DSLR imaging problem

I thought I'd start a new thread on this one, rather than continue an existing thread.

I have two options for scanning-- An Epson v800, and an EOS 90D APS-C DSLR. I don't want to hear from the Sony fanboys that Canon sensors suck. As far as I know, based on some basic math, diffraction shouldn't be an issue until f/11, and only then at one end of the red spectrum. Also, there's a fair amount of anecdotal evidence that DPP4 (Canon's software) does a very good job of compensating for possible diffraction limits.

After some experimenting, I've come up with a couple processes that produce good color-- so that's not an issue. What I'm having problem with, oddly, is a lack of detail when using the DSLR. Or perhaps, it's a lack of contrast (the inverted images do initially seem to be lacking in contrast).

Here are some samples:

First, a 100% crop from the Epson V800, scanned with Silverfast SE+ at 3200 PPI (~ 14 megapixels). Negafix was the only active module. Sharpening applied in post. My personal opinion, there is more apparent detail in the eye on this photo, than anything I've been able to manage on the DSLR.



Next is a processed image (28 MP) from the 90D + 100mm f/2.8 macro @ f/5.6 (Shot RAW, processed with Canon's DPP 4, inverted via Darktable+negadoctor, dust removal and sharpening applied via Affinity Photo):



And then just to be paranoid, here's a 90D RAW f/8 shot, inverted only, no color correction:



I considered a possible sensor or lens issue-- I'm using an inverted tripod with a horizontal beam, leveled with a 2 way spirit level in the hot shoe, and a very nice, heavy, level, workbench. I'm focusing using 10x zoom ('micro focus') and I'm focusing on the grain itself. Still, I did drop my macro lens a couple months ago, so I thought I should check it:



Just a scan of my proof sheet, but the detail is pretty good-- you can even see the yellow anti-counterfeit dots if you look closely. So I don't think it's my lens.

Any ideas? I don't have the light issue totally surrounded (so to speak), so as I said, there's what I consider a mild lack of contrast, but I wouldn't think that would rob detail as much as I'm seeing.

I don't use Affinity Photo, but without knowing more about it, I would suspect the dust removal might be part of the problem with the second one?

How are you releasing the shutter?

runswithsizzers said:

I don't use Affinity Photo, but without knowing more about it, I would suspect the dust removal might be part of the problem with the second one?

How are you releasing the shutter?

Click to expand...

Dust removal is entirely manual-- Zoom to 50%, scroll, find a spot, whack it with healing brush, keep scrolling, repeat.

Releasing shutter with two second delay and wireless remote from about 2 feet away.

grat said:

I'm using an inverted tripod with a horizontal beam, leveled with a 2 way spirit level in the hot shoe, and a very nice, heavy, level, workbench.

Click to expand...

Unless you're using flash to light the neg, the tripod/ horizontal beam is likely causing tiny buzzing vibrations, robbing sharpness. A fairly serious copy stand or similar is probably going to make a significant improvement - under any circumstances the results should be better than the V800, so I'd investigate all possible mechanical vibration sources (even the feet of a tripod on a table is a possible source of wobble).

"Watch Thread"

Thanks for sharing your process.

Lachlan Young said:

Unless you're using flash to light the neg, the tripod/ horizontal beam is likely causing tiny buzzing vibrations, robbing sharpness. A fairly serious copy stand or similar is probably going to make a significant improvement - under any circumstances the results should be better than the V800, so I'd investigate all possible mechanical vibration sources (even the feet of a tripod on a table is a possible source of wobble).

Click to expand...

Dunno-- It's on a pretty heavy-duty workbench on rubber feet, so most of the vibrations should be damped out. It's probably also worth mentioning I'm using "live view", which on the DSLR, means I'm also using mirror lock-up. And if that's the case, why isn't it showing up on the paper test?

I did make one change which improved the detail somewhat-- I flipped the negative over so the emulsion was up.

