What software do you use after scanning negatives with a camera to turn them into positives?

[runswithsizzers]

Yes, there is. I have used it (Vuescan) just for this - scans from a DSLR. But my results with RawTherapee have generally been better.

Interesting! Can you please provide a source for more information about how to open a camera file in Vuescan? Thank you.

 

[grat]

If you're going to use a cheap $50 macro lens with a sophisticated DSLR, why are you bothering? I'm not a lens snob, but there is some truth to the idea of a modern sensor needing a bit more optical clarity than a decades old macro (or reversed prime) will provide. The goal is resolution, after all.

I can easily get 15 MP out of a 35mm frame with the Epson by scanning at 3200 PPI. I could probably bump that up to at least 20 MP by scanning at 4800 PPI. Alternatively, I can "scan" with a DSLR, and get a 28 MP image with one shot.

But I can also load up 21 frames of 135 film into my scanner, set the crop for each frame, turn on ICE, point it at a directory, and go do something else for the next 30 minutes. That's impossible to do with a DSLR (unless you have the cameradactyl gizmo), and the results, as I said, will be better-- but not ncessarily "enough" better to justify the extra mechanical processing (and you've still got to do individual inversions, which the Epson will do a credible job of).

 

[urnem57]

Good debate. I am learning a lot.

 

[albireo]

I guess most of us were looking for ways to easily and quickly digitize large libraries. And (almost) everyone has a Digicam
Digicam scanning is cheap, fast and anyone can do it.

Who are you speaking for? Not for me!

See, there's another camp: hybrid shooters. These are people who don't dislike a digital end-product (a file), but want to get there using film photography equipment and enjoying the fun of development and scanning.

Many of us hybrid shooters don't own a digital camera anymore. I really don't miss my DSLR. I have my camera phone for digital shots.

Anyhow - for me buying an expensive digital camera specifically for scanning makes exactly 0 sense. A good film scanner blows away these mediocre 'DSLR scans' I see around, at a fraction of the cost. A film scanner is also usually more compact than a DSLR scanning rig, and needs less calibration and maintenance.

 

[grat]

You are forgetting about the noise/dynamic range. Scanner users believe they're looking at grain, but in fact they're looking at a grain+noise mixture. Some even invented a term "grain aliasing" to describe the difference between their scans and prints. Camera scans are cleaner.

Hrm. Apparently I'm not arrogant enough to qualify as a DSLR user.

That's silly. DSLR users operate at another level. We scan a 36exp roll in three minutes and we do not need ICE because camera scanning does not attract dust.

.... right.

So... which is flatbed, and which is dslr?



One was scanned with an Epson V800 @ 3200 PPI (14.5 MP), with ICE, manually cleaned up.

One was imaged with an EOS 90D (28 MP), manually cleaned up (it had been in the same room as the Epson, so it was dusty. ).

Both have had mild color tweaks, and a bit of sharpening done in post, and then resized to 1920x1280. No, I did not spend the time to make the colors identical.

 

[grat]

I've never argued that DSLR can't produce higher resolution images. It obviously can, which is why I downscaled both photos, and 1920x1280 was a deliberate choice, as it's a nice size for sharing on the internet.

The argument for flatbed scanners is that out of the box, with minimal effort (not "no effort"-- it takes some practice and work to get reliably good scans), they produce results with less effort than DSLR-- again, once you've got a process up and running, DSLR can be streamlined as well, but there's considerably more work involved.

FYI, my DSLR setup is the 90D, the Canon EF 100mm f/2.8 macro (not the "L"), an inexpensive LED panel off amazon, a pixl-latr negative holder, and a tripod with a hanging right-angle beam. I think the tripod was around $100, the pixl-latr around $60 and so was the Raleno LED panel-- and $2 of ABS plastic sheeting to mask off the light.

If I want to make a large digital print from a negative, I'll use the DSLR-- but for casual use, the scanner meets my needs.

The cat photo was 35mm-- the cat is substantially larger.

 

[Nelari]

Interesting! Can you please provide a source for more information about how to open a camera file in Vuescan? Thank you.

