Let's see your DSLR film scanning setup!

[ArnoG]

@Alain_Oguse Very interesting and at the same time puzzling results. I need to think about that some more. Perhaps we’re seeing some form of interference pattern(???). Anyway, I didn’t find much time recently, but your results inspired me to do a simple quick test to see the effect of collimated light myself and I’m quite impressed. A long time ago, when I started the “collimated light for more sharpness” thread, I was playing around with a set of collimating lenses from an old enlarger when I thought of a simpler test: I purchased a “privacy screen” for a mobile phone. The ones that only allow one to see the screen from a perpendicular viewpoint. They a comprised of a film with vertical bars that only allow light to pass perpendicular to the film. Hence, a poor man’s collimator. I never tried it though, until recently when I came across your findings. If this forum will allow me to upload a video, I can demonstrate my findings (I was impressed). I simply stuck the privacy screen onto my diffuse light source like so:

And the effect is recorded in a video which apparently I cannot upload directly from my phone. I report back and add it once I figure out how to add the video.

Edit: Figured it out I think:

Yes, worked. The resolution of the video is a bit crappy, and I recorded a zoom of a section of Vlad’s target on a monitor, but the effect is very clear. I think I paid maybe €10 for a set of two of those screens (for an old iPhone model, so cheap), and it’s by far the most cost effective upgrade in my scanning system. The effect is impressive and will lead me to revisit the collimator lenses to do this properly, but simply adding a cheap privacy screen to increase the resolution should be a no brainer…

[Alain_Oguse]

@Harry

the V600 scan is probably easier to explain though, credible tests with a USAF 1951 etched target show that it resolves only 1560 ppi

Thank you very much for this information. I was definitely naive to take Epson’s word for it. And I’m shocked to find that they’re capable of worse than I expected… I have the tiny consolation of having had results that didn’t let themselves be fooled

[Alain_Oguse]

@ArnoG

I simply stuck the privacy screen onto my diffuse light source

That’s an interesting and original approach. I like it! In the end, the right solution for color could indeed be a mix between your RGB light system and simple collimation by condensers like on our good old Durst enlargers. Your tests are going to be fascinating!

For my part, I’m starting to think about how to improve my prototype. And I’ve found these lenses that might be interesting?
https://www.edmundoptics.fr/p/100mm-dia-x-150mm-focal-length-pcx-condenser-lens/35971/

[ArnoG]

Interesting that these lenses can be bought. Some years ago I picked up an old meopta(?) enlarger, that has a similar set of condenser lenses, even comparable diameter. It’s by far the cheapest option to extract them from an old enlarger…

[SSelvidge]

Hey @ArnoG thanks for posting that.

Does this screen only work in one plane? For instance when holding your phone vertically, left to right is modified but up and down is not.

Or does it work in all 4 directions?

If yes, I think it works just like the LES (Light enhancement sheets) that Cinestill sells, which I believe are just cut down 3M Light Enhancement films that are often used in light up signage, and as computer privacy screens, etc. @Harry and I talked about it recently here on the forum. But maybe it is something different?!

[ArnoG]

Hi @SSelvidge , I didn’t try the sheets on my phone but suppose that “privacy’ should work in both directions. Will look better when I find more time. The phone sheet does reduce the brightness, so it simply seems to cut off light in non-perpendicular to the sheet directions. If the CS/3M sheets increase brightness, then they will somehow focus non-perpendicular light into the perpendicular direction, which would mean that they contain an array of micro lenses which could lead to funky interference with the micro lenses in camera sensors in the same way as when one takes a digital shot of a computer screen. If the grain spacing/size is of similar dimensions as the micro lens spacing in a sensor then interference could possibly be an explanation why @Alain_Oguse sees funky artifacts with modern grain(?). In general, interference patterns are well known and well understood physical phenomena.