Is there any chance that the lens focuses correctly on the piece of paper but is slightly missing focus on the negative for some reason? The DSLR images to me look more like missed focus rather than vibration blur.

warden said:

Is there any chance that the lens focuses correctly on the piece of paper but is slightly missing focus on the negative for some reason? The DSLR images to me look more like missed focus rather than vibration blur.

Click to expand...

grat said:

I did make one change which improved the detail somewhat-- I flipped the negative over so the emulsion was up.

Click to expand...

Looking at them again on something other than my phone, I think that missed focus is a highly likely cause. Is the camera using phase or contrast detection, or a bit of each in live view mode?

Lachlan Young said:

Looking at them again on something other than my phone, I think that missed focus is a highly likely cause. Is the camera using phase or contrast detection, or a bit of each in live view mode?

Click to expand...

I don't disagree-- but my method has been to use live view, zoom in to 10x focus, and set focus purely manually based on the film grain. I had read elsewhere that this prevents the potentially out-of-focus areas of the negative from confusing the sensor(s). The film has a bit of curl lengthwise, but the gates on the pixl-latr should be holding that relatively flat. Side to side curl is negligible.

I don't do color but I scan 2 1/4 and 4x5 B&W with an Epson 850 and Silverfast studio usually @ 2400 and get very good results. I don't use Negafix. Have you tried stopping the lens down more and perhaps using a negative carrier from an enlarger to hold the film as flat as possible. Having dropped the lens can also be an issue. Sone time ago a friend asked me to photograph him and his wife with his camera. Despite a tripod and careful manual focusing all the negatives were off a little. After the fact he told me he had dropped the lens but there was no obvious damage. He sent it in for repair and lens elements were off and needed adjustment.

http://www.jeffreyglasser.com/

htp://www.sculptureandphotography.com/

Yeah, I need to do some more testing-- Including dragging out some other lenses.

So far, I've seen a mild improvement by shooting the emulsion side (whoops!), and by auto-focusing on in-focus areas of the film. This confuses me, as I was under the impression that focusing on the grain is superior to letting the camera focus.

I'm shooting f/8 already (although not seeing a difference between f/5.6 and f/8).

What tripod are you using?
How are you lighting the negatives? (did I miss it?)
Sandwich the negative between two pieces of AN glass, AN sides to the film.
Switch to a Nikon system.

shutterfinger said:

What tripod are you using?
How are you lighting the negatives? (did I miss it?)

Click to expand...

Tripod is an inexpensive, but decently built Neewer/K&F unit with the ball in the middle of the support column. It's not great, but I've got it at the lowest height, with the center column inverted, and vibration is pretty low. The shot of the paper was taken with the same setup, except the LED panel was off.

As for lighting, it's an inexpensive, but again, pretty decent high CRI LED panel (Raleno PLV-S192) behind a pixl-latr. One of my tests will be to rebuild the rig with a flash unit to see if it makes any noticeable difference. I'm also using some sheets of plastic to mask off the rest of the panel and the other parts of the film strip.

Sandwich the negative between two pieces of AN glass, AN sides to the film.

Click to expand...

I'm not getting Newton rings, and adding glass isn't going to improve sharpness.

Switch to a Nikon system.

Click to expand...

Don't you think I have enough problems already?

As for using a tripod as a camera support when copying film - they are fussy to set up, but it is possible to get sharp results. Before I switched to a copy stand (for convenience), I used a tripod with inverted column, and I tested it for sharpness. At first, my results were very soft because I was pressing the shutter on the camera, but after I switched to using a 2 second time delay for the shutter release, sharpness was good.

My tripod is an entry level Bogen/Manfrotto 3001 which is not particularly robust. And the light source that I was using at the time was not particularly bright, so my shutter speed was often in the 1/8th second range. The 100mm enlarger lens on an APS-C camera resulted in a sensor-to-film range of about 42-43cm to get the magnification needed to copy a 35mm negative. Below is a photo of my first copy rig and a test shot at about the required magnification. (I just noticed the photo shows 120 film being copied, but what I said was about copying 135 film.)

