Noll
Well-known
There's been a fair amount of chatter here about using digital cameras to scan negatives and how they are better/worse than the alternatives. Since I've been using all three of these of these devices for scanning film, I thought I'd post a comparison showing the resulting differences in output. I am using these items mostly on their default settings. My perspective being that I really just want things to work as they come. I don't have the time to buy different software or do tons of post-processing.
This test shot was taken with an OM1 and 35-70 3.5-4.5 zoom at close focus. Film is Arista Premium 400 exposed at 400 and developed in Caffenol CH. It was a thin negative, but all scanning methods could handle it. Maybe not the sharpest film/lens setup in the world, but hey, this is just round 1. More examples will follow
Here is a crop from the V500, scanned at 4800 dpi. No Epson Scan sharpening and minimal if any sharpening applied:
Here is a crop from the full negative, taken with the Panasonic G3, and OM 50mm 3.5 macro:
Finally, for fun, here is one taken with the Panasonic G3 and OM macro at 1:1 reproduction:
It's no surprise the 1:1 macro shot is doing the best, sharp and at such a magnification that there's not much digital noise interfering with the grain. Of course, when you do the math, at 1:1 the 4/3 Panasonic G3 sensor is magnifying at something around 6000+ dpi if my math is correct (someone will probably correct me on this) Downside is that one would need to take 4 good, level shots and stitch them together to cover a full 35 mm neg. Only worth it for the real keepers and requires a decent piece of software to do the mosaicking.
Next best seems to be the G3 shot taken of the entire negative. After cropping the top and bottom, it works out to 13mp. This is how I tend to use my camera for scanning negatives. While it's not lacking much detail from the 1:1 shot, I do see what looks like digital noise interfering with the grain, and it isn't very pretty. This leads me to conclude that even for grainy 400 speed film, 13 mp is inadequate for properly recording all the detail and grain of a 35 mm frame. But in terms of printing, it would have to be a large one to notice this.
Finally, in third, is the V500 scan, which I honestly thought did a surprisingly good job compared to my past experience with this scanner. Maybe it likes this film more than others, we'll see. Still present, however, is the dreaded V500 grain-obliterating "smear" that can only be corrected to a point by sharpening and/or image size reduction.
The Coolscan 5000 comparison is coming...
Larger images of the ones above can be seen on my Flickr set
Next up - color slide film
This test shot was taken with an OM1 and 35-70 3.5-4.5 zoom at close focus. Film is Arista Premium 400 exposed at 400 and developed in Caffenol CH. It was a thin negative, but all scanning methods could handle it. Maybe not the sharpest film/lens setup in the world, but hey, this is just round 1. More examples will follow
Here is a crop from the V500, scanned at 4800 dpi. No Epson Scan sharpening and minimal if any sharpening applied:
Here is a crop from the full negative, taken with the Panasonic G3, and OM 50mm 3.5 macro:
Finally, for fun, here is one taken with the Panasonic G3 and OM macro at 1:1 reproduction:
It's no surprise the 1:1 macro shot is doing the best, sharp and at such a magnification that there's not much digital noise interfering with the grain. Of course, when you do the math, at 1:1 the 4/3 Panasonic G3 sensor is magnifying at something around 6000+ dpi if my math is correct (someone will probably correct me on this
) Downside is that one would need to take 4 good, level shots and stitch them together to cover a full 35 mm neg. Only worth it for the real keepers and requires a decent piece of software to do the mosaicking.Next best seems to be the G3 shot taken of the entire negative. After cropping the top and bottom, it works out to 13mp. This is how I tend to use my camera for scanning negatives. While it's not lacking much detail from the 1:1 shot, I do see what looks like digital noise interfering with the grain, and it isn't very pretty. This leads me to conclude that even for grainy 400 speed film, 13 mp is inadequate for properly recording all the detail and grain of a 35 mm frame. But in terms of printing, it would have to be a large one to notice this.
Finally, in third, is the V500 scan, which I honestly thought did a surprisingly good job compared to my past experience with this scanner. Maybe it likes this film more than others, we'll see. Still present, however, is the dreaded V500 grain-obliterating "smear" that can only be corrected to a point by sharpening and/or image size reduction.
