Flat Field Correction

[JDMXCIV]

I can’t figure this out at all. I’ve read all the blogs, all the threads with no success. I have a calibration frame and the negative I want to correct. Select them both, apply the correction and nothing happens. The files just stay RAW. I’m using a Sony a6300 with a 7artisans 60mm macro. Viltrox light source and Essential Film Holder. The Sony RAW files are .ARW. I just can’t figure this out and it’s killing me.

What really sucks is that I can’t find any videos about this either. I’ve followed the instructions from Adobe, the Pixl-latr article and others in this forum and nothing is working.

I’ll include a dropbox link to the RAW files if anyone out there has the time to help.

https://www.dropbox.com/scl/fo/5uh33m68ku4cn8glz32go/h?rlkey=4pdvbk71sj1pp87dafcd0absz&dl=0

[Digitizer]

Tested FFC with your takes and found out that FFC is not working reliably with Lightroom Classic version 13.2 on macOS 14.4.1 on 5K iMac 2019.

The one way that worked twice needed the following steps

• remove test images from LrC
• restart LrC and tell it to optimise the catalog when quitting
• duplicate the test images and number them to produce a sequence
  
  
• select the leftmost three pairs and apply FFC (in its default settings)
• check results, here is what I got:
  
  
  
  Note that two images are now DNG instead of ARW and that the folder, as seen in Finder, contains more files than the ones we see in Lightroom
• sync the folder to add the missing images
  
  

I never got DNG files with one pair only and when redoing the test without recreating the duplicate files again. All things considered, the implementation in the current version of Lightroom classic is absolutely unusable from an operational point of view.

[JDMXCIV]

Thank you so much for taking the time to do this. I will attempt this method in a few minutes. I’ll also scan the whole roll. I just wanted to test it quickly last night with one image and one calibration. Do I need to interleave the calibration frame between each negative? From so many things I’ve read, most people only need one calibration at the end. But this version of LrC must need multiple calibration frames interleaved. Would using an older version of LrC fix this issue? I’m using a M2 Pro MacBook on its current OS.

[Digitizer]

@JDMXCIV

Do I need to interleave the calibration frame between each negative? From so many things I’ve read, most people only need one calibration at the end. But this version of LrC must need multiple calibration frames interleaved. Would using an older version of LrC fix this issue?

Many questions - and hard to answer, considering the shaky nature of FFC in my tests.

Many users here will love to read what your experience will be, thank you in advance.

[shijan]

your scans are 3 stops underexposed. this is a problem

[JDMXCIV]

I’ll try a few things today and come back with my findings. Thank you for the work you already did to give me somewhere to start.

[JDMXCIV]

Interesting. If I WB, crop and convert the negative normally without FFC, the image looks perfectly fine (aside from the issue I need FFC to fix) . Are you looking at the scan with the black border? I scan all of my shots in aperture priority, F8, ISO 100 and set the exposure comp to ±0.0

[Belinda]

Concerning the underexposure pointed out by shijan, this is indeed a problem. Severe underexposure of the extent illustrated will result in noise and loss of detail.

Auto exposure is rarely a good idea unless the scanned images are all of medium tonality - a tonal mixture that averages to mid-tone grey. Using any auto-exposure mode, images with predominantly brighter tones will be underexposed while darker images will be overexposed.

To obtain a correct digital exposure you should apply exposure compensation until the right side of the histogram is almost touching the right-hand side. This will ensure the capture of maximum available data.

As you are using a good macro lens, I would also suggest using a larger aperture say F4 or F5.6 to take advantage of the lens’ best performance and avoid softening due to diffraction.

Exposure compensation for digital scanning is best applied through shutter speed (unless you have control of the intensity of the light source) as the aperture is effectively fixed and raising ISO is undesirable if noise is to be avoided.

[Digitizer]

Gave FFC another test and this time, it worked. The reference shot was really bright but not clipped and FFC created the DNG file, but again, I had to sync the folder to make the file available in the Library module.

Here, we see the reference shot (edited to show something), the original RAW and the DNG output. The one image between the two images is an inverted quotient between the two images…heavily exaggerated to make anything visible at all. Most of what I see is that the output image has lost the original’s slight greenish tint.