It's very simple. When I start Vuescan, File -> Input -> Source -> File -> Files lets me select a a file for input (click the button with an "@"). It doesn't seem to matter whether the file comes from a camera or a scanner.

Of course, I don't have a scanner at all. Maybe things are more complicated if one does have a scanner?

 

[Nelari]

I am using my own workflow based on a self-written Matlab script.
.................................................
Feel free to pm me if you want more details.

For a newbie like me, pm's are not allowed. But I was inspired by your post to look at Matlab – well, really Octave - and was surprised and delighted to find a good-looking image processing package. So far, I'm struggling with the matrices. But for starters, may I just ask a question about the orange mask of color negatives? Is getting rid of it just something like subtracting one matrix (NxMx3) from another? A linear operation?

 

[Huss]

That is exactly how Negative Lab Pro complicates my workflow in Lightroom.

I digitize film with my digital camera, and import the camera's RAW files into Lightroom. If I don't duplicate the NLP-converted file as a TIFF/JPEG, then normal Lightroom tools work backwards, so further editing with Lightroom tools is difficult or impossible. I use Lightroom's local adjustment tools a lot - adjustments which are impossible using the NLP module. So, using NLP really forces me to duplicate the RAW files as a TIFF/JPEG.

But copying the RAW file as a TIFF/JPEG means having two different versions of the same photo in my Lightroom catalog to keep track of. And if I want to use 16-bit TIFFs for maximum editability, then my storage requirements are doubled.

You really should use a different product due to all those complications.

 

[Huss]

Hrm. Apparently I'm not arrogant enough to qualify as a DSLR user.



.... right.

So... which is flatbed, and which is dslr?

View attachment 273186 View attachment 273187

One was scanned with an Epson V800 @ 3200 PPI (14.5 MP), with ICE, manually cleaned up.

One was imaged with an EOS 90D (28 MP), manually cleaned up (it had been in the same room as the Epson, so it was dusty. ).

Both have had mild color tweaks, and a bit of sharpening done in post, and then resized to 1920x1280. No, I did not spend the time to make the colors identical.

I can understand the colours not being identical, but when one is so far off into the green spectrum - after your mild tweaks - I'm not sure what the point here is.
The one on the left looks better because the colours aren't all messed up. Did you create your own profiles for the green one?

 

[PhilBurton]

That is exactly how Negative Lab Pro complicates my workflow in Lightroom.

I digitize film with my digital camera, and import the camera's RAW files into Lightroom. If I don't duplicate the NLP-converted file as a TIFF/JPEG, then normal Lightroom tools work backwards, so further editing with Lightroom tools is difficult or impossible. I use Lightroom's local adjustment tools a lot - adjustments which are impossible using the NLP module. So, using NLP really forces me to duplicate the RAW files as a TIFF/JPEG.

But copying the RAW file as a TIFF/JPEG means having two different versions of the same photo in my Lightroom catalog to keep track of. And if I want to use 16-bit TIFFs for maximum editability, then my storage requirements are doubled.

Actually once you have done the conversion using NLP, you no longer need to keep the original scanned image RAW in your catalog. For a scanned image, a "RAW" file isn't really a true RAW file unless you do DSLR scanning.

 

[grat]

I can understand the colours not being identical, but when one is so far off into the green spectrum - after your mild tweaks - I'm not sure what the point here is.
The one on the left looks better because the colours aren't all messed up. Did you create your own profiles for the green one?

Yeah-- I'm not sure what was wrong with me when I did the color conversion (which gives away which is which fairly obviously). I think I'll claim it was late at night (it was). I've tried three times to get the colors right on that image, and every time, it's come out looking like.... excrement.

Going back and working with the original RAW, it's evident that there's a blue cast from the original capture-- probably the result of not having the right color temperature when I took the image (I usually set the color temperature on the camera and the panel to the same value). Correcting the color cast is doable, but it results in a drop in contrast in parts of the image-- the final image is fine, until you compare it with the Epson scan, at which point you start wondering why you bothered.

Here's a better version:



Still not as good as it should be-- which is kind of my point. DSLR digitizing can indeed produce superior results-- but a flatbed will produce very usable results for much less effort.