The Coolscan 5000 comparison is coming...
Larger images of the ones above can be seen on my Flickr set
Next up - color slide film
stompyq
Well-known
This is a great comparison. I'am surprised at the output from the G3+macro compared to the scanner. Can you elaborate on your setup? Waiting for the coolscan comparisons
Noll
Well-known
Here's a slide taken on EliteChrome 100, using a Yashica T4 Super.
First up is the V500, scanned at 4800 dpi, mild sharpening in Photoshop (smart sharpen 3 px @100%) :
Crop of same image made using the Coolscan 5000 at 4000 dpi:
Next up is crop of a "full-frame" scan using the Panasonic G3:
Last is a crop from a photo made with the G3 at 1:1 reproduction ratio. Again, 4 images would need to be combined to cover a 35 mm frame:
I'd say it's pretty close between the 1:1 crop and Coolscan 4000, with a little less grain accentuation happening with the G3. I'm happy with both. The full-frame G3 scan is even fine for most shots, but again I feel like there's some grain-noise interference degrading the image somewhat. APS-C or a full-frame digital would probably do better here.
I should add that the Panasonic G3 cost me $150 used, and the OM 50 mm 3.5 macro from KEH was $65. This is cheap compared to high end flatbeds, to say nothing of dedicated film scanners like the Coolscan. And it doubles as a camera, too.
Of course, it's not as convenient as plopping down a strip of negs and not having to worry about paralleling the light table. And color neg film is more challenging this way as well. Perhaps in a few days I'll post round 3.
First up is the V500, scanned at 4800 dpi, mild sharpening in Photoshop (smart sharpen 3 px @100%) :
Crop of same image made using the Coolscan 5000 at 4000 dpi:
Next up is crop of a "full-frame" scan using the Panasonic G3:
Last is a crop from a photo made with the G3 at 1:1 reproduction ratio. Again, 4 images would need to be combined to cover a 35 mm frame:
I'd say it's pretty close between the 1:1 crop and Coolscan 4000, with a little less grain accentuation happening with the G3. I'm happy with both. The full-frame G3 scan is even fine for most shots, but again I feel like there's some grain-noise interference degrading the image somewhat. APS-C or a full-frame digital would probably do better here.
I should add that the Panasonic G3 cost me $150 used, and the OM 50 mm 3.5 macro from KEH was $65. This is cheap compared to high end flatbeds, to say nothing of dedicated film scanners like the Coolscan. And it doubles as a camera, too.
Of course, it's not as convenient as plopping down a strip of negs and not having to worry about paralleling the light table. And color neg film is more challenging this way as well. Perhaps in a few days I'll post round 3.
Noll
Well-known
A picture of the setup I've been using - it ain't pretty. It could definitely use some improvement, but I don't have the time these days to figure out a better rig.
The 35mm negative holder was scavenged from an HP scanner, which I raise off the table slightly to minimize dusty surfaces that are in-focus.
The 35mm negative holder was scavenged from an HP scanner, which I raise off the table slightly to minimize dusty surfaces that are in-focus.
Athiril
Established
Macro copying always works as well as the operator, so of course it can work very well with some effort applied.
Here is one I did years ago from 6x7 FP4+ developed in Rodinal and a reverse mounted G.Zuiko 50mm on a Canon 30D
The top crop is a V500 3200 dpi crop, the bottom is the 30D crop.
35mm is very capable of amazing detail and can astonish you.
Here is one I just did recently, taken on a Canon AE-1 with FD 28mm f/2.8 @ f/5.6
FF
Flatbed scan (3200 dpi crop enlarged to 300%)
Imacon 949 scan (second one is scanned twice and has had SR applied with photoacute)
The chart height is about 1.5-1.6mm
You'd need the highest end of MF digital to beat this, though not really the point, I shoot Tri-X for it's nice grain, etc, but it puts in perspective what you can get out of simple inexpensive gear these days, there's something for everyone.
Pictorial image
Here is one I did years ago from 6x7 FP4+ developed in Rodinal and a reverse mounted G.Zuiko 50mm on a Canon 30D
The top crop is a V500 3200 dpi crop, the bottom is the 30D crop.
35mm is very capable of amazing detail and can astonish you.