FLTR: Reference, Input, Quotient, Output
Photo: Camera scanned Kodachrome transparency, shot and developed in 1972.
Quotient: Layers in Photoshop, treated three times with a tone curve that starts at 250 (instead of 0)

[JDMXCIV]

Thank you again for running another test. I needed those original negatives scanned for a YouTube video so I ended up just scanning on my Epson v600. Thankfully 6x7 and the v600 pulled enough quality. But now that I have time this weekend, I’m going to do an extensive test of exposure and aperture.
Hopefully FFC works for me like it ended up working for you. So you didn’t need to do any of the prior steps before the folder sync in your previous post (remove, optimize, duplicate)? You just applied FFC as it should normally be applied?

[Digitizer]

@JDMXCIV

So you didn’t need to do any of the prior steps before the folder sync in your previous post (remove, optimize, duplicate)? You just applied FFC as it should normally be applied?

Well, this test was done with new photos that I took a few minutes before testing. I set exposure in manual mode for the transparency, then removed it and took other shots with increased exposure time (every thing else should be left alone (lens settings incl. aperture)

Tested FFC with a series starting with the reference and a series ending with a reference, both worked this time, but I’ve not tried reproducing FFC with the same images.

The most annoying component of FFC is, that I had to sync the folder to get the DNG files. That should not be necessary imo and looks like a buggy implementation to me.

---

Update: Removed all DNGs and re-did the test. It worked as it did before.

[JDMXCIV]

Sounds good. I’m going to run different exposure times and a few different aperture tests today. All with a calibration frame at the end and try to FFC. I’ll try FFC with a Calibration at the end, front and both. Then come back with my results.
That is very annoying that you have to sync the library, but at least it’s working the way it should (besides that lol)

[JDMXCIV]

I also got it to work. I did a pretty extensive test tonight. Will share all the details tomorrow.

[JDMXCIV]

I’m going to mention some keywords in this so that others can find this if they are running into the same problem. Obviously it can be vary between different users that have different setups. I personally am using a Sony a6300 APSC camera, 7Artisans 60mm 2.8 Macro Lens, Viltrox L116T light source, and the Essential Film Holder (EFH).

I’ve been battling the Orange Haze/Vignette/Light issue on my scans since I started digital camera scanning. Some rolls are worse than others. But majority of the time I’ve been able to correct it with Manual Vignette options/Masking in Lightroom. But I was determined to figure this out so that all my future scans were perfect and didn’t need some workaround band-aid fix.

I did quite an extensive test with a different roll of film shot on my Pentax 67. This roll didn’t actually show much of the issue, but I was able to get the Flat Field Correction (FFC) to work properly as it should. I also wanted to find the best way to expose for the negative and the sharpest point of the lens in the process. So I tested my usual f8 at “correct exposure” (±0.0), 1 stop over, and 2 stops over. Then I tested f5.6 at “correct exposure” (±0.0), 1 stop over, and 2 stops over because I was seeing some people say that f5.6 was also another option for the sharpest point of this specific lens. I did these first two tests with Aperture Priority like I would usually for quicker workflow and less fiddling (even though I just use the remote app on the phone to not touch the camera). All shot at 100 ISO.

Then I tested Manual Exposure like @Belinda recommended. Same settings as the first two tests. f8, “correct exposure” (±0.0), +1, and +2. f5.6, “correct exposure” (±0.0), +1, and +2.

I also tested f11 just to be sure of the sharpest point of the lens corner to corner. Also “correct exposure” (±0.0), +1, and +2, but only tested one single frame from the roll (yes, I shot the entire roll for every previous test).

Before getting into my findings, I want to address Aperture Priority. Shooting in this mode at ISO 100, I noticed almost no difference in exposure time comparing it to the Manual Exposure. One of the test that I found to work the best in all aspects (Aperture Priority, f8, ISO 100, +1 exposure comp) was completely identical in exposure time to the Manual Exposure test with the same parameters (go figure). There were only two frames of the 9 frames on the roll (yes, my 67 has the frame spacing issue and is out for repair now) that had a different exposure time. But the difference was only by one click.

This difference had no affect on creating the DNG through FFC nor had any affect on the positive images after being converted with Negative Lab Pro (NLP)

f8, ISO 100 and +1 metering/exposure comp appears to be the best at converting the RAW to a DNG with FFC, created the best exposed images when converting with NLP and had the sharpest results corner to corner.

Any test I conducted at “correct exposure” (±0.0) had issues converting all frames from RAW to DNG for FFC.

(you can see that not all RAWs converted to DNG)

It would convert most of the frames, but would leave one or two frames out. I feel that exposing the negative this way confuses Lightroom in identifying the negative. Exposing +1 or +2 worked every time converting all negatives from RAW to DNG. But images exposed at +2 would be very underexposed and have a difference in color once converted to positive in NLP.

