Transparency scaning - why emulsion side up?

[IanG]

Steve,

I don't know why you're defending this nonsense position.

If you put a ground glass between the light and the negative in an enlarger, you can get uniform illumination.
If you put that same ground glass between the negative and the enlarging lens you'll get totally scrambled image.

See the difference?

- Leigh

That's a very good analogy.

Now assume that ground glass is the surface of the emulsion with it's artefacts and that's why emulsion side up is sharper, it's also why wet mounted egatives/transparencies are sharper as well. (of course the gelatin surface only acts weakly in this way but it is enough when scanning)

The issue of emulsion surface artefacts affecting image sharpness and graininess was known about as far back as 1927/8 when there were atricles on wet mounting 35mm negatives to get the bst results withoptical printing, something Ctein practices.

When the first digital minilabs came out there were big problems with graininess with some films, this was due to the surface of the gelatin in the scanning process. Kodak did considerable research to overcome this. If you look at some negatives etc it's almost impossible to tell which is the emulsion side without looking at notch codes, edge markings etc. However older films have a noticeable matt sheen on the emulsion side.

Ian

[ChuckT]

Scanning is based on the assumption that you are doing original 'chromes.

cvt

[chuck94022]

Epson V700, original Epson holder (120), scanned with Epson Scan.

I think until you eliminate the bowing variations caused by the original holder, and can assure that the film is in the plane of sharpest focus for both scans, your scans do not prove or disprove anything. The plane of sharpest focus is *that* thin with the Epson, you really have to get the film *flat*. It might not matter with a large format sheet, because you will scan at low ppi, but with smaller formats, where you have to scan at the scanner's maximum optical resolution to get any decent result, flatness and perfect alignment with the focus plane is critical.

I can easily believe that you will see differences just in the difference in bow between the emulsion being up or down in the standard holder, especially if you haven't adjusted the holder to the precise plane of sharpest focus (almost certain it is not with the standard Epson holders, unless you are a frequent lottery winner).

It would be useful to know from experiment whether orientation makes a difference, but these variables need to be removed.

[chuck94022]

Scanning is based on the assumption that you are doing original 'chromes.

cvt

Whuuut???

[Nathan Potter]

Chuck, a few years back I used a chrome on glass resolution mask that I tilted thru the plane of presumed best focus to get an idea of the best focus plane and DOF for my V750. The resulting plot is somewhere on the forum but I have it below for reference.

Don't infer anything about the ultimate resolution from the plot since the criteria for the resolution given was for near zero contrast for the line pairs chosen. That condition was in order to easily resolve the minimum lp/mm for plotting purposes. The plot only shows the hyperbolic nature of DOF as a function of film plane position.

There turns out to be a greater DOF than I had presumed if one relaxes the maximum resolution criteria slightly.
However some of my films actually bow up to 2000 µm (2mm) when sitting in the Epson holder and there can be considerable loss of resolution as you have suggested.

Once the plane of best focus for the emulsion is captured I confess to not being able to tell the difference between emulsion side up or down but I have this intuitive feeling that I don't like the film base to appear between the emulsion and the scanning lens.


V750RESjpg by hypolimnas, on Flickr

Nate Potter, Austin TX.

[Leigh]

Neat information, Nate.

Question...
Was the target rectilinear, i.e. made up of straight lines and spaces like the USAF 1951 target?

There's a fundamental problem with using that type of target with a scanner.
It was designed for use with continuous-tone media like film, which is not digitized.

- Leigh

[Nathan Potter]

Leigh, a keen observation that I had to think about while getting the data. Didn't do it right at first because the line pairs are in fact just a series of three straight lines. Looking at the Imatest software and ISO 12233 reveals the necessity of considering aliasing between the linepairs on reticle and the linear sensor array when they are collinear. I could see this on the first scans.

So the data shown was obtained by tilting the reticle slightly with respect to the sensor array then with respect to the stepper drive. The result of course shows different resolution numbers along the line pairs. I picked the best section for resolution along the lines to generate the plot. All a bit of a PITA but I think the plot is fairly reliable.

The other pain, as mentioned above, and I'm thinking should be generally considered during resolution tests, is that to derive meaningful MTF data one needs to scan and measure using a full grey scale. Not, if you will a slice thru the grey scale (that is at a single density). Generally we are interested in resolution capability and contrast at a full density range (at least I am). It seems to me that a similar technique can be used for direct evaluation of DSLR lens performance using a distant target and making sure one employs a tilt to the linepairs relative to the sensor array. Also the direct evaluation of macro lens capability by directly photographing the reticle can be done.

Of course there are neat structures in the ISO 12233 pattern to eliminate the aliasing then provide a fourier decomposition of a line edge for the establishment of resolution but I don't want to spend 2 to 3 K$ for that nice package. Any of your insights are always helpful since I'm blundering along with what I have handy for hardware and software.

Nate Potter, Austin TX.

[Leigh]

So the data shown was obtained by tilting the reticle slightly with respect to the sensor array then with respect to the stepper drive. The result of course shows different resolution numbers along the line pairs. I picked the best section for resolution along the lines to generate the plot. All a bit of a PITA but I think the plot is fairly reliable.

Hi Nate,

Sounds like a good way to significantly reduce the problem and improve the accuracy of the results. Well done.

The targets that I've considered using are circles rather than lines. Have not tried this yet, but it should work well.
Edmund Optics has them, listed specifically as scanner test targets, but I'm not interested to the tune of $450.

Thanks.

- Leigh

[sanking]

Just for the record, here is a link to a USAF 1951 chart that Silverfast sells for testing scanners.

http://www.silverfast.com/show/resol...target/en.html

Sandy

[ChuckT]

I've been a Photo Lab Rat for almost 50 years now. And it really bugs me when I've had to scan film flipped instead of emulsion-side-to-lens.

The bloody scanner needs to read the emulsion not the base fer C_sake!

However all is not as it should be some times because scanners specs seldom mention 2 important points -> Does this scanner have a glass lens and does it auto-focus.

If the scanner doesn't auto-focus then I've been stuck with whatever the software enginneers (hah!) "engineered" into the App and that has at times been scanning base-to-lens.

I have this intuitive feeling that I don't like the film base to appear between the emulsion and the scanning lens.


V750RESjpg by hypolimnas, on Flickr

Nate Potter, Austin TX.