How hard is scanning anyway?

[ColSebastianMoran]

http://mfphotography.ca/post/137752399525/my-dslr-scanning-post-processing-workflow-video

All of the images in these two albums were 'scanned' in this way:

https://www.flickr.com/photos/137664763@N07/albums/72157661424754404

https://www.flickr.com/photos/137664763@N07/albums/72157661196247504/with/24351367016/

Michael, thanks for the excellent video.

Question: As the video starts, you have the shot with a bit of blank film and the rest of the color negative. Your blank film is nearly white. The film base in all the C41 films I use is very orange. How did you shoot this so that you get white in the negative film base? Is it just white balance in the camera?

Might be in your other tutorials, but I couldn't find #1 or #2.

I've experimented with this myself. I got my best results by using a color head to give a cold blue light, then shoot with the DSLR.

 

[edge100]

Michael, thanks for the excellent video.

Question: As the video starts, you have the shot with a bit of blank film and the rest of the color negative. Your blank film is nearly white. The film base in all the C41 films I use is very orange. How did you shoot this so that you get white in the negative film base? Is it just white balance in the camera?

Might be in your other tutorials, but I couldn't find #1 or #2.

I've experimented with this myself. I got my best results by using a color head to give a cold blue light, then shoot with the DSLR.

Its a combination of moving the exposure until *just* before clipping, and then backing it off a bit (to avoid clipping the reds) and also WB.

 

[ColSebastianMoran]

Thanks. That was fast.

Can you add a link to your tutorials #1 and #2?

 

[edge100]

Well, it would be very difficult to pin that variable to price - but, from my work in the video industry, I'm convinced that the CRI of some LEDs are better than others. The ones with a good CRI are definitely more expensive, but the more expensive ones do not necessarily have a good CRI.

Unfortunately, I have no idea how to test this on a light pad. It might be best to just build a light pad out of known good LEDs.

With all due respect, who cares? I'm telling you the Autograph LightPad works. In fact, I've linked to two albums full of images showing you this, and I've got literally thousands of other frames confirming this.

Whether or not an LED is accurate to some particular specification is of no relevance here; the whole negative is illuminated by the same light source, and the colour is corrected to the blank film itself. What matters - I would think - is whether this method can produce good scans. And I believe I've convincingly demonstrated that it can and does.

 

[edge100]

Thanks. That was fast.

Can you add a link to your tutorials #1 and #2?

Unfortunately, they were posts on my old blog, now lost to history.

For an updated take, I suggest this:

http://lamfoto.net/2016/01/16/digitizing-negatives-mirrorless-camera/

(Dave and I have worked iteratively on this process, and his method is more or less the same as mine).

 

[edge100]

So my scans - made with an LED source - are good in practice but potentially not good in theory???

It matters because the CRI is a measurement of what the light is, or is not, representing correctly. Here's an image to illustrate what I mean. Theoretically, you could make a lightbox with candles underneath it and set your white balance to correctly represent that Kelvin range. However, all of the wavelength that light source could not represent would also be absent in your scans (all of your blues/greens would be muted, muddy, and indistinct). In the case of LEDs we're talking about oranges, reds, blues, and greens being affected in the skin tone and sky areas (also, orange mask in this case).

The question is not whether or not it works - it's whether or not it works correctly.

Please know, I'm not doubting the DSLR method* in general - so don't think I'm trying to say that. I am simply trying to figure out how to refine the process to working best.

(*I think using conventional scanner is Ludditic)

 

[calebarchie]

So my scans - made with an LED source - are good in practice but potentially not good in theory???

Everybody has their own standards. I personally have seen the effects of poor CRI with my plusteks faulty light source. It is also easily observable with a CD.

It's not a matter whether it is correct or not but about achieving the highest standard.

 

[edge100]

Everybody has their own standards. I personally have seen the effects of poor CRI with my plusteks faulty light source. It is also easily observable with a CD.

It's not a matter whether it is correct or not but about achieving the highest standard.

I'm asking you for your opinion of my scans. I don't doubt one iota of the theory you've described

But my point is: does it matter in practice? I contend, based on thousands of scanned frames, is that the Lightpad is more than sufficient to make outstanding scans of all subjects.

 

[calebarchie]

I'm asking you for your opinion of my scans. I don't doubt one iota of the theory you've described

But my point is: does it matter in practice? I contend, based on thousands of scanned frames, is that the Lightpad is more than sufficient to make outstanding scans of all subjects.

