Collimating light for more sharpness

Hi all,

I’m rather new to DSLR scanning but after getting some already impressive first results compared to my “professional” Minolta Dimage Scan Multi 2, and previous Minolta Dimage Scan Dual (both some 10-15 years ago) I’m getting enthusiastic. My recent switch back to film added motivation. Overall, I’ve been reading the very helpful discussions here, and anything else that could be helpful on the web, and am on a steep learning curve. After some playing around with my iPad, and a custom film guide stuck on opalescent glass, I decided to dive in and build a proper rig. It needs to be smart and cheap (…) but very good and convenient. So far, I’ve come to the following conclusions:

  1. Cheapest good copy stand is an old enlarger
  2. Cheapest good film holder is either from such an enlarger or gluing some film strips as guides on a glass plate and covering that with something flat with a window the size of the to-be-scanned area.
  3. Cheapest in-line dust clean solution are two antistatic carbon brushes for cleaning LP records mounted on top of each other through which film is pulled before entering the scan area. Additionally soft silicon “sticky rollers” could be implemented.

All straight forward and simple enough with many good examples already shown on the discussions here.

What I’m currently stuck at is the physics of the light source, and being a physicists myself, I suspect that I could come up with some solutions that make sense (perhaps…). What I’ve learned so far is that in old enlargers, one usually has two options: Using a “condensor/condenser” which, as far as I could distill, collimates the light source, or in normal words, makes the light travel perfectly parallel and fall perpendicular onto the negative. This will give the sharpest scan (or enlargement). Alternatively, as far as I understood, one would use a “diffuser” (frosted or opalescent glass) to create diffused light (light from all directions) falling onto the negative. This will result in a softer enlargement (and thus scan) with less sharp borders/transitions and softened film grain.

I assume that we all want the sharpest scans possible, so my conclusion was that a well collimated light source is best, which would render all LED panels and the like non-optimal. Hence, I’m embarking on route to generate a proper collimated light source.

One option would be to use the light source in an old enlarger together with its condensor. I don’t know how well collimated this light would be, but I assume that folks that built enlargers with condensors for many decades knew what they were doing. Alternatively, one could use any lens, place a point light source at the position where the film would be, put the lens on infinity, and what should come out of the front of the lens should be perfectly a perfectly parallel beam of light. The negative should then be placed on top the the front of this lens in the collimated light, held perfectly flat, and scanned.

Does this all make sense? Do LED and diffused iPad light sources intrinsically yield soft scans, and does a properly collimated light source indeed yield sharper results? Physics says that a collimated light source should indeed be better, but please chime in or tackle whatever I might have misinterpreted.



Hi @ArnoG and welcome to the forum

All you write is true, specially from a physicists point of view. And then, there is 20/80 or practicality or whatever we might call it…

Enlargers with condensers were used for B/W film mostly, while enlarging colour negatives was mostly done with diffused light. I’ve used both kinds of illumination in my tests with camera scanning and found that both work equally well and produce slightly different character/images. Condensed light produces “grittier” images well suited for B/W - and it makes every grain of dust shine like a lighthouse ;). BTW, if you get a condenser enlarger you’ll want to set it for best results:

  • use a negative and focus
  • replace the negative with a cardboard with a small hole (1/8-1/4in) and adjust the position of the light bulb in such a way as to produce a sharp image of the filament.

Most important (more than whether the light is collimated or not) is the quality of the light when digitising colour film. You’ve read about CRI and TLCI.
Blocking stray light is helpful too.

If you get an enlarger, try to find one that you can also use as a copy stand. This will enable you to use the often seen vertical setup with the camera pointing downwards.

Hi Digitizer,

Thanks for the inputs. Yes, I read about diffused light reproducing “better” the tonality in film, due the Callier effect from collimated light (more scattering at regions with higher silver particle density causing intensity differences), but that is valid for transmitted light onto a paper well beyond the negative. With DSLR scanning, we take a picture of the local brightness of the negative so the Callier effect should not be / be less relevant (???) and the original tonality of the film should be retained, no?

Is collimated light therefore technically not better than diffused light since it should produce a sharper scan? Isn’t the fact that dust is more visible not simply a result of the increased sharpness and contrast in the scan, and is that not technically better? Whether the dust is there in the first place is a different issue.