Since then, I have switched to a copy stand, a better lens, and a brighter light source that results in shutter speeds in the 1/60th second range, and my results improved only marginally.

The purpose for the glass is film flatness. Plain glass will likely add newton rings.
Your tripod is a weak link. A wood tripod or copy stand will be a worthwhile investment if you plan to do DSLR coping of your negatives regularly. ,
Download the test chart(s) http://www.normankoren.com/Tutorials/MTF5.html , print copies at your printers highest resolution on good paper (photo paper is not necessary) , then photograph the charts with your copy setup having a copy of the chart going each direction across the frame so that the same section of the chart is on both sides of the frame.
http://www.takinami.com/yoshihiko/photo/lens_test/USAF.pdf the USAF1951 target.

To avoid deriving false conclusions from your samples, I would try to equalize the processing a bit more. The three scan samples you have shown appear all quite different in colour and contrast - both of which can affect the perception of sharpness. Obviously, the amount and nature of sharpening has a huge effect too. Having said that, based on what I can see, the f8 shot does look a little more detailed than the f5.6 shot. If these were shot without changing focus, then this immediately suggests that the shot is not focused optimally (greater DOF brings it into focus). If it was, and assuming no vibration effects, then with most good macro lenses I would expect a small but noticeable drop in sharpness at f8 compared to f5.6.

There are many possible reasons why the focus may be sub-optimal. At the end of the day, getting focus absolutely spot-on every time for 33mp is next to impossible. The only way you can be sure you are getting best focus is to shoot a focus bracket. Your camera does actually have a focus bracketing/stacking feature, but whether your lens is compatible with this I'm not sure, and I'm also not sure whether it will allow for steps which are fine enough. It may be worth a try. The other way is to do it manually. Normally I would do this by raising/lowering the camera in small steps, but given your setup on a tripod, this probably won't be possible. I believe the EOS utility software allows manual stepping of the AF - this is another option. Again, I haven't tried this, so I'm not sure whether the steps are small enough. The last option is to use the lens' focus ring. How easy this is will depend on the focus throw of the manual focus. In all cases, you need to start with the image definitely slightly out of focus and step through the DOF, so it comes in to focus and starts to go out of focus again. As long as you use small enough steps, one of the images will be at optimal focus. If it still isn't as sharp as you should expect, then you need to look elsewhere for the issue.

One possible reason for getting sub-optimal focus which hasn't been mentioned yet is focus shift. This is where the point of best focus at full aperture differs from that when stopped down. This can definitely be strong enough to have a real effect. The way around this is to focus with the lens stopped down a bit. On a 2.8 lens, I would suggest focusing at f4, at which point most of the aberrations which cause focus shift have gone. The EOS utility allows live view focusing with stopped-down aperture. I wouldn't suggest trying to focus at f5.6 or f8, because the increased DOF reduces accuracy, and lack of light often reduces the fidelity of live view.

Just to check on the vibration issue - I know you are using live view, but is the Silent LV shoot option enabled? If not, it will still close and open the first curtain, which could cause vibration.

I have an Epson V700, which produces the same resolution as your V800, and I have compared V700 scans (absolute best I could achieve) with DSLR scans of the same frames shot with a 600D and Sigma 105mm EX Macro (non OS). Where the film was in focus, the cameras scans captured noticeably more detail than the Epson. This was at f8. The 600D is 18mp, so your 90D and Canon macro should definitely be capable of producing noticeably better detail than your V800. Of course whether this detail is actual image detail, or just sharper grain, will depend on your film shots.

jgboothe said:

To avoid deriving false conclusions from your samples, I would try to equalize the processing a bit more. The three scan samples you have shown appear all quite different in colour and contrast - both of which can affect the perception of sharpness. Obviously, the amount and nature of sharpening has a huge effect too. Having said that, based on what I can see, the f8 shot does look a little more detailed than the f5.6 shot. If these were shot without changing focus, then this immediately suggests that the shot is not focused optimally (greater DOF brings it into focus). If it was, and assuming no vibration effects, then with most good macro lenses I would expect a small but noticeable drop in sharpness at f8 compared to f5.6.

