6x7 film negative scanning help

[boogotti]

hello,
Im new to shooting on film and film scanning. I have two questions what im not sure about.
Im shooting on mamiya RZ67 and using VueScan to scan with @16bit raw tiff, canon 9000F mark ii scanner also using colourperfect to invert negative scan in photoshop.

- When scanning negative is the best way shiny side down?

- Is scanning at 9600dpi then scaling image down to 10000px on longest edge using Bilinear, is that right way?

Im not sure what resolution i should be scaling the image to as 645 film format is 8000px longest edge? (not sure if thats correct?)

Many thanks
Ravi

[Mark D Segal]

hello,
Im new to shooting on film and film scanning. I have two questions what im not sure about.
Im shooting on mamiya RZ67 and using VueScan to scan with @16bit raw tiff, canon 9000F mark ii scanner also using colourperfect to invert negative scan in photoshop.

- When scanning negative is the best way shiny side down?

- Is scanning at 9600dpi then scaling image down to 10000px on longest edge using Bilinear, is that right way?

Im not sure what resolution i should be scaling the image to as 645 film format is 8000px longest edge? (not sure if thats correct?)

Many thanks
Ravi

Normally, one scans negatives with the shiny side up, but check whether the manual with the scanner or accompanying software advises otherwise.

On the scanning resolution, there are two schools of thought on this: (1) "right size" the dimensions and the resolution from the get-go, or (2) scan at maximum optical resolution and then downsize. If you have a very clear idea what are the maximum likely dimensions and resolution you will ever likely need, there is no point bloating your storage with over-sized files relative to what you are ever likely to need. For example if you know that you will never likely print the photo at more than 300~360 ppi and at 20 inch largest dimension, you only need the number of pixels per inch input that will give you those dimensions at that resolution on output.

OR

(2) Scan for maximum optical resolution because you never know what is the maximum you will do with this photo, you don't want to have to rescan it, and downsampling helps reduce noise and artifacts. (But there are other bespoke applications for handling noise and artifacts that you have more control over.) I recommend that it only takes a bit of time to experiment with your own photos and requirements, and see which idea you like better. Some people make a religion of this, but it is really just an empirical question of what you need and looks better for your needs. In my work, I use option (1), which can periodically coincide with option (2).

[boogotti]

Thanks Mark

Is there a standard resolution for 120 film?
Im not going to do big prints but want a resolution that is fair for the 6x7 format?

Should this be around 8000px on the longest edge?

I will have a read on the manual of my scanner, to see which side the negtive works with.

[Bart_van_der_Wolf]

hello,
Im new to shooting on film and film scanning. I have two questions what im not sure about.
Im shooting on mamiya RZ67 and using VueScan to scan with @16bit raw tiff, canon 9000F mark ii scanner also using colourperfect to invert negative scan in photoshop.

- When scanning negative is the best way shiny side down?

Hi Ravi,

Emulsion side towards the sensor. This is to avoid diffusion by the emulsion (dye clouds or silver grain), which then gets internally reflected by the film support/base material. The  film support is usually the shiny side, which is thus on the side of the lightsource.

- Is scanning at 9600dpi then scaling image down to 10000px on longest edge using Bilinear, is that right way?

Scanning at the sensor's native resolution will reduce the risk of grain-aliasing and will thus produce the least grainy image. The Canon scanner you mention supposedly has a native resolution of 9600 PPI, so that's best for quality. That will produce huge files, so one will usually need to down-sample the scan to achieve more manageable sizes. Do not use bilinear if you have the possibility to use something better. Bicubic will e.g. maintain better contrast, and usually produces less aliasing artifacts when down-sampling. Better algorithms will produce potentially even better results, depending on thee scale used.

Im not sure what resolution i should be scaling the image to as 645 film format is 8000px longest edge? (not sure if thats correct?)

Film has no pixels, they are purely the result of digitizing. You typically down-sample to manageable file sizes in function of the output requirements (output size in inches, times 720 PPI maximum, but some output modalities only require 256 PPI).

Cheers,
Bart

P.S. The 9600 PPI is reported as "optical resolution" in Canon's specifications for film, and 4800 PPI for other sizes. VueScan will allow to choose whatever is the true (maximum) native resolution of a scanner.

[boogotti]

Emulsion side towards the sensor. This is to avoid diffusion by the emulsion (dye clouds or silver grain), which then gets internally reflected by the film support/base material. The  film support is usually the shiny side, which is thus on the side off the lightsource.

Thanks Bart for clearing that up.

Im confused as to scanning the negative emulsion side towards the scanner sensor. As what you have said makes sense to me, but reading the canon manual says to scan shiny side down  ? (have the negative reversed when putting in scanner)

I might do some testes to see if that makes a difference?

[Mark D Segal]

I think the guidance of the manual on which way to load the film should be determinative. They are likely aiming at giving you a scan that shows the image as it is intended to be seen. So start with that.