Here is one I just did recently, taken on a Canon AE-1 with FD 28mm f/2.8 @ f/5.6
FF
Flatbed scan (3200 dpi crop enlarged to 300%)
Imacon 949 scan (second one is scanned twice and has had SR applied with photoacute)
The chart height is about 1.5-1.6mm
You'd need the highest end of MF digital to beat this, though not really the point, I shoot Tri-X for it's nice grain, etc, but it puts in perspective what you can get out of simple inexpensive gear these days, there's something for everyone.
Pictorial image
Snowbuzz
Well-known
gulp, wow. Very illuminating samples.
Noll
Well-known
Thanks for the contribution, Athiril. Impressive results from everything above (except the V500
). That Photoacute software appears to be the real deal. Not sure I could pony up for an Imacon, though.I'm finishing off another few rounds of comparisons - on deck are Arista Edu 100, more Elite Chrome (this time with Noritsu uber-scans), and Velvia 50.
Though, I'm starting to think that 1:1 reproduction might be pushing the limits of the OM 50mm f3.5 macro lens...
Godfrey
somewhat colored
The problem with optical capture using a macro lens is that it can be great, but takes rigor and diligent setup to get the best out of it. Whereas a good film scanner like the Coolscan V or 5000 basically nails the job each time, consistently.
G
G
Noll
Well-known
Yowza. Did I read you correctly that this is Tri-X ? Developed in..?
Goodness, I thought I was getting sharp results...
Noll
Well-known
Here is our lovely power meter scene again
(next round will be a landscape, I promise)
Arista EDU Ultra 100, shot with an EOS 50mm 1.8 lens, stopped down to 5.6-ish. Developed in Caffenol C-M 12 min
First up is the V500 crop from "6400 dpi" scan, mild smart sharpen applied:
Panasonic G3 - crop from a full-frame macro-photo:
Panasonic G3 - 1:1 macro photo:
Finally the Coolscan 5000
Here it looks like the Panasonic G3 full-frame macro-photo really can't compete with the Coolscan 5000. The meter face occupies a slightly smaller area on the negative and definitely looks "digitized" compared to the 1:1 photo and Coolscan. Somewhat interesting is that the macro-photo is not picking up on the scratches that the Coolscan 5000 is highlighting. I am not sure why this is.
(next round will be a landscape, I promise)
Arista EDU Ultra 100, shot with an EOS 50mm 1.8 lens, stopped down to 5.6-ish. Developed in Caffenol C-M 12 min
First up is the V500 crop from "6400 dpi" scan, mild smart sharpen applied:
Panasonic G3 - crop from a full-frame macro-photo:
Panasonic G3 - 1:1 macro photo:
Finally the Coolscan 5000
Here it looks like the Panasonic G3 full-frame macro-photo really can't compete with the Coolscan 5000. The meter face occupies a slightly smaller area on the negative and definitely looks "digitized" compared to the 1:1 photo and Coolscan. Somewhat interesting is that the macro-photo is not picking up on the scratches that the Coolscan 5000 is highlighting. I am not sure why this is.
Noll
Well-known
Yes, I agree with that. Though once set up and locked in place, this setup can easily scan a roll in 5 minutes. I don't think the same can be said for most low-end scanners.
Ultimately though, I sense that there is a small but growing niche for a tube that can be readily adapted to different macro lens filter rings, adjustable in length, with adapter on the back end to take either mounted slides or negative strips - and simply be pointed at a light source (or have one built in). This would solve the parallel issues associated with using a tripod and light table and make this method as reliable as a dedicated scanner. (I know there are slide-copying rigs out there but haven't seen anything made to be universally adaptable to a variety of lenses)
Maybe even packaged with a utility for automatically inverting images. I've done enough scripting that this should be pretty easy with Python and GIMP...
Oh if there were only more time in the day ....
mdarnton
Well-known
Alignment really isn't a problem. Put a mirror where the film is and adjust the camera until the reflection of the lens is centered in the viewfinder, and everything is parallel. I can be set up and "scanning" before your film scanner has spit out its first neg. I have the whole rig, minus the camera set up in my digital "darkroom": http://www.flickr.com/photos/mdarnton/7183241686/
I also discovered that I get slightly better results facing the emusion to the camera (possibly a combination of not going through the film base, plus the curvature of the film being slightly more favorable to the lens). I also let my Nikon D300's autofocus find focus for me (via the green dot while I focus manually, using an ancient Micro-Nikkor--I have three of different vintages, and they're all equal in this process).
