DSLR Scanner: Light Sources

[PCC]

Daniel, if you want, PM me your snail mail address and I'll send you one, gratis, for testing.

[Daniel Moore]

How could I say no to that offer : ). PM on it's way.

Ludvig, who is no doubt occupied with his various and many projects, would have something to say about the value of high CRI. He went the mixing chamber route, which is viable with fewer LEDs at higher heat per unit.

I think Imatest could plot the frequency response of individual sensors. Hmm... I'll look into that. Their trial runs for 30 days, FYI. I've been holding off for something juicy to test it on.

Mapping illumination frequency/spectrum to optimum sensor receptiveness should have benefit.

[PCC]

@PCC, as far as I know, no one has tested such LEDs so the jury is still out on CRI requirements/benefits. I did find a ribbon style LED strip that purported to be high CRI but upon powering it up it began smoking and I had to pull the plug on 'em.

Heat is a factor. The LEDs you mention are high output. They will generate a lot of heat in an array layout. You'd need to experiment with the quantity/spacing/size of lamp/diffusion to get the most even illumination for the intended format size. Peter and I have been using RGB LED ribbons from Lee Valley, branded Armacost. They are tightly spaced and low output so not that much heat to deal with. We can also dial in a neutral light relevant to the film type, compensating for color neg mask, tinted B&W, etc. if desired. I don't have any comparisons done with other LED types unfortunately. It's unclear at this stage also how significant CRI actually is to DSLR scanning. I did do a spectral response test and found that the lamp from my much loathed V700 (personal bias? : ) scanner had a pretty psychedelic set of peaks and valleys. Nothing like a black body spectrum such as an incandescent bulb. Not exactly scientific data I'm presenting here, is it . Maybe it's of use.

If you are using this for black and white copying then it's not going to make a difference what CRI the light source puts out. For color you need as much as possible. From personal experience comparing a standard white LED to a high CRI Nichia the reds stand out far better. You'd think that a white LED, which should be representative of all colors , would show all colors equally well but I am not seeing that. Most low CRI LEDs (65-ish CRI) are really blue LEDs with a phosphor that converts the blue light into white light and this white light is very weak on the red side of the spectrum. Red colors appear dull to my eyes with those LEDs while a high CRI LED shows them quite well.

Most white LEDs require approximately 3.2VDC at up to 1.5 amps except for those silly little 5mm ones that cannot be attached to a heatsink. Above 350 milliamps heatsinks are essential to long LED life. Below this drive level you're relatively safe. Heat is still being generated, but it's far less than at higher drive levels. Obviously, higher drive levels give you more light but these LEDs are at their most efficient at around 50mA or so (this varies from model to model).

Incandescent lights are 100CRI. They're terribly inefficient and produces heat in the form of infrared radiation. LEDs produce heat through its contact patch on the MCPCB that it's attaches to. The light has almost no infrared radiation. You can put your hand in front of a 2000 lumen flashlight and it's warm. A 1000 lumen incandescent can light paper on fire at close range to the lens. In a light box I'd rather run a copper strip to pull heat away from the emitters than to have the heat come through the negative that you are scanning. The entire box will heat up from the bulb.

Using a high CRI LED will not give you the color saturation that an incandescent bulb will give you but it won't be bad, neither. A standard low CRI LED will give dull reds and nice blues that you might be able to salvage in post but I'd like to avoid that as much as possible.

[ludvig friberg]

As I have stated before I have found the best method to be a Integrating sphere. That makes it much easier to have many different lightsources for testing and also makes it possible to mix them for different media. It would be interesting to test something like solux halogens against led (http://www.solux.net/cgi-bin/tlistor...olux-more.html). Biggest problem with normal incandescent is for negative color film, the orange cast together with a warm light makes it very problematic without heavy filters. Also it would be interesting to take 10 or so different white high cri leds and mix them in a sphere to see if perhaps they even out each other. One other nice thing with led is the possibility to balance the r,g and b to the native response in the sensor. Meaning less noise in channel x. And to use it for exposure time like a flash, that way I can easy capture HDRI, I do that for my motion film scanners. And if wanted a separate IR channel can be made if the camera is modified, however that changes the focus and makes everything a bit more challenging.

[Daniel Moore]

Back on the 26th of Oct 2012, in this thread (yikes!) Struan Gray wrote:

--------------------------------
"The beauty of LEDs is that you can choose three wavelengths which as much as possible avoid cross-contamination by avoiding the portions of the dye absorbance spectra which overlap. You then adjust the brightness of each type of LED (with modulation, or by adding more or less LEDs to the source) so that each channel gets as much light as it can take without being over-exposed. Post-capture you adjust the strength of the relevant channels *downwards* to balance for grey and overall gamma.

