Use light box to "scan" slides

sbbtim · 01-01-2010, 11:27 AM

I really like what the slides look like in a light box. So last night I tried to "scan" the slide on top of the light box. A tripod, my K-x, and DFA 100mm/2.8 macro are used for that purpose. The results look great. Minimum PP is applied to the following pictures.

brkl · 01-01-2010, 11:47 AM

Those are good results. I might be wrong, but judging by the sides you could have used a lower ISO to minimize loss of quality. What kind of resolution can you get from these?

This should be possible with negatives as well. Fiddling with the colours can be difficult, but it's doable.

sbbtim · 01-01-2010, 12:15 PM

If I understand you correctly, the output resolution could be up to your digital camera's resolution. In my case, I can make it 12M pixels (K-x).

ytterbium · 01-01-2010, 12:40 PM

Very well. I think in such setup it is the best that you can get - light box menans good light source, and macro lens is the best there can be for such purpouse. Even better than those tunnel adapters with think light diffusion glass, dependant on illumination source and cheap macro lens.

The only downside i could think of would be stray light reflecting of the surface of film - like reflections from lens, light sources in the room and similar.

As far as i can tell, very good clarity and colour reproduction.

Talking about negatives, i think the dyamic range of camera could be a problem - because you need a lot of headroom, to remove the orange cast and expand the contrast afterwards. Otherwise working from 12bit raw or 8 bit jpeg may leave you with something like 4bit equivalent DR. Probably true RAW processing is a must.
Im not sure, but i think many scanners have true RGB data for each pixel aswell (monochrome sensor, lit sequentially by R, G, B light at each pixel) - cameras use BAYER.

lithos · 01-02-2010, 02:21 AM

Date stamp! Arrrrgh!

brkl · 01-02-2010, 09:52 AM

You can always do HDR to get all of the dynamic range.

I was wondering what the actual resolution you were getting was, sbbtim. What's the smallest detail you can see? I'd love to see a 100% crop of the sharpest example you can find.

ytterbium · 01-02-2010, 10:29 AM

Originally posted by brkl

You can always do HDR to get all of the dynamic range.

I was wondering what the actual resolution you were getting was, sbbtim. What's the smallest detail you can see? I'd love to see a 100% crop of the sharpest example you can find.

Dynamic range does not always relate directly to the maximum limits as it is with HDR. It is more like a relation between the minimum "step" compared to the maximum range.
Multiple exposures help to cover large range of brightness, but it will be very hard or impossible to increase the smallest resolvable brightness level difference in such way.
If a JPEG is 0-255 grays, and you have noise in the order of 10, you use 200 to subtract the orange cast of negative, you're left with 255-10-200=45 levels of gray with 10/55*100=18% noise level. (Those are not actual calculations, just an example to illustrate the idea).

If you capture two exposures, where the second exposure noise level (10) matches the point where the first one gets blown - you'd get 255+(255-10)=499 level grayscale HDR. You'd still have the same bit depth per brightness difference, the new bits would be covering only additional brightness levels.

brkl · 01-02-2010, 11:20 AM

I don't know, I think you're overestimating the effect of the orange cast. It does make a big difference, but when I was scanning C41, at least 40% and probably more of the image had proper colour data. I don't think you'd ever have to throw away four fifths.

But yeah, okay, you might have a point. I wouldn't know until I tried it. But if it was doable, I would just love decent versions of my film photos among my digital photos in Lightroom. The scans I'm doing now are basically only good enough for Facebook and take a long time to do.

ytterbium · 01-02-2010, 11:33 AM

I wonder what happens if you use the light source trough a cyan filter.

Anyway, this method of image conversion has a huge advantage in conversion speed, which can be important when converting large quantities of film. After you've set everything up, you just feed the film trough and snap away. With conventional scanner, there are various delays - preview, frame detection and the scanning process itself is very slow. A DSLR - 3 and more fps if you can keep up feeding the film

.

Btw, can someone recommend a decent adapter for this kind of conversion, preferably a one that "eats" 120 film and doesn't cost the same as flat bed scanner.