I said everyone has their own standards. That depends on what "Practice" is, for everyday folk, most probably not at all.

 

[edge100]

I said everyone has their own standards. That depends on what "Practice" is, for everyday folk, most probably not at all.

So I'm left to conclude that you think my scans are, in fact, excellent and that your discussion above is largely academic? Does that just about capture it?

I'm always amazed at the ability of Internet forumites to extol the theoretical impossibility of that which has been clearly demonstrated to them.

 

[calebarchie]

So I'm left to conclude that you think my scans are, in fact, excellent and that your discussion above is largely academic? Does that just about capture it?

I'm always amazed at the ability of Internet forumites to extol the theoretical impossibility of that which has been clearly demonstrated to them.

Sorry? I think you are taking this too personally, I have not even looked at your scans.

I have however looked into DSLR scanning options, while they can produce excellent results it's not the thing for me.
EDIT: I am not "DrMcCoy" if you are getting confused here...

 

[mfogiel]

Originally posted by edge100

"I don't follow.

I've tested my D800 'scans' against a 9000ED, an X1, and a Heidelberg drum scan (can't recall the model).

The DR and resolving power of the D800 exceed that of the 9000ED; that is, I can perceive more detail on the D800 shot. It's not massive, and if you have a 9000ED, good on you. Then again, a D800 has a warranty and detaches from the copy stand for use as a camera, too. Can't do that with your 9000ED, can you?/QUOTE]

I am not trying to contradict you - I am more than willing to change my opinions when faced with hard facts, the point is, that IN THEORY this does non make any sense, hence my interest. I am glad to have CS9000, particularly for MF, but a day will come when it will die and it won't be possible to fix it, so I am very much on a look out for alternatives. To get back to my original question: my B&W 35mm scans come out at around 37 MB, say 35 MB of cleanly resolved information, so no wonder a D800 would trump it, but aren't you confused with the resolution comparison of bigger formats? Or perhaps you are stitching the MF neg scans? Have you compared crops from an image size equivalent to a 1mx1m print from a 6x6 negative ( about 20x enlargement) ?
Another question: would it make more sense, in your opinion, to use a camera with a sensor without the AA filter? And more: would a repro lens ( enlarging lens or copy lens for close up) enhance further the result?
Thanks for your patience.

 

[edge100]

"I don't follow.

I've tested my D800 'scans' against a 9000ED, an X1, and a Heidelberg drum scan (can't recall the model).

The DR and resolving power of the D800 exceed that of the 9000ED; that is, I can perceive more detail on the D800 shot. It's not massive, and if you have a 9000ED, good on you. Then again, a D800 has a warranty and detaches from the copy stand for use as a camera, too. Can't do that with your 9000ED, can you?/QUOTE]

I am not trying to contradict you - I am more than willing to change my opinions when faced with hard facts, the point is, that IN THEORY this does non make any sense, hence my interest. I am glad to have CS9000, particularly for MF, but a day will come when it will die and it won't be possible to fix it, so I am very much on a look out for alternatives. To get back to my original question: my B&W 35mm scans come out at around 37 MB, say 35 MB of cleanly resolved information, so no wonder a D800 would trump it, but aren't you confused with the resolution comparison of bigger formats? Or perhaps you are stitching the MF neg scans? Have you compared crops from an image size equivalent to a 1mx1m print from a 6x6 negative ( about 20x enlargement) ?
Another question: would it make more sense, in your opinion, to use a camera with a sensor without the AA filter? And more: would a repro lens ( enlarging lens or copy lens for close up) enhance further the result?
Thanks for your patience.

I explained why.

Yes, for MF and LF, I stitch (which I clearly stated above).

For 35mm, the D800 pulls at least as much info as the 9000ED, and has better DR.

And re: the AA filter, you're right.

 

[kuzano]

Keep low expectations...

Keep low expectations...

Long time reader, first time poster here. I've been really enjoying the discussions as well as the wealth of information on this forum.

As for the topic at hand, I've read countless threads from the archives by now, but a few questions remain. I hope no one will mind if I collect them here in a new thread. I'll try and keep everything short.

Background - I've been shooting more film lately and really enjoying it. I plan to do even more in 2016. Up until now I've had my local lab scan my negs. I've been reasonably satisfied, but a few issues make me wonder if I'd be better off scanning myself:
- Cost. The scans at the lab I'm using are quite expensive. Especially for a reasonable resolution (I generally order their highest res scans at 8MP - good enough for a majority of my needs but given a choice I’d like a bit higher). At their current rate I'd save the entire purchase price of a scanner by scanning just 5-15 rolls myself (5 rolls gives me an entry level flatbed, 15 a dedicated 35mm one)
- Consistency. The lab scans are usually nice enough in isolation, but often both colour and sharpness vary greatly from frame to frame.
- Tonality. The tonality of the lab scans I'm getting are generally a bit harsh for my tastes, and compression makes it hard to make any major adjustments in post.

