (95% being a 5% spot meter but that's probably far too aggressive).
Again, take any of this with a grain of salt. I've tried to tell why I do what I do so that you can make your own mind up on what you want to try.
willie_901
Veteran
cabbiinc
Thank you for spending the time to share your experience.
Thank you for spending the time to share your experience.
thegman
Veteran
Top one looks better to my eye.
Particular
a.k.a. CNNY, disassembler
The preview pane is just produced by vuescan with a default setting to give an approximation of the possible final image (a preview). Don't fret over it. You want the actual scan to be flatter, as it is more likely to contain all the image information. You want the scan to be the best starting point for post processing, it doesn't matter how 'good' it looks out of the box. There is nothing in the preview that isn't also in the final scan.
Bob Michaels
nobody special
Thatspec: You are leading off on the wrong foot by trying to get a scan file that looks good rather than a scan file that contains all the data. Your final result will usually be better (and never worse) with a flat looking scan file rather than one that looks good.
Understand the only thing that comes out of your scanner is a bunch of 1's and 0's that indicate if the red, green and blue light passes through your film or not in a few million places. That is all, no more, no less. Everything else is software manipulation done in the scanner driver (Vuescan, Nikonscan, Esponscan) running in your CPU. This everything includes if the file is shown as a positive or negative, if the RG&B channels are shown individually or merged into a greyscale file, or what the contrast is.
You can choose to output the file as a positive then later convert it to a negative (or vice versa) in your image editor. You gain or lose nothing in Vuescan. Apparently Nikonscan, Epsonscan, etc. make some funky adjustments to the data outputted one way but not the other which you do not want.
You can choose to output the file as an RGB file which is 3 times the size and convert to greyscale yourself in your image editor if you want. Or, you can let your driver do the merge of the similar channels and output the smaller file (which has the same data) I personally find the Vuescan automatic merging of the R,G,&B channels to be ideal so I select to output the greyscale file.
While there is a level of skill involved in scanning, it is important to understand the basics and not fall into the belief that it is some black box voodoo.
But like everything else in photography, if you think you get better scans by closing your left eye and wearing your lucky underwear, do what works for you.
Understand the only thing that comes out of your scanner is a bunch of 1's and 0's that indicate if the red, green and blue light passes through your film or not in a few million places. That is all, no more, no less. Everything else is software manipulation done in the scanner driver (Vuescan, Nikonscan, Esponscan) running in your CPU. This everything includes if the file is shown as a positive or negative, if the RG&B channels are shown individually or merged into a greyscale file, or what the contrast is.
You can choose to output the file as a positive then later convert it to a negative (or vice versa) in your image editor. You gain or lose nothing in Vuescan. Apparently Nikonscan, Epsonscan, etc. make some funky adjustments to the data outputted one way but not the other which you do not want.
You can choose to output the file as an RGB file which is 3 times the size and convert to greyscale yourself in your image editor if you want. Or, you can let your driver do the merge of the similar channels and output the smaller file (which has the same data) I personally find the Vuescan automatic merging of the R,G,&B channels to be ideal so I select to output the greyscale file.
While there is a level of skill involved in scanning, it is important to understand the basics and not fall into the belief that it is some black box voodoo.
But like everything else in photography, if you think you get better scans by closing your left eye and wearing your lucky underwear, do what works for you.
willie_901
Veteran
These are words of wisdom. The goal is to create a initial negative with that best represents the information in the film negative/positive. You are creating a digital model of the original.
Thatspec
Established
Wow! Thanks very much guys, Cabbiinc, Bob, everyone, far more than I expected. There are a lot of variables here. To my eyes, the lower crop is more appealing though the bokeh is slightly pixelated. It's not the preview though, it's what shows up in the preview pane after the scan. Regardless of what I do with the full size saved scan in LR3, I can't ever get it to look as clear and vibrant as (the aforementioned pic straight out of the scanner). Here are a couple of more. Grain reduction is off but sharpening is on. Appears to me grain reduction is on and sharpening off in the lower (crop of the full size output)
Chriscrawfordphoto
Real Men Shoot Film.
