Scan Slides with Kodak Carousel Projector (Hack)

[Dan]

I see some dust problems above.

er, more like land erosion problems 😀

 

[ColSebastianMoran]

For this project, camera-scan of slides, choosing a light source to replace the 300w projector bulb.

The choice is basically between halogen and LED. Specifically, I have these options, all MR16:
- Solux 4700°K halogen (incandescent) bulb.
- Bulbrite 3000°K halogen bulb
- LED MR16 bulb, claiming 90+ CRI.

Anyone else have comparisons? I'm not seeing much of a difference.

 

[ColSebastianMoran]

Here are tests of the bulbs, shooting a Color Checker. Sony camera. Lightroom, Profile: Camera Neutral. (Other profiles give more saturation and punch).
- Left is the LED, claiming 90+ CRI
- Right is the Solux 4700°K halogen bulb, by definition, 100 CRI

LR eyedropper says the color temps are 3100°K and 4650°K, about as expected.

Both of these look good. I thought I would prefer the Solux, but these are close enough, I'm going to go with the LED to keep down the heat, and because the beam on the Solux is too narrow ("narrow flood" 24°). The Carousel projector needs a bulb with wider throw.

 

[predicolous]

wow impressive results.

I can't compare it to the color checker but from the image I'd rather go with the LED source. The blues and the orange seem to be more saturated and more to my liking.
Or is this rather due to the different exposure?

in either way it might be good to compare some "real" slides with both light sources : )

 

[ColSebastianMoran]

wow impressive results.

I can't compare it to the color checker but from the image I'd rather go with the LED source. The blues and the orange seem to be more saturated and more to my liking.
Or is this rather due to the different exposure?

in either way it might be good to compare some "real" slides with both light sources : )

First, the exposures were quite close. Only minor LR adjustments to Exp and Contrast.

I have the same questions. In my opinion, Kodak and photo finishers decided that consumers wanted more saturated and punchy colors. I found I was immediately inclined to prefer that in my test shots.

Note that this is with "Camera Neutral," for both bulbs. If you take most any of the other choices in LR, including the default "Adobe Color," you'll get even more sat & punchy.

My net of all this is that good LED bulbs are fine for color photography and should be fine for my Carousel-slide-scanner hack.

But, if you are going to try this at home, the easier route will be to find a 20w or 35w MR16 GU5.3 120v bulb that plugs right into the unmodified projector. These 120v bulbs aren't carried at your local store, but they are available.

More soon.

 

[dmr]

We (still) have an Ektagraphic slide projector in our AV closet. I might be tempted to try something like this, but everything I do currently is loose film, whether negative or reversal.

 

[Moto-Uno]

What a stellar idea 🙂 . Have a couple of projectors and crappy weather here , I might have a new project ! Thanks . Peter

 

[ColSebastianMoran]

Making good progress. Started seeing very ugly and unusual artifact in the middle of some slides.

Remember, I've removed the 250w bulb and replaced with a 50w equiv LED, so the slide lighting is two stops lower. Ugly artifact appears on under-exposed slides, bright band across any dark area in the middle.

The artifact is the projector's focusing lamp which shines a glancing blow across the face of the slide. Disabling the lamp cures the problem. Artifact is probably present on any slide, but far less noticeable in properly exposed slides.

 

[ColSebastianMoran]

I think my rig is now ready for production. Here's a sample, a Kodachrome from 1977, in the Grand Teton National Park. Camera-scan of my slide in my modified projector, shot with 24MPx Nikon D7200 and 105mm f/2.8 Micro-Nikkor. This is the straight out of camera jpg, except for reduced resolution.

From the RAW, I can tune this further, recover some shadow detail on the left and perhaps improve the clouds. I compose a bit wide, then crop in post. No HDR on this one, but I do find that some slides benefit from HDR. Matrix metering does a good job in this application.

Here's the example, straight out of camera.

[ EDIT: I had previously posted the full size file, and you'll see comments on that below. But, something was wrong; the resolution was flawed, more to come later. ]

 

[ColSebastianMoran]

Summary and how-to.

Basic:
- Find a projector. Check that "advance" works. This is the usual failing of Carousels.
- Remove the lens
- Add a diffuser to the light path. Careful, it can get hot in there! I used white plastic about 1mm thick, cover of a Sterilite storage container. See #37 above for positioning of diffuser.
- Disable the focus bulb (shines glancing light across the slide) by opening the twist-on electrical connector or cutting the wire. Tape the wire to avoid electrical contact with anything.
- Point your camera into the lens hole and shoot. It's very bright, 1/1000th at f/8.
- Use a 100mm macro lens on a crop body. The Carousel projector positions slides accurately for focus, left-right positioning varies just a bit. I prefer manual focus, but YMMV.
- Matrix metering works well. Some slides will benefit from bracketing & HDR.

Better -- Substitute a lower wattage bulb:
- You can find 50w 120v MR16 with GU5.3 base online. You want a broad flood, not narrow. Might have to thin the pins to make bulb fit in ceramic socket.
- Or, switch to 12v where you have both halogen and LED options. Rig 12v power to the bulb. 50w equivalent is about right. I have a CRI 95 LED wide flood brand name SORAA with "Vivid Light".

Anything else? I'm enjoying slides that I haven't looked at in 20 years!

 

[ColSebastianMoran]

One more. Again, Tetons, 1977 Kodachrome, Teton range from Jackson Lake Lodge. Some adjustments in LR. Reduced to screen resolution.

 

[pyeh]

Great detail, Colonel. Good results and pictures well worth digitising.
Thank you very much for sharing.