As for the business about native resolution, I would subject the grain-aliasing risk to an empirical test. See whether anything like that shows in your prints by "right-sizing" the scans, before hogging huge amounts of storage with vastly over-sized files relative to your requirements. Bart can explain or point you to references about the risk of grain-aliasing given his expertise in the matter, but what matters most in the final analysis is what you see.

[rgs]

I have several years of 6x7 negs that I have scanned. I use an Epson v700 and VueScan. I want to archive my better work at a useful size but a 6x7 chrome at the v700s optical resolution (4800 if I remember correctly but I always have to look it up) is huge and I'm unlikely to need it that big so I scan at about 1200-1800 using multiscan techniques to optimize quality and limit noise. I do not apply any noise reduction, sharpening, or Digital ICE to the scan. All that can be done better in post or (dust spotting) by hand. I just want the best basic scan I can get. I file the chromes where I can easily re-scan if I need a bigger file. I always scan to TIFF, not JPEG. VueScan also offers DNG but it's not quite the same as a DNG from a DSLR's RAW file and I don't trust Adobe (they may make me really mad or someone may write better software that does not read DNG) so I avoid DNG.

The best thing you can do is apply whatever multi-scan techniques the combination of your scanner and VueScan can provide. Multi-scanning will reduce noise without degrading image quality and extend dynamic range. Study VueScan's multi-scan options with your scanner.

BTW the RZ is not a 645 camera (unless you have a 645 back), it's 6x7. And if you do have only a 645 back and you plan on doing more, try to find a 6x7 back. The IQ difference is huge.

[LesPalenik]

In practical terms (and depending on your scanner), scanning a medium format film at 1600-2400 dpi is the best choice.

[Bart_van_der_Wolf]

I think the guidance of the manual on which way to load the film should be determinative. They are likely aiming at giving you a scan that shows the image as it is intended to be seen. So start with that.

I agree with Mark that there may be a reason why Canon specifies it as they do, but it may also be something as trivial as an over-simplification in their software that doesn't allow to see the image right side up, mirrored. I don't know their software well enough. Anyway, VueScan has all those possibilities, so I'd go for the common practice of positioning the film with emulsion side towards the sensor, unless there is a mechanical reason like in the film holder e.g. to avoid Newton's rings.

As for the business about native resolution, I would subject the grain-aliasing risk to an empirical test. See whether anything like that shows in your prints by "right-sizing" the scans, before hogging huge amounts of storage with vastly over-sized files relative to your requirements. Bart can explain or point you to references about the risk of grain-aliasing given his expertise in the matter, but what matters most in the final analysis is what you see.

This is the original article that warned about grain-aliasing when scanning at lower that native scanner resolution. Here are some comparisons between grain-aliasing from scanning at a lower PPI and downsampling to that same PPI.

Norman Koren has also written about grain-aliasing, and also mentions the influence of the resolution limitations of the scanner optics to even resolve such level of detail (the lens limitations can function as a high-pass filter), and noise reduction software to mitigate the artifacts.

In my experience, there is also film resolution to be had above 4000 or 4800 PPI. My 5400 PPI Minolta scanner extracts quantifiably more resolution from the same image of a test target compared to lower (4000 PPI) resolution scans. I got the following resolutions from three different scans of the same test target shot on film: Nikon LS2000: 45.4 lp/mm, Nikon LS4000: 62.1 lp/mm, Minolta SE5400: 76.1 lp/mm. That additional resolution also helps if one wants to use noise reduction software, because that usually loses a bit of resolution. Of course not all images that were shot handheld will have such resolution, and not all (camera as well as scanner) lenses perform as well, but reducing grain-aliasing also allows to use more sharpening.

Cheers,
Bart

[Paul80]

Hi

I scan all my film, neg & slide on my old Epson 4490.  I scan at the optical resolution of the scanner, 4800dpi & tavern as TIFF.  I have 3 different scanner software , Epson, Vuescan & Silverfast. When scanning negatives especially colour, Silverfast gives the best result by a long way. Especially if I just select generic for the film make and type.

Paul

[boogotti]

Thank You Guys!

I have a 120mm 6x7 back for my RZ67

Yes Bart, i think Canon have said that because using there software (which im not using/ using vuescan instead) , when you scan the negative shiny side down it scans correctly orientation (no need to flip after)
But im pretty sure from what you said scanning emulsion side down will give best results

multi-pass scanning
One query i have is when using vuescan with the Canon 9000F ii, i tried using multi-pass scanning which i set to 4. i found that comparing a scan with 1 pass. The 1 pass scan is much sharper i have read that the registration of the scanner is not perfect and can blur the scan? How you guys find using muti-pass scanning? would like to here your experience on this.

Ravi

[Bart_van_der_Wolf]

multi-pass scanning
One query i have is when using vuescan with the Canon 9000F ii, i tried using multi-pass scanning which i set to 4. i found that comparing a scan with 1 pass. The 1 pass scan is much sharper i have read that the registration of the scanner is not perfect and can blur the scan? How you guys found with using muti-pass scanning? would like to here your experience on this.