The process is very quick, and better than any flatbed. This fall I'll have a D7100, and I expect the quality to jump accordingly.
To the OP: be aware that the Oly 50/3.5 macro has a floating element in it, and at this range should always be used fully-extended. I used to produce slide shows and filmstrips in the early 70s and used this lens for all the copying, in preference to the Componons on the fancy (was it a Bencher?) copy camera the facility had, and my results were much better. I managed the extension problem by having a full set of short tubes, so that the lense would always be all the way out between 1:2 and 1:1.
I also discovered that I get slightly better results facing the emusion to the camera (possibly a combination of not going through the film base, plus the curvature of the film being slightly more favorable to the lens). I also let my Nikon D300's autofocus find focus for me (via the green dot while I focus manually, using an ancient Micro-Nikkor--I have three of different vintages, and they're all equal in this process).
The process is very quick, and better than any flatbed. This fall I'll have a D7100, and I expect the quality to jump accordingly.
To the OP: be aware that the Oly 50/3.5 macro has a floating element in it, and at this range should always be used fully-extended. I used to produce slide shows and filmstrips in the early 70s and used this lens for all the copying, in preference to the Componons on the fancy (was it a Bencher?) copy camera the facility had, and my results were much better. I managed the extension problem by having a full set of short tubes, so that the lense would always be all the way out between 1:2 and 1:1.
Monochrom
Well-known
Hi, ver very interesting!!!
Thanks for posting!
I left film because i didn´t have the time for developing nor scanning...
Thanks for posting!
I left film because i didn´t have the time for developing nor scanning...
fdarnell
Well-known
Very enlightening series - thanks for all your efforts here. When you think about the Coolscan - think of it as being a condenser enlarger. You have an LED sourced projected through collimated optics onto the sensor strip as it is moved past the film.
In your digital camera scans, think of it like a diffusion enlarger or like the old "cold-light" heads - like printing black and white on a color enlarger. The white background diffusion screen produces and even scattered light for even illumination of the film which is then imaged on the sensor.
Things like dust and scratches were minimized by the diffusion enlargers and more noticed on condensers. Condensers tended to produce sharper grain images, which you could see with the grain magnifying focus aids.
Overall, it's a matter of personal preference. Both the Coolscan and the shots were you would stitch the image together reveal the actual grain structure of the film. I would think that's the goal.
Godfrey
somewhat colored
A device like this works nicely on my Micro-Nikkor 50mm f/3.5 for copying mounted 35mm slides onto 16x24mm, except that I have to space it out a ways for the 35mm to APS-C 1:2 reduction distance.
http://www.bhphotovideo.com/bnh/controller/home?O=&sku=37453&is=REG&A=details&Q=
I've seen other, similar devices with a built-in, dedicated macro lens and zoom capability, but usually they zoom from 1:1 to 2:1 so you can crop 35mm slides for 35mm format recording. What you need for APS-C recording is the ability to run from 1:3 to 1:1 format.
Of course, I'd like something more general as well for capturing 6x6 negatives to APS-C or FF digital. For when speed is the most important factor. However, they're not going to beat a dedicated film scanner for consistency, quality and ease of use IMO.
(I've done all of this stuff over the past 20 years... ! One of the few benefits of lots of years .... )
G
http://www.bhphotovideo.com/bnh/controller/home?O=&sku=37453&is=REG&A=details&Q=
I've seen other, similar devices with a built-in, dedicated macro lens and zoom capability, but usually they zoom from 1:1 to 2:1 so you can crop 35mm slides for 35mm format recording. What you need for APS-C recording is the ability to run from 1:3 to 1:1 format.
Of course, I'd like something more general as well for capturing 6x6 negatives to APS-C or FF digital. For when speed is the most important factor. However, they're not going to beat a dedicated film scanner for consistency, quality and ease of use IMO.
(I've done all of this stuff over the past 20 years... !