 

[predicolous]

it's so great to see how a simple idea can unfold and lead to amazing results.
Congratulations! I'm deeply impressed with the results

do you plan to digitize your whole slide archive with this set-up?
What about the "automatic" mode with both timer on the Carousel and the camera?

 

[dmr]

Here's a sample, a Kodachrome from 1977, in the Grand Teton National Park. Camera-scan of my slide in my modified projector, shot with 24MPx Nikon D7200 and 105mm f/2.8 Micro-Nikkor. This is the straight out of camera jpg.

I have to admit that the scan looks impressive! At least on the monitor here.

Have you tried printing any of your projector scans at 8x10 or larger?

Here's the example in full resolution, straight out of camera. If you view at 100%, you'll see that there isn't 24MPx of real data in the image; probably more like 6-10MPx.

However, viewing it full res, on this monitor at least, I don't see the detail of the dye clouds clearly.

When I scan a Kodachrome in full res on the admittedly-13-going-on-14 year old KM SD IV and inspect the .tif or the raw file closely I can clearly see the dye clouds ("apparent grain") in there in more or less crisp detail.

Multiplying things out, the scans I get are 15 million pixels or so.

However, one comparison I've made is to compare the scans I've done on slides and negatives to the shots from a digital camera of about the same resolution, which I happen to have, the Fuji HS30. Fuji claims 16 megapixels on that one.

In looking at prints of Kodachrome scans and HS30 images, at 8x10 they are very close. At 8.5x11, they still appear close, however I think the prints of the Kodachrome scans are a bit cleaner and show just a wee bit more information/detail.

When looking at larger, as in 13x19 prints, the largest my printer will do, the Kodachrome scans are clearly superior to those from the HS30. I admit that both need to be run through Neat Image to produce a 13x19 that will stand up to close viewing, and not all images from scans or from the HS30 will produce such a print.

My conclusion is that the Kodachrome scans are equivalent to at least a 10 megapixel digital image, probably more, as in 12-13 or so.

The reason I'm interested in your project is, as I (think I) said, that I know the film scanner I have is living on borrowed time. Eventually, it will die, or my "Media PC" will die and I will not be able to replace it with one that will drive the scanner. I'm thinking of going with some kind of "scanning" with a DSLR (or in my case, a mirrorless, the newer Fuji, the X-T20, which Fuji claims to be 24.{mumble} megapixels).

If I want to splurge, I may go with the "Toaster", except that's very expensive. The people at the Film Toaster booth at Photo Plus assured me that yes, with a good quality camera and lens, it will indeed resolve the real or apparent grain quite easily. I think that the point at which a "scan" clearly resolves the grain or dye clouds is the point at which we can say that all of the available information in the negative or slide has been captured.

 

[ColSebastianMoran]

... do you plan to digitize your whole slide archive with this set-up?
What about the "automatic" mode with both timer on the Carousel and the camera?

Yes, my whole archive. Starting with good images I've previous selected years ago.

No, to automatic. Here's why:
- I look at each image, place the AF focus point.
- Click with auto-exposure
- Check the histogram
- If needed, revise exposure or shoot a bracket different exposures.

Also, I am composing to get the full 35mm image plus a little. If a slide in the carousel is vertical, not horizontal, I will rotate it.

So, no to automatic.

 

[ColSebastianMoran]

However, viewing it full res, on this monitor at least, I don't see the detail of the dye clouds clearly...

Thank you for this comment. I don't see them either. [ UPDATE: 3/20/2019 @dmr is right, see updates below. ]

I'll investigate. Quick google says the dye clouds are 1.25 to 4 microns. I highly doubt I can resolve these in my 24MPx crop sensor body (pixel pitch 3.9 microns and my shots are at magnification 1:1.5). And, from the measurements at CoinImaging.com, it appears that my 105mm lens will only resolve details at about 8 microns (at this magnification).

Hard to decide how much detail to try to include in these camera scans. I have been guessing that 35mm chromes would deliver up to 10MPx, and I have 20+ in these camera-scans. I might try greater magnification and stitch to see if there is more.

Thanks for the question. Will do some more with this.

 

[dmr]

Thank you for this comment. I don't see them either.

Here is an example of what I am talking about. This is a crop of a very typical film scan, done at full res, which according to TFM is 3200 dpi.

Note that the grain/dye cloud/whatever is very obvious.

I found out the hard way that scanning some films at a lower res can cause "aliasing" of the grain/whatever and makes it appear far more prominent than what it should be.

 

[ColSebastianMoran]

Thanks for the example. Much appreciated.

 

[predicolous]

Thank you for this comment. I don't see them either.

I'll investigate. Quick google says the dye clouds are 1.25 to 4 microns. I highly doubt I can resolve these in my 24MPx crop sensor body (pixel pitch 3.9 microns and my shots are at magnification 1:1.5). And, from the measurements at CoinImaging.com, it appears that my 105mm lens will only resolve details at about 8 microns (at this magnification).

Hard to decide how much detail to try to include in these camera scans. I have been guessing that 35mm chromes would deliver up to 10MPx, and I have 20+ in these camera-scans. I might try greater magnification and stitch to see if there is more.

Thanks for the question. Will do some more with this.

I guess, when talking about megapixels we need to take the Bayer Array of the camera‘s Sensor into account.
This source suggests / calculates, that for a 10 MP line scan you would need 40MP digital camera resolution to get the same amount of detail in the red and blue channels.
But if I’m honest ... that’s beyond my knowledge (at the moment)haha

 

[ColSebastianMoran]

However, viewing it full res, on this monitor at least, I don't see the detail of the dye clouds clearly.

I now find that that capture wasn't optimal sharpness. Don't know why yet, but further exploration (different lens & body) shows I can get a lot more image detail out of that slide.