I must say that my best scans so far were with an iPad (CRI whatever since only B&W so far) and an opal-glass diffuser on 6x6. Tonality was indeed great and it seemed plenty sharp, but not so sharp that I could see film grain (Zeiss 100mm makro planar and a Nikon D4 sensor should resolve these).

So, yes, diffused light seems to work pretty well, but I am just wondering whether collimated should be technically better before I put together something proper. Once collimated light is available, it is easy to make that diffuse, but the other way around does not really work (though >1000 lines/mm gratings could help a bit).

There are also quite a few discussions on enlargers that want a point light source with a condenser to increase the sharpness and detail. For, say, 35mm scanning, this could easily be achieved with a point light source and a lens at infinity, or by using a smaller light source in an enlarger.

Another advantage of using an enlarger head with a condenser is that they usually use a frosted tungsten light bulb as far as I understood. These are continuous spectrum, contrasting all LED sources, even those with high CRI. Here in the EU, tungsten light bulbs are now illegal to be sold, but there must be some left and wattage is not so critical for scanning.

What you write on the cardboard with small hole means the the light bulb should be placed in the focal point of the condenser lens set, and then the light should be properly collimated?

Overall, the proof should be in the pudding. I’ll start looking for an enlarger with a condenser and put something together. I’ll follow up and update with a comparison of collimated and diffused scans, to see if theory holds.

Whatever you do, physics win. The question is, if it’s worth to go through all the effort.

Get the sun as a backlight and you’ll have it all, or at least almost. Remember to use a triplet to avoid chromatic aberrations and some heavy duty protective eyeglasses…

Here’s a simpler experiment: Lighting and Detail

Hi Arno
A couple of details occurred to me - in your film holder setup you mention a glass window to hold the film flat. You might need to ensure the film is held as flat as possible but also does not slide against any surface and scratch due to any debris on it…

If the iPad has a high colour index (and is therefore useable for colour negs or slides), then I wonder if the layer of iPad plus an acrylic diffuser on top would in fact be equivalent to a collimated source, due to the flat array of LEDs?


@Digitizer : thanks for link. Interesting differences with different light indeed. Perhaps the conclusion in the end would be that diffused light would work as a dust and scratch removal filter without loosing too much detail on the negative?

@Andy : Here’s how I created my opal glass diffuser to remove iPad screen details:

I sourced a sample piece of opal glass since apparently that isn’t made anymore these days. I wanted glass because it will not scratch and is smooth and flat. I then went to the hardware store and found a dark plastic cover for a light switch that happened to have a 5.5x5.5 cm square window (that is perfect for 6x6 negatives) and a flat front. I cleaned that up with a dremel, file, and knife and glued two strips of negative with tapered ends as side-guides on the glass with drops of superglue (=thin). The light switch cover was then glued on top with some drops of superglue and I could pull a strip of negative through that (emulsion side up) which is held perfectly flat this way. The edges of the light switch cover are rounded and the glass is a bit larger than the light switch cover, yielding a smooth entrance and exit. The whole setup costs €7, which was the price of the light switch cover. Cheap, simple and works very well. Results using a Zeiss 100mm f2 Makro Planar (planar=very flat field) and Nikon Df (D4 sensor) hanging off a tripod above this are:

The first two shots are just from my computer screen, with the zoom at 100%. The second two are the actual scans made with NLP with “some settings”. Overall, I was quite impressed compared to the scans I got from my Minolta Dimage Scan Multi ii, which were very crummy in comparison.

I now sourced a box of darkroom stuff from someone for €30, which includes a Meopta Opemus 5 enlarger that has a condenser. I’ll pick that up next weekend and plan to replace the large light bulb with a small car light to create a wide-spectrum tungsten point light source and a well collimated beam, which I can also diffuse with my little diffuser above and will make some clean comparisons that I’ll post here.

@Andy : Perhaps the iPad does create something that remotely represents collimated light, but the opal glass (and any other diffusing layer) will scatter the light and make it fully diffuse.

Yes, thinking about it the iPad plus diffuser would be flat intensity across the surface but locally would be diffuse. It will be interesting to see the results of your condenser vs collimated tests.

Unless you just like to tinker, all these issues have been addressed commercially with devices such as the Essential Film Holder, a purpose-engineered device sold worldwide by Englishman, Andrew Clifforth: << about the EFH

I used every sort of lamphouse in the photo industry over the course of 50+ years, 33 in labs. Condenser enlargers are best when you want B&W to look gritty. Diffusion dichroic enlargers are best for color, and for B&W variable contrast printing papers. Cold light sources are the softest for B&W, but don’t work for color.

