How does scanning software improve my scans?

[papo]

I keep hearing about this and want to understand what the software does, that the actual scan doesnt.

I have Photoshop and Lightroom - arent they good enough to do modifications on scans?

Thanks you guys

 

[dmr]

About the only "feature" I regularly use in scanning software is multiple samples. The scanner I have does not have infrared noise suppression so I have to fix all of that in post.

 

[papo]

I have an Epson V550 and use the standard software thant came with it - what could i do to improve my scans?

 

[Pentode]

In my (admittedly limited) experience, scanners often work well with the software they came with, but different sets of software may have different feature sets that better fit your working style.

I use a V500 and I've gotten my best results with Epson's native software, EpsonScan. I did use a trial version of VueScan for a little while, though, and it had options the EpsonScan software didn't such as improved previewing options and a way to indicate clipping that was very, very handy.

For my needs, EpsonScan ended up giving me better results, but if I had spent more time with VueScan it may have matched or surpassed it, with better features to boot.

Ultimately, I suspect just about any scanning software can get you what you need, it's just a question of how easily you get there. No software package is going to improve the actual optics of the scanner itself.

 

[mani]

In my view, nothing that any scanning software can do in terms of image manipulation can match the features and accuracy of Photoshop or Lightroom.

Scanning software is best when used to capture a neutral, raw scan with as much detail from the original preserved as possible (which may include multi-passes, or other techniques), but which doesn't alter the character of the scan before outputting the file.

 

[willie_901]

In general, use software options that result in a file with the highest amount of information content. This is not necessarily the most convenient way, but it provides you with the highest degree of flexibility during post-production.

If convenience is more important just set everything to 'auto' and hope for the best.

Select the least compressed output-file format possible... probably TIFF. There are two types of file compression. JPEG output files use lossy compression. The JPEG file does not contain all of the original data. TIFF file compression can be lossy or lossless. Always select lossless. A 'flat' lossless TIFF file does contain all the information recorded by scanner.

Do not make any adjustments to the preview image rendering. This results in a 'flat' or linear rendering. Cropping or straightening (rotation?) are not rendering parameters. The scanner software rendering adjustments do not affect the scan, they are applied after the scan is complete. This does not apply to spot removal and some other functions such as focus adjustments (when available).

Do not use noise filtering (a.k.a. noise 'reduction').

Some software supports multiple scan modes that are averaged in clever ways to improve IQ.

Use grain aliasing control and dust spot elimination (some dust elimination tools don't work with B&W negatives).

With this approach the initial image rendering might not look very good. However using LR and Photoshop can produce excellent results. It took me some time to develop a satisfactory post-production workflow. Color-temperature adjustments for color negative film scans is tricky at first. I always started in PS in invert the image and manually adjust the color temperature. Then I did all other rendering adjustments in LR. People who are skilled with PS will consider LR redundant. I am not skilled with PS and I use LR for file and project organization. Also, the NIK suite of LR plug-ins can be very useful.

 

[MIkhail]

About the only "feature" I regularly use in scanning software is multiple samples. The scanner I have does not have infrared noise suppression so I have to fix all of that in post.

Do you find that multiple passes give you a better result?
I tried number of times and it never looked any different to me...
I have Nikon Coolscan ED

 

[MIkhail]

I keep hearing about this and want to understand what the software does, that the actual scan doesnt.

I have Photoshop and Lightroom - arent they good enough to do modifications on scans?

Thanks you guys

I use Vuescan rather than Nikon software that came with scanner, specifically for the reason that (it seems to me anyway) Vuescan is less invasive to the image, so to speak.
It gives me most neutral picture that I can process further in LR or PS

 

[kj_]

It probably depend what medium you shoot (BW / color neg or slide film). From my research the biggest difference between apps seems to be in colour correction and adjustments including inversion. More sophisticated softwares contain more sophisticated tools for adjustments and presets for different film stocks. Getting accurate colour scans can be challenging and having good presets makes a big difference. If you shoot mostly BW or slide film that will obviously have less of an impact.

 

[Rangefinder 35]

I use Minolta Multipro scanner with Vuescan software. This independent software (it cost $79 when I bought it few years back) allows me to scan in RAW, yielding a file with all the info of the original transparency, which then I can match, or improve over the original in Photoshop. I use multiple sampling feature only if I get bands in the scan, which happens very seldom, mostly in dark, uniform density originals. Otherwise a single sampling is better. No problem with misalignment of samples there. Using RAW scanning let me pull the faintest highlight detail from the slide, which in normal scanning would've been lost.

 

[MIkhail]

I use Minolta Multipro scanner with Vuescan software. This independent software (it cost $79 when I bought it few years back) allows me to scan in RAW, yielding a file with all the info of the original transparency, which then I can match, or improve over the original in Photoshop. I use multiple sampling feature only if I get bands in the scan, which happens very seldom, mostly in dark, uniform density originals. Otherwise a single sampling is better. No problem with misalignment of samples there. Using RAW scanning let me pull the faintest highlight detail from the slide, which in normal scanning would've been lost.