 

[etn]

For a newbie like me, pm's are not allowed. But I was inspired by your post to look at Matlab – well, really Octave - and was surprised and delighted to find a good-looking image processing package. So far, I'm struggling with the matrices. But for starters, may I just ask a question about the orange mask of color negatives? Is getting rid of it just something like subtracting one matrix (NxMx3) from another? A linear operation?

Yes it is indeed as easy as that. Not subtracting, but dividing:
let's assume R(m,n), G(m,n) and B(m,n) are your matrices representing the red, green and blue values of your pixels, each normalized so that the values of each pixel are within the 0 (dark) .. 1 (bright) range.
If the color of the mask is represented by the mr, mg, mb values you can remove the orange mask by computing
R(m,n) / mr
G(m,n) / mg
B(m,n) / mb

I realized that applying histogram equalization after this step improves things. Still, the colors are not always 100% accurate after inverting.
Applying Photoshop's "Auto color" function as a last step always make them great - obviously Adobe has tight lips about the auto color algorithm...

Too bad about pm, let me try to send you one from my side, you can take it from there if you wish to discuss more in pm.

 

[runswithsizzers]

Actually once you have done the conversion using NLP, you no longer need to keep the original scanned image RAW in your catalog. For a scanned image, a "RAW" file isn't really a true RAW file unless you do DSLR scanning.

My digital camera copies are RAW files. And if I delete the RAW file, then I will no longer have the option to redo the conversion with NLP, right? I think there is a very good chance that either the software, or my skill using the software will improve, so it's quite possible I may want to reconvert the file in the future. Or use it as starting point for a manual conversion. Make sense?

 

[runswithsizzers]

It's very simple. When I start Vuescan, File -> Input -> Source -> File -> Files lets me select a a file for input (click the button with an "@"). It doesn't seem to matter whether the file comes from a camera or a scanner.

Of course, I don't have a scanner at all. Maybe things are more complicated if one does have a scanner?

Thank you. I was able open a RAW file from my Fuji camera in Vuescan. However, the preview did not look at all normal, and I was not able to do anything practical with it. Maybe because it is a Fuji X Trans file and not Bayer? I was hoping to see if Vuescan could be used to do the color conversion to positive, using a RAW camera copy of a color negative - but apparently, not?

 

[PhilBurton]

My digital camera copies are RAW files. And if I delete the RAW file, then I will no longer have the option to redo the conversion with NLP, right? I think there is a very good chance that either the software, or my skill using the software will improve, so it's quite possible I may want to reconvert the file in the future. Or use it as starting point for a manual conversion. Make sense?

Agree with your point about your RAW files, but from the perspective of LR, you might want to keep the true RAW files outside of LR, and keep in the LR catalog only the files processed with NLP, since they are the starting point for all normal DEVELOP actions.

 

[grat]

One of the complaints I've always had with Lightroom (and Darktable) is that they feel you should have exactly *one* "catalog" that contains every single photo you've ever worked with or wanted to. I'd love to have separate collections such as "Masters", "archival", etc.. At the moment, I've got some 8,000 pictures in my catalog-- some are duplicates, some are pictures sent to me by other people, some I keep because I don't want to lose, some I keep because I'm too lazy to delete-- but for day-to-day, I don't touch 90% of those files. I can select "rolls" or "files" or "folders"-- I can select by tags, or lens, or camera, or aspect ratio-- but having to have everything in a single monolithic catalog is a bit irritating.

 

[Nelari]

Thank you. I was able open a RAW file from my Fuji camera in Vuescan. However, the preview did not look at all normal, and I was not able to do anything practical with it. Maybe because it is a Fuji X Trans file and not Bayer? I was hoping to see if Vuescan could be used to do the color conversion to positive, using a RAW camera copy of a color negative - but apparently, not?

My copy of Vuescan tells me Fuji RAF (*.raf) files are accepted. Unfortunately, I'm not at all familiar with Fuji's formats. Are there several? Both Canon and Nikon have two entries on Vuescan's list of formats accepted, but there is only one for Fuji.

