I'm making my own film scanner and will attempt to sell it

[mshchem]

Prototype is handsome!

 

[brbo]

That’s because at the end of the day, it doesn’t change how you process it or change the white balance on a raw file.

Frankly, I don't care for WB. It's the ability to set gain independently to individual photosites. Now, I don't follow the digital camera scene, but that's easily got to be worth at least 1 stop in DR even for a largely ISO invariant sensor. Astrophotographers must've went gaga when this stuff was introduced, yet I can't find anything about it...

 

[Richard Man]

Best of luck!!!

 

[Adrian Bacon]

Frankly, I don't care for WB. It's the ability to set gain independently to individual photosites. Now, I don't follow the digital camera scene, but that's easily got to be worth at least 1 stop in DR even for a largely ISO invariant sensor. Astrophotographers must've went gaga when this stuff was introduced, yet I can't find anything about it...

Again, they're probably not aware of it because you really only see that behavior on the extreme ends of the Kelvin tonal scale. As long as you're +-1500K (approximately) of the sensor's native white balance, it literally will never show up, and even if it did, unless you wrote your own raw processing program and was actually looking at the relevant values, you still probably wouldn't notice because it doesn't change what you have to do to get correct white balance. How much hoopla do you hear about digital cameras having a native white balance? The vast majority of photographers literally have no clue that there is in fact a point on the Kelvin scale where their digital camera is applying the minimum amount of digital gain to the red and blue channels to get correct white balance for that kelvin setting, and it is at that point that the digital camera has the most dynamic range and least amount of noise. The vast majority of photographers don't know and don't care because they just use Adobe Lightroom (or shoot jpg), and the performance is still good enough that they'd rather just shoot pictures and not worry about it.

 

[gswdh]

So I have an update!

I have a working scanner and I have the first images off of it off a roll of Delta 400. The images were taken on a dark day at around f2 to f2.8 on a Nikon F3 with a Nikon 50mm 1.8d Please bear in mind these are the very first images I've scanned, the focus isn't perfect and there's so much work to do but it is lovely to see images!!!

At the moment the images are ~2000dpi, 8 bit. The sensitive part of the line scanner is offset a few millimetres wasting some of the resolution, I just need to make something to shift is a reliable distance. There's also this strange vertical noise / distortion issue that I intuitively don't understand yet so any suggestions on that are very welcome!

At the moment my main focus (sorry) is sourcing a good lens at the right price. I think I have found one that can be sourced at €120 a piece at quantity and has a resolution working out to around 3000-4000dpi if you use the Nikon 50mm 1.8d as a reference for the MTF values.

Edit: Sorry for those interested in the exposure times, it works out to about 1ms or 1/1000th per line with the Nikon 50mm at f11. The light source is 5 dual LEDs running around 15W through a 5mm thick opal diffuser. I scanned a 36 frame (plus extra at the ends) roll of 35mm at 2000dpi and timed it to completion at 1m 45s.

 

[Adrian Bacon]

The lines likely is dust on your sensor, or a defective/unresponsive or miscalibrated sensel.

 

[koraks]

There's also this strange vertical noise / distortion issue that I intuitively don't understand yet so any suggestions on that are very welcome!

I assume you're talking about the staggered diagonals etc., and not the longitudinal lines? The latter I agree with @Adrian Bacon are probably due to something like dust. The former looks like a timing/synchronization issue between the film transport and the sensor sampling. It looks like you're sampling several times at the same position, then step a decent distance and then sample several times again.

 

[gswdh]

The lines likely is dust on your sensor, or a defective/unresponsive or miscalibrated sensel.

Yes I'm aware of that. I'm referring to what koraks mentioned.

 

[gswdh]

I assume you're talking about the staggered diagonals etc., and not the longitudinal lines? The latter I agree with @Adrian Bacon are probably due to something like dust. The former looks like a timing/synchronization issue between the film transport and the sensor sampling. It looks like you're sampling several times at the same position, then step a decent distance and then sample several times again.

Yes, maybe it could be the resolution of the stepper motor. Ages ago when I designed this board I used the DRV8825 stepper driver which is notoriously non linear when used in microstepping. I was already looking at a better alternative like the Trinamic controllers.