That's a bit about what I'm seeing. As for the scanners I've been eyeing, I'm mainly interested in a Plustek Opticfilm 8200i since I'm shooting mostly 35mm. If I'd be able to get reasonable resolution (good enough for 20x30cm prints, the largest I ever do with any regularity) from a flatbed I'd consider going for that, since I occasionally shoot 120 as well (the Canon 9000F probably my prime candidate).

Moving on to the questions:

- Ease of achieving quality. Are my expectations unreasonable in thinking that I'd be able to get comparable results to the lab scans by scanning myself (with a reasonable amount of practice and effort, i.e. after a few rolls)? Should I completely disregard the flatbeds for my resolution needs?

- Speed. What is a reasonable expectation for the time it would take to scan a 36 exposure roll, excluding post work? An hour, two? Does it differ greatly between a flatbed and the Plustek.

- B&W vs Colour. I'm currently shooting both. Is achieving reasonably accurate colour scans far harder than getting B&W scans of comparable quality? I could see myself shooting B&W only while coming to terms with scanning, if it simplifies the workflow greatly. Does Silverfast’s Negafix profiles offer good enough results for colour consistency (mainly asking since it’s included with the Plustek)?

Sorry if that got a bit long winded or if any of the questions seem basic. I'm trying to learn and thankful for any input.

Regards / KJ

I am a diehard, but flatbed scanning "KICKED MY A__!!"

HUGE PITA.

But I finally realized the problem. My expectations were far too high.

After 4 new scanners, I realized that I actually expected high quality results, so I use labs and often drum scans for occasional scans.

I do not have, nor will I devote the time for the results obtainable by consumer scanners, particularly flat bed. My expectations can only be met by someone with much more expensive equipment than I am willing to buy, and who does scanning for a living, or as a part of his/her job.

Drum scans are surely expensive until you factor in equipment cost and time spent. If the difference is a salable piece of photography vs. just so/so forget scanning at home. :bang:

 

[edge100]

I am a diehard, but flatbed scanning "KICKED MY A__!!"

HUGE PITA.

But I finally realized the problem. My expectations were far too high.

After 4 new scanners, I realized that I actually expected high quality results, so I use labs and often drum scans for occasional scans.

I do not have, nor will I devote the time for the results obtainable by consumer scanners, particularly flat bed. My expectations can only be met by someone with much more expensive equipment than I am willing to buy, and who does scanning for a living, or as a part of his/her job.

Drum scans are surely expensive until you factor in equipment cost and time spent. If the difference is a salable piece of photography vs. just so/so forget scanning at home. :bang:

The answer is in this thread.

Scanning does not need to be a trade off between ease of use, cost, and quality.

 

[vbsoto]

I've been scanning a back catalog of negatives which obviously is my only option for getting digital files. In contrast, I have film cameras but rarely shoot film now because of the convenience. I don't think I could return to a film only workflow for new work. I do enjoy the scanning process periodically time permitting and the enjoyment of seeing old work. I used a Epson 3170 for years until it broke and now have an Epson 4490 ICE function. I've used both VueScan and Epson scan software. Both produce best results with frame by frame adjustments prior to running the batch. Figure 10 minutes of prep for each batch run.

 

[ColSebastianMoran]

This is a tricky question, and has been from the dawn of photography.

My take: No photograph is a true measure of the light values of the scene. Produced by the scene, but not a straight measure of the scene. B&W has a curve (with a toe and a knee). Color is much more complicated, our photo-gear produces good color because of good engineering. Films have personality, so does a DSLR (often with options for rendering style). It's not just a clean measurement of the scene.

Are the colors in your photo accurate? No, they aren't. They've been massaged by a process or a response curve or both. So to an extent, I'll agree, who cares if it's theoretically accurate?

The important question to me is: Do the colors in your photo deliver on your photographic (artistic) intent? Obviously, when trying for a realistic rendering, if the colors are "off" that won't be an effective photo. Of course, colors that are really "off" can produce an intentional non-realistic rendering that works.

The scans by @edge100 look pretty good to me. Good resolution, sharp, realistic renderings, and none of the color or tonality looks "off."