Ergo, I use the diffuser built into my EFH, which I mounted in a 10” by 15” sheet of half-inch railroad board (black foam core mounting board). I cut a window under the EFH and mounted a Viltrox L-116t LED panel there, using Velcro strips. I raised the board up on three sides with L-shapes made from scrap 1x2 maple, and connected an AC adapter.

My 50+ year old negatives from high school yearbook photography never looked this good on silver halide paper! The EFH keeps film FLAT.

I have the 35mm and 120/6x9 film guides on the EFH. If I take off the 35mm guide plates that mount over the 6x9 aperture guide, I can Velcro my 35mm and 6x6cm Omega B-22 negative carriers on there, to handle single cut negs, or strips too short for the EFH.

The Callier Effect occurs at the silver emulsion and the scattering is random. Using a camera to image the emulsion thus illuminated will not eliminate the effect; highlights will pass less light to the imaging medium (enlarging paper, digital camera) in a non-linear function. The Callier Effect from collimated light scatters light passing through silver emulsions, and has a more pronounced effect as the density of the silver increases. Thus denser areas (negative highlights) are effectively “held back;” in a print this effect is reversed, thus producing harsh highlights artificially too light. We used to develop our film to a lower gamma to compensate for the Callier Effect, which meant less graininess due to lower silver density, more important in small size film (35mm) than larger formats. But the effect on tonality is some
harshness, which we now think of as the 35mm film “look.” My advice is to go with a diffused, high CRI light source and add a little more sharpening in post processing if you think you need it; there are many and subtle ways to do this intelligently. BTW, you cannot get to a 1:1 copy of a 35mm film with the 100mm Macro Planar and a full frame Nikon camera. It’s a great lens but limited by design. Adding tubes or close up attachments gives noticeably poorer results. Watch out for Newton rings with your glass set-up. Keep experimenting.

Hi Burkphoto,

Thanks for the inputs and for sharing your extensive enlarger experience. Yes, I read about the EFH and considered purchasing it, but ended up putting something together quickly since it was easy and since I did want opalescent glass as a diffuser. I have to admit that indeed I’m a bit of a tinkerer :wink: and am now planning to place a slide holder inside my 5.5x5.5 cm window to enable also 35mm film in the same configuration.

Being of a certain age, I grew up with film, but to my regret now, I never embarked on developing or enlarging myself. Eventually, I purchased my first computer for the sole purpose of using it as a digital darkroom, but was always disappointed with scans of my negatives, whether I scanned them myself or sent them out to have it done. Then digital cameras got fancy and I never looked back, but kept on wondering why some folks kept talking about negatives as “real RAW”. Fast forward to now, when my D700 broke and Nikon suggested I throw it away because it was 10 years and therefore “old”, which seriously upset me since my old trusty F801 was still working as new. Since I had “matured” with digital I was looking for some new excitement, and now have two 70 year old film cameras (a Zeiss Ikon Super Ikonta for MF and a Contax iia for 35mm) that needed some overhaul but now seem to be working as new again. This triggered the desire to get proper scans and here we are.

I think you hit the nail on the head by your statement that your 50-year old scans never looked this good, and I was similarly amazed by how good DSLR scans looked in comparison to the crummy self- and commercial scans that I have of my film stock. Hence, film still rocks, and as a bonus one gets to play with some wonderful historical cameras.

I’m still not sure where I’ll land with respect to lighting for scanning, but the inputs here seem to converge to diffuse light as default, with collimated light with its Callier disadvantage for gritty B&W. Hence, there might be no single “best” solution for all purposes.

I do appreciate very much everyone’s inputs as I’m learning a lot as part of the exercise and that is great fun.



Hi Jeffstev,

Thanks for your inputs.

I think you’re correct and the Callier effect will also be present with DSLR scans, so what I currently have might suffice as a good solution, though I still would want a tungsten continues spectrum light source as opposed to a discrete one, even when it’s high CRI, just for peace of mind. It shouldn’t be too complicated to arrange that.