Hi Ravi,

I don't know how the Canon scanner hold its film, but I suspect that the film is moving (due to temperature change) between the scans. VueScan in principle doesn't reposition the scan-head/sensor, it just repeats the scan multiple times with the scan-head in the same position, and averages the scans to reduce random (electronic) noise, before moving to the next scan-line position.

Cheers,
Bart

[Alan Klein]

Don't the multiple scans blur the scans to eliminate the grain and effectively soften the focus?   Can't  you eliminate grain in post and have better control of sharpness?

[boogotti]

Don't the multiple scans blur the scans to eliminate the grain and effectively soften the focus?   Can't  you eliminate grain in post and have better control of sharpness?

oh right, i got mixed up with multi-pass scanning and multi-exposure scanning, as i think canon 9000F ii doesn't have multi-exposure scanning (in VueScan?)

Can the Canon 9000F ii do multi-exposure scanning?

[Mark D Segal]

This is the original article that warned about grain-aliasing when scanning at lower that native scanner resolution. Here are some comparisons between grain-aliasing from scanning at a lower PPI and downsampling to that same PPI.

Norman Koren has also written about grain-aliasing, and also mentions the influence of the resolution limitations of the scanner optics to even resolve such level of detail (the lens limitations can function as a high-pass filter), and noise reduction software to mitigate the artifacts.

In my experience, there is also film resolution to be had above 4000 or 4800 PPI. My 5400 PPI Minolta scanner extracts quantifiably more resolution from the same image of a test target compared to lower (4000 PPI) resolution scans. I got the following resolutions from three different scans of the same test target shot on film: Nikon LS2000: 45.4 lp/mm, Nikon LS4000: 62.1 lp/mm, Minolta SE5400: 76.1 lp/mm. That additional resolution also helps if one wants to use noise reduction software, because that usually loses a bit of resolution. Of course not all images that were shot handheld will have such resolution, and not all (camera as well as scanner) lenses perform as well, but reducing grain-aliasing also allows to use more sharpening.

Cheers,
Bart

Hi Bart, Ravi,

Yup, read all that stuff in the past, good research and doesn't change in the least my recommendation Ravi that you do real-word tests with real-world photos and see what works best for you. And Bart: you are comparing a 189 dollar scanner with products worth well over a thousand when they were marketed. When you are talking about LP/mm this is an outcome includes for everything in the imaging path, not only the stated resolution of the sensor - light handling, lens quality etc., etc. Could well be apples and oranges, why I think an empirical test with the user's own equipment, environment and needs is the best possible advice.

[Mark D Segal]

Don't the multiple scans blur the scans to eliminate the grain and effectively soften the focus?   Can't  you eliminate grain in post and have better control of sharpness?

I agree with your point that one has better control over grain mitigation and sharpening using post-scan applications that handle these tasks well and give the user full control over the effect being achieved. While some scanning applications may have reasonable tools for these tasks, I prefer not to use them because a scan bakes-in everything we've done, whereas all of this stuff can be done on layers in Photoshop or reversibly in Lightroom, making the process non-destructive of options. As well, for those who like using a multi-stage sharpening workflow, this is best achieved with a bespoke product for this purpose from start to finish.

[Mark D Segal]

oh right, i got mixed up with multi-pass scanning and multi-exposure scanning, as i think canon 9000F ii doesn't have multi-exposure scanning (in VueScan?)

Can the Canon 9000F ii do multi-exposure scanning?

Yes, the scanner is capable of multi-exposure provided the software you use can operate it. I know that SilverFast can; for Vuescan, open your scanner, open Vuescan, go to the Input tab and see whether there is a function for multiple exposure (not "Number of samples" - this is not the same function).

[Doug Fisher]

>>but I suspect that the film is moving (due to temperature change) between the scans. <<

With the moderately priced flatbeds, the softness during multi pass scans is usually caused by the fact that the scanning head is unable to start each pass at exactly the same position.  Heat could also be a factor to a lesser extent.

Doug

[Bart_van_der_Wolf]

Yes, the scanner is capable of multi-exposure provided the software you use can operate it. I know that SilverFast can; for Vuescan, open your scanner, open Vuescan, go to the Input tab and see whether there is a function for multiple exposure (not "Number of samples" - this is not the same function).

To be clear, the number of samples does not move the scan head, and only removes electronic noise by averaging (1/4th of the noise at 16x sampling). This should improve image quality, provided that the film doesn't move during the iterations (because it takes 2x, 4x, 8x, or 16x longer to expose).

Multiple passes requires accurate repositioning of the multiple scans (which not all scanners can do), if the scanner hardware doesn't support multiple samples.

Multiple-exposures allows to penetrate dense film better (if scanners allow to control this), but that is something different, and is usually not required for negative film.

Cheers,
Bart