Overall I believe that @Alain_Oguse has pointed out something very important that at least I completely missed: Diffuse light will effectively smear out edge definition, and smear out the grain edges as well, causing them to look different than what @Alain_Oguse was used to seeing in analog world. Collimating the light will not increase sharpness (which is defined by the optics and sensor combination), but will make the edges better defined, which gives in the end higher resolution, and should thus be better for scanning. Picking up more dust and scratches eventually means scanning with improved definition. One just has to be more prudent to avoid dust and scratches.

Thanks for pointing out the 3M sheets, which might work better than the phone screen I came across. In the end though I’m getting more and more convinced that proper collimated light should yield the highest resolution, and also makes the light very efficient since it’s fully directional. The source will need to be a proper point source and for B&W a single wavelength would be ideal (perhaps a laser source???). For color I could simply reprogram my LED array to use only a single RGB LED and remove the diffusing glass. I still have a set of collimating lenses from an old enlarger so it will just take time to do some experimentation. This is all quite theoretical though, and if it will create a mess with modern grain film as @Alain_Oguse points out, perhaps some removable 3M sheets would be a more suitable solution.

The bottom line is that by adding a sheet of something that make the light more directional, one increases edge definition and this leads to a scan with visually improved resolution as is evident from the video I posted above. It’s a simple, cheap, and easy way to improve any diffuse light panel it seems. What sheet is optimal remains to be investigated.

[Alain_Oguse]

I answer to myself because the latest results I obtained on the digitization of tabular grain films are completely false. Indeed, I subsequently realized that the horrible aspect obtained resembled what I had already obtained in the past, less serious, and of which I had been able to identify the cause as being a focus defect. So I started my tests again
yesterday and I was able to obtain excellent results this time. Which bodes well for color work

I still have to understand how this focus defect could have been possible? Because I had carefully ajusted and tested it using a TriX, then, without changing anything, tested the Tmax. So either my prototype is causing some kind of slip that has yet to be identified, or the thinning of the gelatin sought by film manufacturers for the tabular grain would be enough to shift the focus… I have certainly already noticed a hyper sensitivity of the focus in collimated light, but at this point it seems excessive to me to be possible.

So here’s a picture of the corrected result. The band on the right is the Tmax with correct focus. In the middle, the focus is wrong. On the left, for visual reference, the TriX. Admittedly, despite a clear improvement, I still don’t have as good a result with the Tmax as with the TriX. And I still don’t understand why.

I’ll keep looking…

With apologies for my mistakes last week!

[Mentalkisiel]

Hi,

Soligor slide duplicator and DIY box with the backlight

[ArnoG]

Hi all,

motivated by @Alain_Oguse, I attempted above (post 341) a simple privacy screen for a phone on top of my diffused light source and posted a video to show how effective this was. I then ordered some CineStill CS-LiteBrite+ sheets since that seemed purposely build to provide some collimation and increased brightness. I now tested these and frankly, the results are rather disappointing: Yes, it increases the light by some 25% or so, but it does not seem to provide any collimation (in contrast to CineStill claims). If anything, it makes the high contrast edges even worse (in my case). Here’s a video showing the effect to adding a CS-LiteBrite+ film on top of my light source:

In comparison, I re-tested also the phone “privacy screen” to make sure that what I showed previously was indeed correct:

Indeed, the high contrast edge definitions become less “smeared” using the privacy screen, but not so with the purpose-built CineStill sheets. So my conclusion is, short of doing proper collimation, to advice to purchase a phone privacy screen (they can be had pretty cheap for older generation phones) since this provides a cheap and simple way to increase the resolution when scanning with a diffused source. The one I used was made by JETech for an iPhone 6.1-inch and it seems a proper slab of glass (it had two of them in one box):

One detail that was asked above is whether it works in both directions: Yes, when it’s placed on my phone, it blocks the view in all side directions. These screens do reduce the brightness somewhat, but that is easily countered by a longer shutter speed.

EDIT: Read also my post below: Let's see your DSLR film scanning setup! - #356 by ArnoG

[Digitizer]

@ArnoG ,

does the protector improve apparent sharpness and how does that change depending on the distance?