With regards to 1:1 with my 100mm Zeiss: yes, I’m aware. So far I did my 35mm with an old Nikon slide copy adapter that I had in a drawer for some 25 years or so, combined with a Nikon EL 50mm 2.8 enlarger lens that I used for macro since it should be optimized for field curvature:

The results were not (yet?) as good as the scans from my MF film though so I’ll keep experimenting. I do have some old manual macro rings and bellows to get the Zeiss to 1:1 will see how that works.



This series of images illustrates what happens when the diffuser is moved between the light source and the “negative”. Between the light source and the negative, I put a diffusor and moved it from close to the light to close to the negative while the distance between the light and the negative remained a constant 25 cm. The practical side of this setup is, that I can change the looks of the negatives by simply moving a diffuser…

The leftmost image still shows the matrix of white and yellow LEDs from the backlight. Moving the diffusor towards the negative increasingly reduced the grittiness of the resulting image - without sharpness per se. The shots at left might look sharper but it’s looks only and nothing that cannot be compensated for with Lightroom’s sliders.

Here’s a photo that shows the whole rig:

That is an interesting comparison. I wonder what’s going on optically there. If I understand well, then your “negative” is a ermmm “geodreieck” (triangular plastic ruler in English perhaps?) that has a thickness substantially larger than the emulsion side of a negative? It seems that placing the diffuser, i.e. light source further away creates some shadows from the backside of the ruler. It reminds me of portrait shooting: A larger and/or closer light source provides “softer”, or better diffused, light, without evoking loss of sharpness, but with less specular highlights and softening of details. I’m not sure how that compares to perfectly collimated versus perfectly diffused light?

I was lucky to work with Kodak technical service reps in the late 1990s, early 2000s, as they developed high speed, high resolution scanners. Their Bremson HR-500+ was the pinnacle. It was a $50,000 beast that could scan film up to 70mm wide and 100’ long. We had nine of them in our lab in 2003, and 12 more in our other three labs. They scanned each school portrait in four seconds. By 2007, they were obsolete and we had ripped out our film processors!

Those moved film between a 6002-pixel sensor bar and light source. The software, DP-2, was a giant imaging database and rendering engine that ran on a Microsoft SQL Server. I think we are very lucky to be using Negative Lab Pro, because it does a similar job as that setup, a LOT more easily.

The HR-500 light source was a Tungsten-halogen lamp. The scanner had to be re-calibrated after each lamp change, which was NOT a fun task, as it involved “slope” adjustments for over- and under-exposed films. It was also incapable of self-adjusting for different film emulsions. We standardized our business around Portra 160 film, so that took care of that issue…

Anyway, the only issue I see with Tungsten filaments is that they yellow and dim as they age. NLP should compensate for the color shift, since you white balance the film base in Lightroom Classic. But you may see a gradual exposure shift over time. If you’re like me, and expose everything based on the film base density plus fog = black, there will be a slight adjustment between lamps.

I’m using my Viltrox light source at 100% brightness, 4400K setting, which uses all the yellow and all the blue LEDs at roughly equal intensity. That seems to provide the best color response on slides and color negative films. With black-and-white, it doesn’t really matter.

Yes, it’s easy to get lost in the black hole of post-processing. With LrC/NLP and Ps, I can do more than I knew how to do in a darkroom. Exposures I could never figure out how to print are usable now.

A source of limited size emitting diffused light emits collimated light when seen from far away…and yes, it’s a geodreieck, old battered brittle plastic too :sweat_smile:

No advantage in condenser light source.

I purchased and experimented with Omega B enlarger condensers for NLP/digitalCamera “scans” and found little to be gained. Obviously any difference in contrast between condenser and diffuse lighting can be eliminated in Lightroom. I expected to find more detail, that is, more information with the collimated light. I did not, at least with a 24Mpx Sony Nex-7.

The condenser lighted images did show horrible dirt and scratches on the filmstrip that were completely absent in the diffuse lighted images.

Digitizer I am a bit confused by your description of condenser proper focus. I focussed the lamp in the aperture plane of the objective lens. This description of a slide projector shows what I did:

The condenser set L1 with focal length f1 was focussed such that 
1/f1 = 1/u1 + 1/v1

Screen Shot 2021-07-21 at 5.49.13 PM

Think of taking a portrait with undiffused flash light. Your subject will probably not like it.

This is how I sort of remember focusing the projectors for slideshows. It’s quite a while since I did it last time though. It was all about getting the most and most even light onto the screen. Maybe that focus needed to be changed to show the filaments…try it out