Thatspec,
It doesn't matter that the final output doesnt match the preview. The most important thing to remember is that the file you get from the scanner is NOT - NOT - NOT a final, ready to use image. Its a starting point, and it has to be edited to get the correct contrast and tonal values, as my examples show. You seem to be latching on to the idea, which is false and will not give you the best quality, that you can and should somehow set vuescan in such a way as to get a perfect final image right from the scanner.
I took your unsharpened file and edited it to get tonality that should be pretty close to the original scene. I sharpened it after editing. You should scan at full 4000dpi resolution with no sharpening, so you can print it any size you want. Sharpening should be done only after resizing to final output size (save as a separate file so you do not lose your full-res original).
I edited it in Photoshop using curves adjustment layers. I flattened the file to make a jpeg to post here, but if you want to see the layered file so you can see what I did, email me and I'll email it to you.
It doesn't matter that the final output doesnt match the preview. The most important thing to remember is that the file you get from the scanner is NOT - NOT - NOT a final, ready to use image. Its a starting point, and it has to be edited to get the correct contrast and tonal values, as my examples show. You seem to be latching on to the idea, which is false and will not give you the best quality, that you can and should somehow set vuescan in such a way as to get a perfect final image right from the scanner.
I took your unsharpened file and edited it to get tonality that should be pretty close to the original scene. I sharpened it after editing. You should scan at full 4000dpi resolution with no sharpening, so you can print it any size you want. Sharpening should be done only after resizing to final output size (save as a separate file so you do not lose your full-res original).
I edited it in Photoshop using curves adjustment layers. I flattened the file to make a jpeg to post here, but if you want to see the layered file so you can see what I did, email me and I'll email it to you.
Attachments
cabbiinc
Slightly Irregular
It sounds to me like LR is doing something on import. Have you tried opening in Photoshop (or another pixel editor) and seeing if it's closer to what the output from Vuescan is giving you? Are you ouputting as a RAW file or as a Tiff?
Thatspec
Established
Thanks for chiming in Chris,
I will send you an email.
Cabbiinc, I have it set to save as a .tif, I've tried saving as .dng (too big) and .jpg as well with no discernible difference. I'm starting to wonder if it's the downsizing that the stock OSX picture viewer is doing (called preview, not to be confused with the preview pane of Vuescan which shows both a pixelated mess when you hit "preview", and then what I consider to be the best output from Vuescan after you hit "scan")
Edit: Seems I've had sharpening checked under the filter tab and that sharpening is being applied to the post scan preview pane (that I like so much as my starting point), but is not being applied to the saved file for unknown reasons... or perhaps it is being applied to the saved file at it's full size but when viewed at the same size as the post scan preview pane appears much softer. It's a slightly overly aggressive sharpening but I can't seem to approach it in LR without wrecking the picture. I realize sharpening is another can of worms undoubtedly already beaten to death
I would rather not go back to using Photoshop for PP (I think all I have is the original version of CS), but I will give it a try in an effort to explore all the options.
I think that at least solves the mystery for me
Many thanks to you all!
I will send you an email.
Cabbiinc, I have it set to save as a .tif, I've tried saving as .dng (too big) and .jpg as well with no discernible difference. I'm starting to wonder if it's the downsizing that the stock OSX picture viewer is doing (called preview, not to be confused with the preview pane of Vuescan which shows both a pixelated mess when you hit "preview", and then what I consider to be the best output from Vuescan after you hit "scan")
Edit: Seems I've had sharpening checked under the filter tab and that sharpening is being applied to the post scan preview pane (that I like so much as my starting point), but is not being applied to the saved file for unknown reasons... or perhaps it is being applied to the saved file at it's full size but when viewed at the same size as the post scan preview pane appears much softer. It's a slightly overly aggressive sharpening but I can't seem to approach it in LR without wrecking the picture. I realize sharpening is another can of worms undoubtedly already beaten to death
I would rather not go back to using Photoshop for PP (I think all I have is the original version of CS), but I will give it a try in an effort to explore all the options.
I think that at least solves the mystery for me
Many thanks to you all!