If the spectral absorbances of the dyes in the film overlap at all wavelengths, there will always be *some* cross contamination. But with better signal to noise in the blue channel you are in a better position to do something about that with further processing.

The central point is that just because colour negative film was optimised for analogue printing with an incandescent light source, it doesn't mean that will be the best light source for capture with a DSLR."

------------------------------

This echoes Ludvigs statement above "One other nice thing with led is the possibility to balance the r,g and b to the native response in the sensor."

I'm not suggesting the RGB LEDs Peter D. and I are currently employing are that, specifically, tuned to a DSLR sensor, but I do want to suggest that high CRI may not be the primary target for DSLR scanning. In fact, I have read about the poor color response to tungsten lighting that DSLRs can suffer. I can't find that info at the moment (there is certainly such a thing as too many bookmarks) but I've seen it for myself. Black body spectrum might be perfect for color slide film but very likely isn't perfect for other film types.

Mind you, I'm no scientist. But even they can't always agree : ).

Final note for the night. Anyone interested in developing a light source would do themselve a huge favor by reading the absolutely phenomenal input made by some very knowledgeable folk through out this thread. Start at the beginning, it's a hell of a ride, as I reminded myself for the last two hours : ).

[drtebi]

As I have stated before I have found the best method to be a Integrating sphere. That makes it much easier to have many different lightsources for testing and also makes it possible to mix them for different media. It would be interesting to test something like solux halogens against led (http://www.solux.net/cgi-bin/tlistor...olux-more.html). Biggest problem with normal incandescent is for negative color film, the orange cast together with a warm light makes it very problematic without heavy filters. Also it would be interesting to take 10 or so different white high cri leds and mix them in a sphere to see if perhaps they even out each other. One other nice thing with led is the possibility to balance the r,g and b to the native response in the sensor. Meaning less noise in channel x. And to use it for exposure time like a flash, that way I can easy capture HDRI, I do that for my motion film scanners. And if wanted a separate IR channel can be made if the camera is modified, however that changes the focus and makes everything a bit more challenging.

Regarding Solux Halogen bulbs:
I have built a DSLR Scanner with a lightbox that uses a Solux Halogen bulb. I am quite satisfied with the results. The colors are great, a bit on the warm side, but that's easy to edit out (e.g. much easier than a color cast).

What is much more difficult with a halogen light source is to get an even illumination. Eventually I decided to solve that problem with software. At each scanning "session" I also take an image of the light source alone, with no negative/positive. I then use this as a layer in PhotoLine (similar to Photoshop) and set the layer to "divide". That layer usually also needs a bit of exposure adjustment, to avoid overexposure after the division adjustment.

A similar technique was mentioned in this thread here:
http://www.largeformatphotography.in...=1#post1146865

Here are a few links to my results:

Fujifilm Reala 100 shot with Fuji GA670: http://1drv.ms/1TIa37L

Fujifilm Superia 200 shot with Yashica Electro 35 GSN: http://1drv.ms/1PguOEx

Fujifilm Reala 100 shot with Plaubel Makina 670: https://www.flickr.com/photos/drtebi...57653529648714

And here a couple of shots taken with the Plaubel Makina 670 on Fujifilm Provia 100F:
https://www.flickr.com/photos/drtebi/19044135543/
https://www.flickr.com/photos/drtebi/18980441304/

My DSLR is a Nikon D810, I use the ColorPerfect plugin for negatives, and I do a tiny bit of post editing in LightZone. When comparing my results to those of my Minolta Dimage Multi Pro film scanner, I have to say the DSLR does a much better job. The scanner can still resolve more detail, but the light source of the scanner is definitely not as nice as the halogen bulb. I also get a lot less grainy-looking images, which I believe is due to the better diffusion of the light with the DSLR setup.

Please don't hesitate to criticize the outcome of these images, technically speaking (I am no professional photographer...). I would love to hear what you think.

[Peter De Smidt]

Charles Krebs has a good DIY write-up on making high CRI LED fixtures. See: http://www.photomacrography.net/foru...=27265&start=0

LED is a Bridgelux COB, "Decor Ultra Vero 10", with a CRI of 97, part number: BXRC-30H1000-B-2 .

[Joseph_B]

Hello thanks for adding me to the forum - I hope this thread is still active, so I won't make a new one to ask this. I have a couple of questions to ask you guys, as I have just embarked on this process myself.

My setup: Nikon D810 - Tokina 100mm macro - Kaiser Rs2 copy stand - Artograph A920 LED light pad - 6x7 neg carrier.