Click to expand...

Yeah, The last shot, I didn't perform any color correction (other than inversion), because I was concerned that my color correction was causing detail loss. f/5.6 vs. f/8 doesn't show any differences in sharpness according to the image tests over at the-digital-picture.com. I think it is a focus issue, as the DoF of f/8 appears to have helped.

There are many possible reasons why the focus may be sub-optimal. At the end of the day, getting focus absolutely spot-on every time for 33mp is next to impossible. The only way you can be sure you are getting best focus is to shoot a focus bracket. Your camera does actually have a focus bracketing/stacking feature, but whether your lens is compatible with this I'm not sure, and I'm also not sure whether it will allow for steps which are fine enough. It may be worth a try.

Click to expand...

It's configurable. I've done some very close-up macro work with this camera (and tripod) using the same lens (and extension tubes). I can rule out vibration, tripod instability, and poor technique.

One possible reason for getting sub-optimal focus which hasn't been mentioned yet is focus shift. This is where the point of best focus at full aperture differs from that when stopped down. This can definitely be strong enough to have a real effect. The way around this is to focus with the lens stopped down a bit. On a 2.8 lens, I would suggest focusing at f4, at which point most of the aberrations which cause focus shift have gone. The EOS utility allows live view focusing with stopped-down aperture. I wouldn't suggest trying to focus at f5.6 or f8, because the increased DOF reduces accuracy, and lack of light often reduces the fidelity of live view.

Click to expand...

Haven't encountered that problem. I should also mention I've tried this process with MF and LF film, without issue, so I suspect the ultimate problem is film curl.

I have an Epson V700, which produces the same resolution as your V800, and I have compared V700 scans (absolute best I could achieve) with DSLR scans of the same frames shot with a 600D and Sigma 105mm EX Macro (non OS). Where the film was in focus, the cameras scans captured noticeably more detail than the Epson. This was at f8. The 600D is 18mp, so your 90D and Canon macro should definitely be capable of producing noticeably better detail than your V800. Of course whether this detail is actual image detail, or just sharper grain, will depend on your film shots.

Click to expand...

Yeah-- The Epson gets a bit of a bad rap, and I'm not sure it's deserved. I tried a scan at 6400 PPI (insanity!), and actually, it came out OK, after a bit of post-sharpening (I don't let Silverfast do the sharpening, I do it myself). I think I'll try using the Epson holder with my DSLR (the Epson holder does a better job of holding the negative flat). At the moment, though, I think I'm getting the most result for least effort from the Epson, rather than the DSLR.

I still need to finish my DSLR "scanning" rig for MF and LF, though.

grat said:

I thought I'd start a new thread on this one, rather than continue an existing thread.

I have two options for scanning-- An Epson v800, and an EOS 90D APS-C DSLR. I don't want to hear from the Sony fanboys that Canon sensors suck. As far as I know, based on some basic math, diffraction shouldn't be an issue until f/11, and only then at one end of the red spectrum. Also, there's a fair amount of anecdotal evidence that DPP4 (Canon's software) does a very good job of compensating for possible diffraction limits.

After some experimenting, I've come up with a couple processes that produce good color-- so that's not an issue. What I'm having problem with, oddly, is a lack of detail when using the DSLR. Or perhaps, it's a lack of contrast (the inverted images do initially seem to be lacking in contrast).

Here are some samples:

First, a 100% crop from the Epson V800, scanned with Silverfast SE+ at 3200 PPI (~ 14 megapixels). Negafix was the only active module. Sharpening applied in post. My personal opinion, there is more apparent detail in the eye on this photo, than anything I've been able to manage on the DSLR.