 

[Nelari]

Yes it is indeed as easy as that. Not subtracting, but dividing:
let's assume R(m,n), G(m,n) and B(m,n) are your matrices representing the red, green and blue values of your pixels, each normalized so that the values of each pixel are within the 0 (dark) .. 1 (bright) range.
If the color of the mask is represented by the mr, mg, mb values you can remove the orange mask by computing
R(m,n) / mr
G(m,n) / mg
B(m,n) / mb

That looks simple, indeed. But how does one determine the mr, mg, mb values?

Too bad about pm, let me try to send you one from my side, you can take it from there if you wish to discuss more in pm.

This may become my post #20, which should make me eligible for private messages...

 

[runswithsizzers]

My copy of Vuescan tells me Fuji RAF (*.raf) files are accepted. Unfortunately, I'm not at all familiar with Fuji's formats. Are there several? Both Canon and Nikon have two entries on Vuescan's list of formats accepted, but there is only one for Fuji.

Interesting. I will research this more.

Have you (or anyone else) used VueScan to open a camera-copy RAW file of a color negative, and then converted that to a positive in Vuescan?

Would anyone care to speculate what would cause VueScan to render a RAF like this?

 

[etn]

That looks simple, indeed. But how does one determine the mr, mg, mb values?

I had the DSLR frame set up a bit wider than the picture itself, so as to include part of the unexposed negative.
The script would then read all values on, say, the first 5 rows ; average them ; and use this for mr, mg, mb.

I considered important at the time to sample mr, mg, mb on the same DSLR picture as the one I wanted to process, as the values are dependent on the exposure (which varies from one picture to the next in my workflow.)
Just thinking, the following might be worth a try: converting the RGB "scan" into a different color space such as YCbCr and removing the base color in Cb Cr only. This could make it exposure independent. Maybe we can even find standard correction factor for each kind of film: measure it once and re-use everywhere.

If you have any ideas or find good tricks, please share, thank you

EDITS: improving sentences and correcting mistakes...

 

[wiltw]

One of the complaints I've always had with Lightroom (and Darktable) is that they feel you should have exactly *one* "catalog" that contains every single photo you've ever worked with or wanted to. I'd love to have separate collections such as "Masters", "archival", etc..

You can have any number of LR Catalogs sitting on harddrive...each one with a unique name, but only ONE of them is ACTIVELY in use by LR! File Open (catalog selected by user)

Back to OP, I have tried built in capability within Paintshop Pro to convert negative to positive. The result came out inferior

to what the built-in 'scan negative' in the Canon-provided MP Navigator scanning program (supplied with my Canon flatbed many years ago) could do

[

As a result, I will not shoot negatives with my dSLR and then convert with postprocessing software. I wish someone could identify a program that does a reasonable job, comparable to my flatbed scanner application.

 

[wiltw]

Not following. Several people, myself included, have "identified" several programs that do an excellent job, with full-sized examples, in this very thread. Why do you keep wishing?

A number of software solutions were proposed, but it seemed that most lacked any commentary about how successful their particular suggestion was at the conversion. About the only suggestion thatt seemed to carry a hint of personal satisfaction was RawTherapeed, so maybe I will try than. Otherwise, it seems the comments thus far has been 'my scanner does a better conversion', as I had also experienced. The point was not merely to get suggestions of products, but a personal ENDORSEMENT of said product!

In giving some further though to the process of shooting negs with dSLR, and then using some software to convert to a positive image, the thought does arise about the Color Temperature of the backlight source used while photographing the negative. Setting the camera WB does nothing to the RAW data which is stored, so just how is any postprocessing software going to deal with the unknown of WB setting...any 4 photographers could be shooting with 4 different WB sources!

 

[grat]

You can have any number of LR Catalogs sitting on harddrive...each one with a unique name, but only ONE of them is ACTIVELY in use by LR! File Open (catalog selected by user)

That's good to know. I should bug the Darktable people.

 

[wiltw]

Maybe it got lost in the comments, but Negmaster and Negative Lab Pro both deserve a closer look for inverting color. I have provided both the endorsement and full-sized sample scans you can evaluate.

Thanks for pointing this out. Gotta go back and re-read your posts in this thread.