 

[albada]

Have you considered using a DC servo instead of a stepper? It has the advantages of being quieter and smaller, and probably cheaper.

 

[gswdh]

Have you considered using a DC servo instead of a stepper? It has the advantages of being quieter and smaller, and probably cheaper.

Are you talking about a 3 phase brushless with feedback? I'm not sure there's many around that operate at a very low RPM with good linearity. Maybe the would work well with a gearbox. Stepper motors and their drivers are already very cheap. You can get a NEMA 11 for around a few $ in quantity, the highest end driver is around $3 for 1k.

 

[koraks]

Maybe the would work well with a gearbox.

This will likely be the key anyway in getting good resolution, even if you use a fairly precise stepper.

 

[gswdh]

So I have verified the stepper motor non linearity is the issue. I calculate ~8.91 pixels pass per stepper motor step of 1.8 degrees, I have 1/32nd microstepping enabled. The average spacing between the vertical artefacts in the image are on average 8.5 pixels apart. A new stepper motor is on the way with a 27:1 gearbox reduction. Maybe it's not quite enough but it'll surely be better.

 

[gswdh]

The new stepper motor has fixed the issue. I now have a fully working 2000 dpi film scanner. Please follow the link below for the full resolution image, looks real nice.

Google Drive Image

The new design is in progress. I have a lens coming to me from Schneider-Kreuznach that looks like a great fit at the right price but I need to trial it.

I will post some pics of this new design soon. I intend it to be representative of what someone will be able to buy at some point. Looking at specs:

- 3750 or 5400 dpi depending on the sensor chosen.
- Around 10 minutes scan time for colour.
- 14 bit ADCs.

 

[Steven Lee]

@gswdh Congrats! The feeling of accomplishment must be pretty great.

 

[Romanko]

- 3750 or 5400 dpi depending on the sensor chosen.
- Around 10 minutes scan time for colour.
- 14 bit ADCs.

This is quite impressive! An equivalent camera scanning setup would require a 39 megapixel camera (assuming 5400 dpi) with a 14 bit sensor.

Is the scan time for a single 35 mm frame?

Please follow the link below for the full resolution image, looks real nice.

Love this image. Well done!

 

[gswdh]

This is quite impressive! An equivalent camera scanning setup would require a 39 megapixel camera (assuming 5400 dpi) with a 14 bit sensor.

Is the scan time for a single 35 mm frame?



Love this image. Well done!

A 10m scan time would be for 36 frames. This current prototype (2000dpi) does 36 frames in 1m 45s.

Thank you!

 

[Archiloque]

That's really great news. Are you still considering an infrared pass or - let's dream - an implementation of ICE ? Are they even still licensing their technology ?

 

[gswdh]

That's really great news. Are you still considering an infrared pass or - let's dream - an implementation of ICE ? Are they even still licensing their technology ?

I'm not sure yet. The sensors that are available have a BW channel with some IR sensitivity however I'm not sure if that'll work. There are few sensors on the market and the ones that are are limited. I've been doing my tests with a roll of film that's just been sitting on my desk, I give it a quick wipe before a scan and get very little dust. I wonder if a super simple and cost effective solution would be to use a film cleaner just before the scanning element. This is still an area of unknown and requires a fair amount of work.

 

[koraks]

I wonder if a super simple and cost effective solution would be to use a film cleaner just before the scanning element.

I'd be very hesitant to introduce additional elements that contact the emulsion side of the film. It might work well at first, but what happens after a while as materials age, get contaminated with debris etc.?

 

[Archiloque]

I'm not sure yet. The sensors that are available have a BW channel with some IR sensitivity however I'm not sure if that'll work. There are few sensors on the market and the ones that are are limited. I've been doing my tests with a roll of film that's just been sitting on my desk, I give it a quick wipe before a scan and get very little dust. I wonder if a super simple and cost effective solution would be to use a film cleaner just before the scanning element. This is still an area of unknown and requires a fair amount of work.

I understand, but if I may elaborate, the last main incentive to use scanners nowadays, in this era of camera scanning is ICE. I understand it doesn't work with b&w, but for color negatives it really does wonder.