View attachment 273335

Next is a processed image (28 MP) from the 90D + 100mm f/2.8 macro @ f/5.6 (Shot RAW, processed with Canon's DPP 4, inverted via Darktable+negadoctor, dust removal and sharpening applied via Affinity Photo):

View attachment 273336

And then just to be paranoid, here's a 90D RAW f/8 shot, inverted only, no color correction:

View attachment 273337

I considered a possible sensor or lens issue-- I'm using an inverted tripod with a horizontal beam, leveled with a 2 way spirit level in the hot shoe, and a very nice, heavy, level, workbench. I'm focusing using 10x zoom ('micro focus') and I'm focusing on the grain itself. Still, I did drop my macro lens a couple months ago, so I thought I should check it:

View attachment 273338

Just a scan of my proof sheet, but the detail is pretty good-- you can even see the yellow anti-counterfeit dots if you look closely. So I don't think it's my lens.

Any ideas? I don't have the light issue totally surrounded (so to speak), so as I said, there's what I consider a mild lack of contrast, but I wouldn't think that would rob detail as much as I'm seeing.

Click to expand...

that is missed focus. I have a 90D that I use for scanning. Your 90D has dual pixel AF that is extremely good when using live view. At the start of the roll, put the edge of the first frame inside the focus box and push the AF on button until you get focus confirm, then simply scan the roll. Depending on how you’re triggering, you may need to go into the menus and change the settings so it doesn’t try to focus when you take a photo. While you’re in there, change to first curtain electronic shutter if it isn’t already on.

secondly is your aperture, 5.6 is pretty much the minimum I’d consider as you need enough to DOF across the whole frame. If you can, shoot at f/8 or smaller. A little diffraction is more desirable than not in focus because the DOF is too small. It’s all a balancing act.

third, your lens of choice isn’t ideal for APS-C. The rendered resolution in the center of the frame is less than 20MP. What makes your lens so great is it maintains its resolution performance all the way out to the edges on a full frame camera, so for full frame, it’s quite high performance, but for APS-C, there are better choices.

I've had much better results using a copy stand ($150 on ebay) than a tripod. Also I use AF in LiveView. It focuses on the grain in an instant. I get much better results than manual focus.
This looks out of focus.

Also, I use F11. 5.6 is too far open, remember this is a macro shot.

Great info from jgbooth BTW.^^^

Your Canon scans aren't bad, but part of you're seeing is that color negative scans don't come close to maxing the dynamic range of the camera's sensor (quite the opposite in fact) so all your tonal values are kind of squashed. Look at your histograms and you'll see what I mean. In post-processing, try applying an s-curve to your image and see if that helps.

If you want to see for yourself where your lens's optimum aperture is, try a series of camera scan images at different apertures, and I think you'll pretty clear see one or two settings which give crispest images. And in the case of my 50/2.8 FD macro, it's definitely f/5.6 or f/6.3. By f/8, diffraction softens the image.

4season said:

If you want to see for yourself where your lens's optimum aperture is, try a series of camera scan images at different apertures, and I think you'll pretty clear see one or two settings which give crispest images. And in the case of my 50/2.8 FD macro, it's definitely f/5.6 or f/6.3. By f/8, diffraction softens the image.

Click to expand...

When you are shooting a macro image, like when scanning film, you need to stop down much further than that. While the lens performs ideally at f5.6 (this is for nearly all lenses), that is at regular shooting distances, not macro photography. At macro level DOF is so shallow that shooting at what seems to be an optimal aperture for lens performance is not useful if the DOF cannot compensate for any lack of film flatness.

Nikon's D850 has a special film converting mode. When you use that, it automatically sets the lens at f8. There is a reason for that.

Adrian Bacon said:

that is missed focus.

Click to expand...

After some testing, I can confidently say that film curl sucks

Basically, the pixl-latr, which does fine on medium format and 4x5, doesn't constrain 35mm as much as I would like. As soon as I put the film into the Epson holder, and used that for imaging, the focus came right back in.

third, your lens of choice isn’t ideal for APS-C. The rendered resolution in the center of the frame is less than 20MP. What makes your lens so great is it maintains its resolution performance all the way out to the edges on a full frame camera, so for full frame, it’s quite high performance, but for APS-C, there are better choices.