Never even occurred to me to scan RAW... I will give it a try.

 

[Prest_400]

Epson V550 and Epsonscan user here too.

Do not make any adjustments to the preview image rendering. This results in a 'flat' or linear rendering. Cropping or straightening (rotation?) are not rendering parameters. The scanner software rendering adjustments do not affect the scan, they are applied after the scan is complete. This does not apply to spot removal and some other functions such as focus adjustments (when available).

That is interesting. I followed Colton's guide to set white and black points with some headroom, and set output from the (10, 240) default to (0, 255). Maybe hit autoexposure (before the above really) to get WB into the ballpark.
End with a flattish TIFF that gets adjusted in LR afterwards.

The point about epsonscan's output (seen below the histogram) being default at 10, 240; I recall was not something good, but forgot the reason now.

I've read that a Linear Flat Scan (Negative without inverting) and using Colorperfect to invert works wonders.

Portra does scan well, properly exposed. Out of curiosity, I tried to do the "airy" lab look, from denser negs, and the scanner (bet the operator also) often can't keep up. A part of it is density adjustment in scanner (in epson scan, how?) and I guess also some LR post. I did notice that some highlights (deep shadows) in negative film do not seem to have the detail supposed to, even with headroom.

 

[Ronald M]

I use Vuescan rather than Nikon software that came with scanner, specifically for the reason that (it seems to me anyway) Vuescan is less invasive to the image, so to speak.
It gives me most neutral picture that I can process further in LR or PS

I found the same with Minolta software & 5400. Sadly I can not use it with current Mac OS.

My aim is to get a color balanced and properly exposed scan. All mods/refinements are done in Photoshop. I even do HDR scans of the same neg and merge in photoshop.

I have negs that I printed with Leica enlargers and best lenses that need no dust spotting. Scans of the same show enough spots that it is an hours work to clean them up, i.e. a scan for print is better than an optical print.

If I have a bad neg, I use the IR clean up and it does not seem to destroy the image.

Grain reduction is best done with programs made to do it, i.e. Grain Surgery, which is no longer available. Noise reduction programs are not nearly as effective.

All said and done, it you want an electronic image, use a digital camera. I use Nikons and Leicas. For darkroom prints, I use film with same brands of cameras.

There are no quality scanners available new for a reasonable price so I believe we are beating a dead horse trying to use this old tech.

 

[PaulCooper]

I found the same with Minolta software & 5400. Sadly I can not use it with current Mac OS. .

Ed Hamrick, the developer of Vuescan software, updates the SF quite often to keep up with new OSs. You can check compatibility of your system on Vuescan website. It's worth the try.

 

[mani]

All said and done, it you want an electronic image, use a digital camera. I use Nikons and Leicas. For darkroom prints, I use film with same brands of cameras.

There are no quality scanners available new for a reasonable price so I believe we are beating a dead horse trying to use this old tech.

This is such utter garbage I can't even be bothered to go over the facts again. Why bother. Stick to a topic you know something about.

 

[willie_901]

Epson V550 and Epsonscan user here too.


That is interesting. I followed Colton's guide to set white and black points with some headroom, and set output from the (10, 240) default to (0, 255).
...

The white and black points are important. If the white point is set too low and, or the black point is set to high 'clipping will occur'. A symptom of clipping is spikes in the scans histogram.

I don't if setting the white and black points too high or too low can degrade he scan file. I'm thinking it won't because the histogram white and black points can be set during post-scanning rendering adjustment.

It's possible wider than necessary white and black points 'waste' bit depth. I'm not sure about this.

 

[Prest_400]

The white and black points are important. If the white point is set too low and, or the black point is set to high 'clipping will occur'. A symptom of clipping is spikes in the scans histogram.

I don't if setting the white and black points too high or too low can degrade he scan file. I'm thinking it won't because the histogram white and black points can be set during post-scanning rendering adjustment.

It's possible wider than necessary white and black points 'waste' bit depth. I'm not sure about this.

Had to recheck his guide to see the process again. I do follow up the first one most, but set manually the extreme points.
Usually leave a headroom, clipping is no good. The TIFF resulted is a rather flat photo. Though in Post I tend to adjust exposure +-0.25 in post.

My V550 does seem to lose detail (Density issues) in deep dense areas, shadows in positives and extreme highlights in negatives. Overexposed negs don't help, correctly exposed and low contrast scenes give no problem and with very little adjustment look nice.

The bit loss makes sense, but how much of a problem may it be? Checking out the usual "airy" scans, there is often a lot of tonality located in higher lights, with less emphasis on mids, and then some deep shadows. That seems tonality "stretching". Some folks, and I would like to know if some labs do, are editing over JPEGs.

Just realized this above thing sounds like audiophilia. 😀

There's a photo fair around here in a few weeks and a lab is bringing their SP3000 along with an operator. So far I've reserved my scanner time and should test. (Just need the event pass which is processing)