Dust is like genes, some get a lucky draw, some don't. My place is filled from top to bottom with dust. I'd spend hours with the stamp tool in Lightroom if it wasn't for ICE.

The lack of IR dust detection would be a serious drawback for what appears to be the most promising scanner project.

 

[albada]

I wonder if a super simple and cost effective solution would be to use a film cleaner just before the scanning element. This is still an area of unknown and requires a fair amount of work.

You might not need ICE if you have (1) air blowing across both sides of the negative before the sensor, and (2) a diffuse light source. Based on my testing, I know that a diffuse light source hides scratches that my Nikon Coolscan IV ED outputs (in B&W) due to its collimated lighting. A blower will remove dust without touching the negative. But dust embedded in the emulsion (while drying) will not be removed this way.

Mark

 

[Mr Bill]

I wonder if a super simple and cost effective solution would be to use a film cleaner just before the scanning element. This is still an area of unknown and requires a fair amount of work.

I spent more than a few years in serious efforts to deal with dust on film in photo labs. It's not necessarily that simple.

A quick primer in case you're not familiar. Films are generally pretty susceptible to building up static electricity charges, and a lot of dust may be attracted to such charges. If one is in a low relative humidity environment it can be real difficult to get that dust off. Wiping with a brush may just move it around, etc., and there may be more dust just waiting to jump onto the film.

So first you wanna be in an environment with relative humidity around 50%, give or take 5 or 10%. (More than this could encourage rusting.) This will help tremendously with the static situation. It'll be much easier to brush or perhaps blow the dust off. In more severe situations it can be useful to use a "air ionizer" to help bleed the static charges off. In this situation some type of "film cleaner" can be pretty useful.

It's preferable to NOT wipe the film; this can induce a charge or even scratch film. It's real common to use some sort of compressed air or perhaps a hand blower for obvious dust particles. But... very tiny particulate does NOT blow off readily, if at all. The situation is not unlike using a hi-pressure washer (coin operated car wash) to wash your car; there may well be some very fine dirt down low that simply doesn't come off under the water stream alone. Same thing can happen on your film.

That said, the closest thing I've ever seen to a "magic bullet" for this sort of thing is something originally developed by Kodak called PTRs, for particle transfer rollers. They are slightly tacky and when rolled against film can generally take just about all the loose dust particles away from the film, even against a static charge. A lot of professional equipment was made using such rollers. The standard routine is to have the rollers offset from each other so that the film path gets a partial wrap around one roller then the other. Then, because of the possibility of the roller already having a particle stuck as it rolls over another particle on the film, it's usual to use 2 sets of rollers. When dirty just rinse with water and let dry.

I'd say it is well worth considering PTRs. I have no idea about the price or availability.

One last note about the SOURCE of debris. A great deal comes in on people. But a possible big issue is in the film guides - the part that keeps the film from moving side-to-side. Since the sides of the film have been slit to size there is the possibility of slightly rough edges where bits of gelatin may flake off. Especially if there is a slight burr on the guides. So you wanna be real careful with this. It may be worth using a couple rollers on the sides set slightly narrower than the guides to establish film tracking.

Best of luck

 

[Adrian Bacon]

Dust is like genes, some get a lucky draw, some don't. My place is filled from top to bottom with dust. I'd spend hours with the stamp tool in Lightroom if it wasn't for ICE.

One thing that helps a lot is running a large HEPA filter in the film areas. I run a large HEPA filter in my film processing area and another large HEPA filter in the film scanning area and have the fan speed turned up so that it circulates a lot of air. It does wonders for knocking down the total amount of dust that gets on your film. Coupled with regularly wiping down surfaces with a damp lint free cloth and covering equipment when not in use, dust just isn't that big of an issue. Yes, it's not completely gone, but it's down to having an occasional frame with a spec or two of dust to spot out.

 

[foc]

I wonder if a super simple and cost effective solution would be to use a film cleaner just before the scanning element.

There are antistatic units available like this.



I have used them for years and they worked perfectly. Removed the dust and the static charge from the negatives and never scratched or marked. They just needed a cleaning every evening.