I see a raster in both of your screen captures above. How susceptible to moire is your lighting setup? After all, there’s a stack of three rasters (iPhone/protector/sensor) that begs for moire.

[ArnoG]

@Digitizer My light is a custom built trichromatic RGB light, not an iPhone, and I scan with a Minolta 5400 scanner lens, in theory capable of 80 lp/mm, but my limit is around 60 lp/mm since I use a 24 MPix APSC sensor. My setup is described above:

Let's see your DSLR film scanning setup!

Hi all, I upgraded my scanning rig with a NEX7 APS-C body and a Minolta Dimage Scan Elite I scanner lens. [IMG_7822 2] [IMG_7823 2] [IMG_7826] Rationales and details can be found here: https://www.arnogodeke.com/Blog/Using-a-scanner-lens-for-DSLR-scanning

I did some more tests and what is seen in the videos is actually NOT a by me detectable increase in resolution, but a reduction in the brightness due to the privacy screen, which reduces the “glow” around the highlights and gave the impression of increased definition. In my current setup, I hit the limit of the sensor, and I cannot detect an increase in resolution due to the privacy screen, nor due to the CS screen. I will correct this in my video posts above. I do think that an increase in collimation (light directionality) should increase the resolution, but with my current 24 MPix sensor, I cannot confirm it. I would therefore appreciate it if others could try a privacy screen and/or CS screen to see whether it is indeed a cheap solution to increase resolution, but for this at least 36 MPix are needed. With my 24 MPix, I hit around 60 lp/mm maximum and, hence, cannot confirm the effect myself. The actual shot files look as follows (legend in the top left corner) at 300%:

As you can see, also increasing the distance has no visible effect on the resolution, but it does cause a larger light fall-off in the corners.

Inspired (again) by @Alain_Oguse I tried a fully collimated light as well (on a shoe string…), using collimator lenses from an old enlarger, and a microscope lens stuck on a flash light to create a point source:

The last pic is simply shot with my phone on the screen that I use for scanning. I noticed the following with (ad-hoc) fully collimated light:

1. It is extremely efficient with respect to light
2. As also @Alain_Oguse found out, everything is in focus (DOF is huge: I struggle still explaining the optics reason for that) and in order to focus with a collimated light source, one has to add a diffuser for focusing, which is then removed for taking the picture. This explains why with collimated light one picks up more dust and scratches.
3. One picks up EVERYTHING due to the huge DOF, and since the taking lens is focused on the point light, I also noticed whatever imperfections there are on the microscope lens. Not good.

Hence, imho, an ideal setup would start with a diffused light source, on which something is placed that makes the light more directional, to ensure that high contrast edges are better defined. Unfortunately, with my current 24 MPix sensor, I cannot verify whether this theory is indeed correct. A 36 MPix sensor is not available in APSC, so I would need to use a full-frame camera, at which point I start to significantly deviate from the magnification factor that my Minolta 5400 lens is designed for.

For now, I will simply stick the CS sheets on my diffused light source, since indeed each sheet does amplify the brightness by around 25% which is always a good thing.

[Digitizer]

@ArnoG

My light is a custom built trichromatic RGB light,

…and you’re experimenting with point light and condensers on a Fossil box
…which sparks the idea of using one of the spare lenses instead:

Obviously not aligned yet, but okay as a proof of concept.
Lighting: Desk lamp illuminating a wall 2 yards away.
Reversed the zoom for practical reasons.
Negative: Old, battered “Geo-Dreieck”, about 55 years old. The box (ARDA) is 3x14 mm.

[ArnoG]

@Digitizer let’s see if I understand this correctly: Diffuse light source into a lens focused at infinity, geo at “film plane” for the lens, so at focal point. So the lens picks up parallel light and focuses that onto the geo. The taking lens is focused on the geo and projects onto the sensor. I think I need a sketch to see where the beam paths go.