One of the few benefits of lots of years .... ) G
2WK
Rangefinder User
Clearly the winner here is the v500!
ahaha....I just bought a v750, so I have to say things like this as seeing a coolscan comparison always bums me out a little bit.
ahaha....I just bought a v750, so I have to say things like this as seeing a coolscan comparison always bums me out a little bit.
Faintandfuzzy
Well-known
Athiril
Established
Chart height is 1.5-1.6mm on the negative.
Sorry I'm an idiot, I forgot to mention that is Adox CMS 20 stand developed in Perfection XR-1 (way too thin, needs less dilution for better results, but still manages high detail, would probably require grade 5 printing in the dark room), not Tri-X, the point was while you need MF digital at it's best to beat this light weight cheap equipment combo (sans scanner), that that isn't everything and I still really enjoy Tri-X with it's grain and all.
While you can overexpose a lot, and develop in a speed-losing contrast-enhancing developer to force somewhat more detail than normal, you can't jump from normal Tri-X to those kind of results like that I would think, at least not that clinically clean, even T-Max 100 wouldn't be that clean.
I have managed that in modifying C-41, and getting it from 50 lp/mm recorded on Lucky 200 to 70 lp/mm recorded (which means the film res would be 80 lp/mm to record that much from the lens I used).
Once I have it fine tuned I'll try it on Ektar, to see if I can exceed slide results. Plus be able to expose Ektar at a much lower speed without having to suffer contrast loss.
Here is what my Tri-X looks like (and this is 6x6)
PhotoAcute works well on my film scans, just 2 passes. You have to lie to it and put in a digital camera + lens combo in the profile settings, despite having a GH2 profile, I have so far not managed to get increased detail out of it. Though it may be worth trying on digital camera film copying (2 separate shots of the same area required), you should be able to exceed both normal sensor and lens resolution due to the way multi-image superresolution works.
Sorry I'm an idiot, I forgot to mention that is Adox CMS 20 stand developed in Perfection XR-1 (way too thin, needs less dilution for better results, but still manages high detail, would probably require grade 5 printing in the dark room), not Tri-X, the point was while you need MF digital at it's best to beat this light weight cheap equipment combo (sans scanner), that that isn't everything and I still really enjoy Tri-X with it's grain and all.
While you can overexpose a lot, and develop in a speed-losing contrast-enhancing developer to force somewhat more detail than normal, you can't jump from normal Tri-X to those kind of results like that I would think, at least not that clinically clean, even T-Max 100 wouldn't be that clean.
I have managed that in modifying C-41, and getting it from 50 lp/mm recorded on Lucky 200 to 70 lp/mm recorded (which means the film res would be 80 lp/mm to record that much from the lens I used).
Once I have it fine tuned I'll try it on Ektar, to see if I can exceed slide results. Plus be able to expose Ektar at a much lower speed without having to suffer contrast loss.
Here is what my Tri-X looks like (and this is 6x6)
Thanks for the contribution, Athiril. Impressive results from everything above (except the V500
I'm finishing off another few rounds of comparisons - on deck are Arista Edu 100, more Elite Chrome (this time with Noritsu uber-scans), and Velvia 50.
Though, I'm starting to think that 1:1 reproduction might be pushing the limits of the OM 50mm f3.5 macro lens...
PhotoAcute works well on my film scans, just 2 passes. You have to lie to it and put in a digital camera + lens combo in the profile settings, despite having a GH2 profile, I have so far not managed to get increased detail out of it. Though it may be worth trying on digital camera film copying (2 separate shots of the same area required), you should be able to exceed both normal sensor and lens resolution due to the way multi-image superresolution works.
Noll
Well-known
...........
Last edited:
Noll
Well-known
With the cold and darkness I've been doing some more scanning tests - this time with color slide film!
The camera was a Yashica T4, shot on Elitechrome XC. I had the film processed and uber-scanned by NCPS on their Noritsu minilab - in addition to all the scans I've done afterward...
Here is our detail-rich scene as scanned by the Nikon Coolscan:
I chose to focus the comparison below on an area just below center due to the combination of high detail and dark shadow areas. Here they are from worst-ish to best-ish.