Vertex Ninja · 01-02-2010, 12:18 PM

Originally posted by ytterbium

...
Im not sure, but i think many scanners have true RGB data for each pixel aswell (monochrome sensor, lit sequentially by R, G, B light at each pixel) - cameras use BAYER.

This is true. A CCD scanner is just like a camera with a microscope or macro lens attached, but with a true 3 line RGB sensor without Bayer interpolation. If cameras did not have this interpolation or an AA filter AND had a lens with a perfectly flat field and the sensor had sufficient dynamic range, it would be possible to get quality equal to the best CCD scanners... but that's a lot of "ifs". Still, even with all these stacked against it, I'd imagine your camera should get quality better than most consumer or prosumer flatbeds.

If you wanted increased resolution you could increase magnification and using a moving copy stand, take multiple images and combine them in photoshop.

sbbtim · 01-02-2010, 12:31 PM

I attached two pics here...the first one is the original, and then a 100% crop.
I could put the slide closer to the camera, if I would try to squeeze more details.

Originally posted by brkl

You can always do HDR to get all of the dynamic range.

I was wondering what the actual resolution you were getting was, sbbtim. What's the smallest detail you can see? I'd love to see a 100% crop of the sharpest example you can find.

architorture · 01-03-2010, 01:01 PM

I tried this same thing a little while ago. It is of course true that your output size is limited by your camera resolution (so for 24x36mm film I get at best 10MP with my k10d versus potentially ~30MP with a scanner) and flatness of film and lens field are crucial.

However, the potential is very high - at slightly greater than 1:1 with a 100mm macro lens and extension tube I found that the detail captured was greater than my Epson v700 could deliver (if I recall correctly, the two were at ~equivalent size - though not the full image, the macro shot was about the size of a 3200dpi scan). And scanner software is not as easy to use as a raw file in ACR.

My big curiosity is what you would get with this technique and a medium-format digital back & view camera. That would be impressive - though of course no longer cheaper than a high-quality scanner such as an Imacon or drum scanner.

Vertex Ninja · 01-06-2010, 10:13 AM

Originally posted by architorture

...
However, the potential is very high - at slightly greater than 1:1 with a 100mm macro lens and extension tube I found that the detail captured was greater than my Epson v700 could deliver (if I recall correctly, the two were at ~equivalent size - though not the full image, the macro shot was about the size of a 3200dpi scan)...

This sounds exactly right, as it's been proven that the V700/50 don't get more than 2100-2600 SPI of real resolution even in the best of conditions. It's a real shame that scanner development has come essentially to a standstill. Professional flatbeds were getting 2-3x this resolution 10 years ago. I wonder what happened to the trickle down of technology?

brkl · 01-06-2010, 02:31 PM

Those professional scanners probably didn't get the resolution they reported either, but yeah, this stuff should be easy and cheap by now, but it really isn't.

Vertex Ninja · 01-06-2010, 02:49 PM

They got/get 95-99% of what they advertised and are still being manufactured relatively unchanged. Fudging numbers wouldn't fly in the professional market.
Other than the cost of optics everything should be significantly cheaper to make, but without real demand or competition it just never trickled down.

01-01-2010, 11:27 AM	#1
sbbtim Senior Member Join Date: Oct 2009 Photos: Albums Posts: 125	Use light box to "scan" slides I really like what the slides look like in a light box. So last night I tried to "scan" the slide on top of the light box. A tripod, my K-x, and DFA 100mm/2.8 macro are used for that purpose. The results look great. Minimum PP is applied to the following pictures.