Click to expand...

I've had this lens for a very long time-- it was the first "quality" lens I bought for myself (ie, one where the review didn't say "It's a good lens for the price..."). Didn't realize the resolution was quite that weak on APS-C, and of course, the 90D is going to push that even harder.

grat said:

After some testing, I can confidently say that film curl sucks

Basically, the pixl-latr, which does fine on medium format and 4x5, doesn't constrain 35mm as much as I would like. As soon as I put the film into the Epson holder, and used that for imaging, the focus came right back in.

I've had this lens for a very long time-- it was the first "quality" lens I bought for myself (ie, one where the review didn't say "It's a good lens for the price..."). Didn't realize the resolution was quite that weak on APS-C, and of course, the 90D is going to push that even harder.

Click to expand...

Yes your DOF is way smaller when doing 35mm, due to the higher magnification, so keeping it flat is critical if you want good performance across the frame. It's not easy.

Not sure exactly which lens you have (version of Canon 100mm 2.8 macro?), but I wouldn't assume it's going to be in any way bad or not 'good enough' on APS-C just because it's a full frame lens. It's probably better than the Sigma 105mm macro I used on APS-C, and that exceeded the V700 output quite easily. I have also used the later Sigma OS version (a much better lens) - also a full frame lens, which performed very well on a 50MP 5DS r. The pixel pitch on your camera is not that much higher than the 5DS r. I would guess that keeping most of the frame within the plane of focus will be your biggest limiting factor.

grat said:

Yeah, I need to do some more testing-- Including dragging out some other lenses.

So far, I've seen a mild improvement by shooting the emulsion side (whoops!), and by auto-focusing on in-focus areas of the film. This confuses me, as I was under the impression that focusing on the grain is superior to letting the camera focus.

I'm shooting f/8 already (although not seeing a difference between f/5.6 and f/8).

Click to expand...

It just occurred to me that the problem could be created simply by photographing the WRONG SIDE of the film.
A principle long held in photography is that ALWAYS the emulsion should face the emulsion.

• In enlarging, the emulsion side of the film faces the emulsion side of the paper, as one example.
• In contact printing, the the emulsion side of the film faces the emulsion side of the paper.
• In projecting slides, the emulsion side goes toward the lens (and the screen)

If you photograph the emulsion side, you are focusing on the image forming surface, not on the base material-emulsion interface. In photographing a slide, putting emulsion toward sensor results in a horizontally flipped image, but that is easilty corrected in postprocessing the dSLR image...remember the image on film IS already FLIPPED from reality, when viewed from the emulsion side.

The slide duplication methodology of decades past, with special adapters that held the slide afixed to the camera in front of the optics precisely located (no motion/vibration consideration, whereas motion/vibration caused by camera mounted on tripod photographing a slide which is not mounted as a unit with the camera. Additionally there was NO contrast reducing light leakage possible between the slide and the optics!

jgboothe said:

Yes your DOF is way smaller when doing 35mm, due to the higher magnification, so keeping it flat is critical if you want good performance across the frame. It's not easy.

Click to expand...

I suspected it was the problem early on, but wanted other opinions to see if I'd overlooked anything.

Not sure exactly which lens you have (version of Canon 100mm 2.8 macro?), but I wouldn't assume it's going to be in any way bad or not 'good enough' on APS-C just because it's a full frame lens.

Click to expand...

It's the Canon EF 100mm f/2.8 Macro USM, and there's a pretty comprehensive review including MTF charts at the Digital Picture website:

https://www.the-digital-picture.com/Reviews/MTF.aspx?Lens=107

The MTF charts in particular, as Adrian points out, show that the lens is somewhat challenged in the 50 lp/mm region. Of course, I've seen MTF charts that "prove" the Epson v700/800 series isn't capable of anything more than 700 PPI, so...