Epson V500 @ 4800 dpi:
Panasonic G3 + 50mm f/3.5 macro @f/8, "1-shot scan" (about 1:4 reproduction on the G3)
Nikon Coolscan 5000 @4000 dpi:
Panasonic G3 + 50mm f/3.5 macro @f/8, "2-shot scan" (about 1:2 reproduction on the G3, photomerged)
Noritsu minilab best quality:
Last but not least, Panasonic G3 + 50mm f/3.5 macro @f/8, "4-shot scan" (about 1:1 reproduction on the G3, photomerged)
My 2 cents... as expected the V500 is by far the worst in terms of sharpness. It has decent color rendition of the actual scene, as long as you don't care about ever having a sharp-looking image beyond a 4x6 print.
Using the G3 + macro lens for a 1-shot scan is definitely more viable than flatbed scanning. Still, the images ended up having a slightly mushy texture at 100, and is definitely leaving at least half the detail on the slide.
The Nikon Coolscan does a respectable enough job. Detail is close to the Noritsu, but I dislike the color output from this scanner. It tends to look a little too pastel for my taste, and I often see color speckling at 100%
The Noritsu does a great job. There is none of the color speckling seen in the Coolscan, and the detail has some "bite" to it. Super sharp, and the best shadow detail though the color balance came back way on the green side and oversaturated compared to the slide. Considering that NCPS only provides jpeg, this can be a challenge to correct.
Finally there are the Panasonic G3 2-shot and 4 shot panorama-scans. Practically speaking, I'd say these are identical in terms of detail. As I've probably mentioned elsewhere, I seem to be hitting the limits of this lens. Shadows are not as good as on the noritsu, though colors are a bit more accurate to the slide - if a bit underwhelming. Overall, though I think the Noritsu still does a slightly better job, color balance aside.
Well there you have it! If anyone wants to see more you can view the entire scene for each crop above at my Flickr: http://www.flickr.com/photos/68243197@N04/sets/72157633582603921/
The camera was a Yashica T4, shot on Elitechrome XC. I had the film processed and uber-scanned by NCPS on their Noritsu minilab - in addition to all the scans I've done afterward...
Here is our detail-rich scene as scanned by the Nikon Coolscan:
I chose to focus the comparison below on an area just below center due to the combination of high detail and dark shadow areas. Here they are from worst-ish to best-ish.
Epson V500 @ 4800 dpi:
Panasonic G3 + 50mm f/3.5 macro @f/8, "1-shot scan" (about 1:4 reproduction on the G3)
Nikon Coolscan 5000 @4000 dpi:
Panasonic G3 + 50mm f/3.5 macro @f/8, "2-shot scan" (about 1:2 reproduction on the G3, photomerged)
Noritsu minilab best quality:
Last but not least, Panasonic G3 + 50mm f/3.5 macro @f/8, "4-shot scan" (about 1:1 reproduction on the G3, photomerged)
My 2 cents... as expected the V500 is by far the worst in terms of sharpness. It has decent color rendition of the actual scene, as long as you don't care about ever having a sharp-looking image beyond a 4x6 print.
Using the G3 + macro lens for a 1-shot scan is definitely more viable than flatbed scanning. Still, the images ended up having a slightly mushy texture at 100, and is definitely leaving at least half the detail on the slide.
The Nikon Coolscan does a respectable enough job. Detail is close to the Noritsu, but I dislike the color output from this scanner. It tends to look a little too pastel for my taste, and I often see color speckling at 100%
The Noritsu does a great job. There is none of the color speckling seen in the Coolscan, and the detail has some "bite" to it. Super sharp, and the best shadow detail though the color balance came back way on the green side and oversaturated compared to the slide. Considering that NCPS only provides jpeg, this can be a challenge to correct.
Finally there are the Panasonic G3 2-shot and 4 shot panorama-scans. Practically speaking, I'd say these are identical in terms of detail. As I've probably mentioned elsewhere, I seem to be hitting the limits of this lens. Shadows are not as good as on the noritsu, though colors are a bit more accurate to the slide - if a bit underwhelming. Overall, though I think the Noritsu still does a slightly better job, color balance aside.
Well there you have it! If anyone wants to see more you can view the entire scene for each crop above at my Flickr: http://www.flickr.com/photos/68243197@N04/sets/72157633582603921/