01-02-2010, 12:18 PM	#10
Vertex Ninja Inactive Account Join Date: May 2007 Location: Indianapolis, Indiana Posts: 652	Originally posted by ytterbium ... Im not sure, but i think many scanners have true RGB data for each pixel aswell (monochrome sensor, lit sequentially by R, G, B light at each pixel) - cameras use BAYER. This is true. A CCD scanner is just like a camera with a microscope or macro lens attached, but with a true 3 line RGB sensor without Bayer interpolation. If cameras did not have this interpolation or an AA filter AND had a lens with a perfectly flat field and the sensor had sufficient dynamic range, it would be possible to get quality equal to the best CCD scanners... but that's a lot of "ifs". Still, even with all these stacked against it, I'd imagine your camera should get quality better than most consumer or prosumer flatbeds. If you wanted increased resolution you could increase magnification and using a moving copy stand, take multiple images and combine them in photoshop. Last edited by Vertex Ninja; 01-02-2010 at 12:35 PM.

01-03-2010, 01:01 PM	#12
architorture Senior Member Join Date: Apr 2008 Location: New York, NY Posts: 152	I tried this same thing a little while ago. It is of course true that your output size is limited by your camera resolution (so for 24x36mm film I get at best 10MP with my k10d versus potentially ~30MP with a scanner) and flatness of film and lens field are crucial. However, the potential is very high - at slightly greater than 1:1 with a 100mm macro lens and extension tube I found that the detail captured was greater than my Epson v700 could deliver (if I recall correctly, the two were at ~equivalent size - though not the full image, the macro shot was about the size of a 3200dpi scan). And scanner software is not as easy to use as a raw file in ACR. My big curiosity is what you would get with this technique and a medium-format digital back & view camera. That would be impressive - though of course no longer cheaper than a high-quality scanner such as an Imacon or drum scanner. Last edited by architorture; 01-03-2010 at 01:02 PM. Reason: add info

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Slides - Scan or Copy?	Khukri	Photographic Technique	20	09-12-2014 07:17 PM
Amazon "Open Box" Lenses	brofkand	Pentax SLR Lens Discussion	17	10-06-2010 12:57 AM
treasure trove. the box of "vintage Pentax" items	CEWren	Pentax SLR Lens Discussion	22	09-02-2010 12:03 AM
DSLR advice from one of those "other" photo forums (also, Pentax and low-light)	shuttervox	Pentax DSLR Discussion	38	02-25-2009 05:50 PM
"Hunger for a DA50-135?" or "The DA50-135 as a bird lens!" or "Iron age birds?"	Douglas_of_Sweden	Post Your Photos!	4	08-13-2008 06:09 AM

01-01-2010, 11:47 AM	#2
brkl Veteran Member Join Date: Oct 2008 Posts: 497	Those are good results. I might be wrong, but judging by the sides you could have used a lower ISO to minimize loss of quality. What kind of resolution can you get from these? This should be possible with negatives as well. Fiddling with the colours can be difficult, but it's doable.

01-01-2010, 12:15 PM	#3
sbbtim Senior Member Join Date: Oct 2009 Photos: Albums Posts: 125 Original Poster	If I understand you correctly, the output resolution could be up to your digital camera's resolution. In my case, I can make it 12M pixels (K-x).

01-01-2010, 12:40 PM	#4
ytterbium Veteran Member Join Date: Jan 2008 Posts: 1,076	Very well. I think in such setup it is the best that you can get - light box menans good light source, and macro lens is the best there can be for such purpouse. Even better than those tunnel adapters with think light diffusion glass, dependant on illumination source and cheap macro lens. The only downside i could think of would be stray light reflecting of the surface of film - like reflections from lens, light sources in the room and similar. As far as i can tell, very good clarity and colour reproduction. Talking about negatives, i think the dyamic range of camera could be a problem - because you need a lot of headroom, to remove the orange cast and expand the contrast afterwards. Otherwise working from 12bit raw or 8 bit jpeg may leave you with something like 4bit equivalent DR. Probably true RAW processing is a must. Im not sure, but i think many scanners have true RGB data for each pixel aswell (monochrome sensor, lit sequentially by R, G, B light at each pixel) - cameras use BAYER.

01-02-2010, 02:21 AM	#5
lithos Inactive Account Join Date: Oct 2007 Location: Brisbane, QLD, AUS Posts: 3,261	Date stamp! Arrrrgh!