I do suspect there are color theory violations when we use white balance to correct for the film base, use an imperfect light source, and just invert the data (which has been gamma encoded). And, we aren't doing specific calibrations for each film stock. I suspect a shot of a ColorChecker might not be "accurate." I suspect that if we were shooting for a catalog or archival shots for a museum, we would have a better chance getting accurate product colors with a major brand DSLR than by using this process on C-41 film. Mini-labs had good engineering for producing color that customers would accept. Shooting a negative with a DSLR doesn't have the benefit of that engineering.

I believe the people at ColorNeg have addressed these theory questions with considerable skill, but they disclaim using their product with DSLR pseudo-scans, only with "flat" scanner scans. I would love to see a software product which would use their math on DSLR pseudo-scans.

I've experimented with a process pretty much like yours. Here's sample, not too bad, but not perfect, and it didn't take much work.

 

[kj_]

A busy weekend has mostly kept me away from the Internet. I've checked this thread a few times, but no time to write anything until now.

I'm amazed and thankful for all the great input. So many thoughtful replies and a lot of valuable insights in this thread. I almost feel like replying to each entry individually, but that's obviously not practical for anyone.

I guess my key takeaways so far are:

- Don't underestimate the tedium.
Almost everyone brings up the amount of time scanning takes. I need to include that in my planning and in the structure I chose. I also need to find a feasible way to identify what negs to scan, instead of always aiming for entire rolls. The idea of shooting the film strips with my phone and then inverting the results sounds like an interesting approach to this issue.

- There are few absolute truths
I probably shouldn't be surprised by this, but writing the original post I expected there to be some pretty clear answers to some of the questions. Obviously that's not going to happen. Like everything in photography (and life in general) choices in approach and equipment depends on many factors. Tradeoffs between quality, resolution, price, convenience, speed etc makes the decision matrix enormou and a great choice for one person might be awful for another. It's also pretty much impossible to know for sure without getting your feet wet. I would've never known wether I like or dislike shooting rangefinders (love), medium format (like) or pro SLR:s (dislike) without trying. Likewise I probably just need to give scanning a go at a level that seems reasonable and adjust from there.

- Dust will be an issue
No further comments.

I'm sort of closing in on a decision, or at least a starting point. I'll keep reading as long as anyone is writing as well as keep you posted on the wetting of my feet.

Regards / KJ

 

[edge100]

Well, that's why my original question was about the CRI of the LEDs in the pad that you are using. They do "look good" - but that's a subjective measurement of the digitised negatives. However, I am talking about objective, measurable accuracy of the representation of the colours. Even though you, or others, might actually like how your digitised negatives look, there is a distinct possibility, because of the high volume of cheap, low-CRI LEDs available on the market, that they could look different as soon as high-CRI LEDs begin to get used. If this is the case, it is misrepresenting the film stock (B&W included).

It may not matter to a hobbyist, such as yourself, who are typically more interested in seeing the final product, and then making your own valuations - but for professional photographers there is more on the line. A professional has to satisfy someone else's requirements and expectations. For instance, if I am going to be setting up a shot for a product in an ad, I have to make sure that the product is accurately represented. It will negatively affect my business if I can't deliver that. However, part of my interest also lies in identifying and utilising tools well. I can look at lower CRI lightpad digitisation as a potential tool in getting a look I want (like selecting a Ektar over Portra, or the Canon 50 1.5 over the 50 1.8) because it's influencing the curve of the film.

We really don't know until we can compare a negative that has been wet-printed or we somehow find out the CRI of the LEDs in the lightpad you are using. If it's somehow found that the LEDs in that pad are high-CRI, then we can say "oh yeah, his digitised negatives are a great representation of the ideal results from this method".

All valid points.

On the other hand, I wouldn't shoot a commercial job with colour negative film. Interpretation of the negative has always been a part of shooting negatives, whether they were wet printed or otherwise. There is no objectively "correct" colour from a negative; there is only interpretation, which may or may not be true to what the manufacturer intended.

A commercial job, shot on film, to me suggests reversal film and a colour managed scanning workflow. Which, incidentally, is entirely compatible with DSLR scanning (IT8 targets work just as well here).

 

[jnclde]

Buy yourself a Pakon f135 and enjoy your time shooting rather than scanning. The quality is there. And if you really want to high resolution you can use software like blow up 3. Or if the negative is that great you can get it done.

The latest image are all my samples from the Pakon.

https://www.flickr.com/photos/80442584@N07/