So basically shoot a roll at the sharpest aperture of your lens, at the base ISO of your camera and set your exposure time to be +1 stop over on your metering or exposure comp (whether you want to manually expose or use aperture priority). Then take a calibration frame after the roll is shot by taking your film holder off the light source (I used a frosted acrylic pane over the light source to diffuse the light since the Viltrox has visible LEDs). Upload your images into Lightroom, White Balance, Crop, select all images including the Calibration Frame at the end and select Flat Field Correction in the Library tab.

Here are my FFC settings

Then just convert/adjust/make copies as you normally would in NLP

Lens Sharpeness:
First, comparison between f5.6 & f8, ISO 100, +1
Left is f8 & right is f5.6

Zoom is at 200%. Images have no edits or sharpening, straight out of conversion. You can see that f5.6 gets a little blotchy in the edges/corners (can’t see the grain well)

Next, comparison between f8 & f11, ISO 100, +1
Left is f8 & right is f11

Zoom is at 200%. Images have no edits or sharpening, straight out of conversion. It appears that f11 also gets blotchy in the edges/corners (can’t see the grain well)

Sorry for such a lengthy post. I really wanted to help others and hopefully those with these issues find this helpful. I had so many issues trying to find a resolution online. There are hardly any videos that show this process and damn near no videos that show this process when it comes to film scanning. There are some articles online about this like ( Defeating the Orange Haze - Lightroom Flat Field Correction - An Essential Tool for Digitising Film? - pixl-latr ) which helped a lot. But doesn’t go into such detail or explain what could cause one to run into the issue of their RAWs not converting to DNG.

I’ve gotten into making YouTube videos recently about my film photography and now consider making this a video since there are no other resources out there explain this in detail. I know some people would appreciate a visual version of this test/guide like I would have.

@Digitizer

[Digitizer]

If I read your post correctly, these are your findings

General

• use the lens at its optimal aperture for sharpness
• shoot in aperture priority mode
• overexpose by one stop

Flat Field Correction

• after scanning, don’t touch the lens
• shoot the backlight as above to get the FFC reference
• apply FFC, then convert all images with NLP

Doing the above gave you

• converted images with colours you liked best
• flat field corrected dng files for all scans

[JDMXCIV]

To sum it up, yes.
I personally like using Aperture Priority for quick workflow, but you can manually expose if that’s what you prefer. I found no difference. Overexposing +1 at the aperture I found sharpest provided DNGs in FFC every time and gave me the best NLP positive converted results.
I’ll do a run of 35mm that I need to develop tomorrow to confirm these findings.

[Harry]

I haven’t downloaded your pictures yet but do you know what is responsible for your needing to use Flat Field Correction? Squinting at your thumbnails it looks like slight vignetting from the lens rather than anything induced by the interaction of your light panel and film holder. If it is lens vignetting then you could provide an appropriate adjustment using the radial ND filter in Lightroom and apply it using a setup.

On the subject of aperture priority exposure v manual, Nate recently suggested that he bases his exposure so that the light source with no negative in place is just at the point of clipping and then uses that manual exposure for the whole roll. That does at least provide a constant ‘base’ exposure which you can adjust to personal taste if you want to. This is for colour negative of course.

[Digitizer]

@Harry

Nate recently suggested that he bases his exposure so that the light source with no negative in place is just at the point of clipping and then uses that manual exposure for the whole roll.

Indeed, and this makes scanning straightforward, specially if all negatives are similarly dense. Aperture priority exposure can be equally useful, if we watch out for clipping.

Please note that LrC applies some hidden rolloff and that some sensor have a “linearity limit”. Both areas above these should be avoided.

Last but not least: in my tests, I found that correction went in the wrong direction, making some areas brighter instead of darker. I’ll have to check that again later on.

[JDMXCIV]

I haven’t seen that post about Nates test on basing the exposure off the light source first. Sounds interesting and I’ll have to try it out some time.
My issue with using masks or vignette tools in LR is that I’ve had different results each time. At one point, I built presets for certain formats I’ve scanned and they would not work universally/consistently. I would prefer a method that corrects the issue for each specific roll that I’ve scanned and that’s where FFC comes in. It corrects the specific vignette I got with each individual roll I’ve scanned by providing a calibration frame at the end of each roll I’ve ran through.

[Harry]

Well if you’ve already tried it and didn’t work consistently that’s fair enough, surprising that it wouldn’t be consistent for the same format, lens and lens aperture though. In theory I thought it would it would be easier than messing around with FFC but there’s theory, and then there’s practice…