01-02-2010, 09:52 AM	#6
brkl Veteran Member Join Date: Oct 2008 Posts: 497	You can always do HDR to get all of the dynamic range. I was wondering what the actual resolution you were getting was, sbbtim. What's the smallest detail you can see? I'd love to see a 100% crop of the sharpest example you can find.

01-02-2010, 10:29 AM	#7
ytterbium Veteran Member Join Date: Jan 2008 Posts: 1,076	Originally posted by brkl You can always do HDR to get all of the dynamic range. I was wondering what the actual resolution you were getting was, sbbtim. What's the smallest detail you can see? I'd love to see a 100% crop of the sharpest example you can find. Dynamic range does not always relate directly to the maximum limits as it is with HDR. It is more like a relation between the minimum "step" compared to the maximum range. Multiple exposures help to cover large range of brightness, but it will be very hard or impossible to increase the smallest resolvable brightness level difference in such way. If a JPEG is 0-255 grays, and you have noise in the order of 10, you use 200 to subtract the orange cast of negative, you're left with 255-10-200=45 levels of gray with 10/55*100=18% noise level. (Those are not actual calculations, just an example to illustrate the idea). If you capture two exposures, where the second exposure noise level (10) matches the point where the first one gets blown - you'd get 255+(255-10)=499 level grayscale HDR. You'd still have the same bit depth per brightness difference, the new bits would be covering only additional brightness levels.

01-02-2010, 11:20 AM	#8
brkl Veteran Member Join Date: Oct 2008 Posts: 497	I don't know, I think you're overestimating the effect of the orange cast. It does make a big difference, but when I was scanning C41, at least 40% and probably more of the image had proper colour data. I don't think you'd ever have to throw away four fifths. But yeah, okay, you might have a point. I wouldn't know until I tried it. But if it was doable, I would just love decent versions of my film photos among my digital photos in Lightroom. The scans I'm doing now are basically only good enough for Facebook and take a long time to do.

01-02-2010, 11:33 AM	#9
ytterbium Veteran Member Join Date: Jan 2008 Posts: 1,076	I wonder what happens if you use the light source trough a cyan filter. Anyway, this method of image conversion has a huge advantage in conversion speed, which can be important when converting large quantities of film. After you've set everything up, you just feed the film trough and snap away. With conventional scanner, there are various delays - preview, frame detection and the scanning process itself is very slow. A DSLR - 3 and more fps if you can keep up feeding the film . Btw, can someone recommend a decent adapter for this kind of conversion, preferably a one that "eats" 120 film and doesn't cost the same as flat bed scanner.

01-06-2010, 10:13 AM	#13
Vertex Ninja Inactive Account Join Date: May 2007 Location: Indianapolis, Indiana Posts: 652	Originally posted by architorture ... However, the potential is very high - at slightly greater than 1:1 with a 100mm macro lens and extension tube I found that the detail captured was greater than my Epson v700 could deliver (if I recall correctly, the two were at ~equivalent size - though not the full image, the macro shot was about the size of a 3200dpi scan)... This sounds exactly right, as it's been proven that the V700/50 don't get more than 2100-2600 SPI of real resolution even in the best of conditions. It's a real shame that scanner development has come essentially to a standstill. Professional flatbeds were getting 2-3x this resolution 10 years ago. I wonder what happened to the trickle down of technology?

01-06-2010, 02:31 PM	#14
brkl Veteran Member Join Date: Oct 2008 Posts: 497	Those professional scanners probably didn't get the resolution they reported either, but yeah, this stuff should be easy and cheap by now, but it really isn't.

Thread Tools	Search this Thread
	Search this Thread: Advanced Search

01-06-2010, 02:49 PM	#15
Vertex Ninja Inactive Account Join Date: May 2007 Location: Indianapolis, Indiana Posts: 652	They got/get 95-99% of what they advertised and are still being manufactured relatively unchanged. Fudging numbers wouldn't fly in the professional market. Other than the cost of optics everything should be significantly cheaper to make, but without real demand or competition it just never trickled down.