My thoughts findings so far - I have found that I easily clip the red channel in the camera when taking the photograph on the DSLR of the negative. It makes me wonder would using a blue thread on filter on the lens help? I also assume that clipping the red channel gives you less room to manoveure later when you are trying to convert your image into a positive using color perfect, or photoshop or whatever.

If I used a filter, I think that I would reduce light by a couple of stops, meaning that I would have to have a longer exposure? Is this a problem? I am already at about 1/5s a second, iso 64 - shooting on the copy stand

And does a flash give less peaks across the spectrum than an LED light?

Does anyone think that the LED light source that I am using is not as suitable as using flash and a diffusion box. With flash, the benefits would probably be a stronger light source and so better exposure times (minimising potential vibration, or trouble moving around?) But I think flash might give more of an uneven light spread compared to the Lightbox?

To recap
- Should I use a filter on my DSLR for capture?
- Will a flash work better than an LED lightbox
- What software do you guys use to conver? ColorNeg by ColorPerfect, or Photoshop, or Vuescan?

Cheers,

[drtebi]

- Should I use a filter on my DSLR for capture?

I don't think that would be a good idea. You will probably make the conversion process more difficult than without.
I am using ColorPerfect as well, and am getting really nice results. I am using a Solux halogen light bulb though, and with my D810 I am setting the color temperature manually to 4700K.

- Will a flash work better than an LED lightbox

I haven't tried a flash, but in my earlier experiments I did use an LED lightbox similar to yours. Then I did a lot of reading about light sources, and my conclusion was, that a "black body" (edison-type light bulb, halogen lightbulb etc.) would give me the best colors. Especially the warmer colors of the spectrum are looking much more natural to me with a halogen bulb.

It is a bit tricky to evenly diffuse the light of a halogen light bulb, but it is possible to remove any irregularities in post processing. You can do this by taking an image of the light source alone, and then put that image as a layer on top of your negative image. Then put the light source image layer into "Divide" and, if necessary, adjust the histogram of the light source layer to avoid over-exposure. This technique works great for me, here are some examples:
6×7: http://1drv.ms/1TIa37L
35mm: http://1drv.ms/1PguOEx

But if you want to stick with the LED lightpad, I think the most crucial thing is, to set the color balance right. Maybe try to do that with a positive (slide), which will make it easier to judge the color balance.

I also noticed that not all film is easy to scan. While I got great results with Fujifilm Reala, Kodak Portra 160 looked really off.

- What software do you guys use to conver? ColorNeg by ColorPerfect, or Photoshop, or Vuescan?

I use ColorPerfect. It usually gives me great results without changing any of its settings (except the film type).

I have tried many other methods to convert a negative to a positive, and while I did get nice results at times, I could never repeat the same process on other negatives to get consistent results. Especially not once you change to a different film. I really wanted to use my own custom conversion, or just another program, but I gave up... ColorPerfect just does it better and faster than anything else I have tried.

By the way, ColorPerfect also works with PhotoLine. PhotoLine is more or less like Photoshop, but costs much less.

For final post editing (e.g. after the conversion), I use LightZone. I really like that program despite the couple of quirks it has. It makes post editing very quick, and I find the modules very smart and powerful, yet easy to use. It's not for everyone, but worth a try. Oh, and it works on Linux, too (Java-based).


Another thing I would like to mention... stitching can get you more detail if you get it right. I have my lightbox mounted on an X-Y table, and some images I scan 2×2 and then stitch the four pictures together with Image Composite Editor, which is free, easy to use, and most of the time works just perfectly.

Here is a picture of my setup:


If you would like more info on my setup, just ask here.

Cheers,
DrTebi

[ropel]

I have been following along on this topic - quite a read by now :-)

I'm looking to work towards a scanner setup as well, hopefully including a nice XY table.
My current set-up is a Canon FD bellows with duplicator, Canon FD 50mm macro lens and Sony A7 attached.
Playing around with ColorPerfect, which so far seems a very good piece of software indeed.

So many messages on the light sources that I am not quite sure what direction to go.
Lots of posts aim at high CRI, but since I'm primarily focussing on B&W, that probably is not my biggest concern.
Currently I'm using my tablet, and that makes me think in the direction of a lightbox (e.g., http://www.kaiser-fototechnik.de/en/...ge.asp?nr=2447)
Seems a good option for even illumination, but I'm unsure if the light output will be sufficient? (My tablet certainly is not, slow shutter speeds > 1 sec required to get the shot at f8).
Can anyone share experience in lightbox vs tablet light output?

If a lightbox turns out to be insufficient, I'm in doubt about halogen vs LED.
Any recommendation for B&W setup at reasonable price?

Thanks!

Roy