The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

LCCs or radial filters for correcting lens casts...

John Leathwick

Well-known member
(This post could have easily gone into the image-processing thread, but I have put it here because it tackles an issue that mostly crops up for users of medium format technical cameras.)

Quite a lot of my recent shooting with my Fuji GFX 100s/F-Universalis has been with the three wider lens options I have available to me, i.e., an S-K ApoDigitar 35mm L-88, a Digitar 47mm, and an Apo Componon HR 60mm. In using these lenses, I have become increasingly disenchanted with Adobe’s current implementation of lens cast calibration images (LCC’s) in Lightroom (LR), which they term flat field correction. In LR’s earlier implementation it was possible to highlight a set of images, each with their corresponding LCC, and it would reliably return corrected images converted to .dng format, with the original raw files optionally left in the directory. In the current implementation, I can only rely on it to perform as expected when I select a single image and its LCC; sometimes it works well, but at other times it seems to correct for illumination but not colour casts, even if the colour cast correction option is checked.

This set me thinking about whether I could build a smoother workflow to correct for vignetting and colour casts using Lightroom’s radial masks. In my initial experimentation, I started with my S-K Apo Componon 60mm, a lens with moderate vignetting, particularly when shot wide open. I started by shooting LCCs at a range of different apertures, and after importing these, experimented with settings for inverted radial masks until I found one for each LCC that minimised the width of the histogram and corrected for the blue colour cast in the outer part of the image. I then saved each mask as a named preset, one for each cardinal aperture from F/4 through to F/11. Correcting images taken at any of these apertures was then easily achieved simply by applying the preset after import, with results virtually indistinguishable from those achieved when applying an LCC using LR’s flat-field correction option.

By way of qualification, for some images taken with moderate shift, I needed to adjust the position of the mask to centre it on the estimated centre point of the lens, guided by the evenness of the correction, particularly in the corners. In some cases, where I didn’t want to apply full correction, I found that the slider that allows control of the strength of a mask enabled me to back off from full correction to leave a mild vignette.

I then extended this approach to my Digitar 47mm and ApoDigitar 35 L-88. Finding settings for the latter was the most challenging, requiring compensation not only for exposure and temperature (y-b) but also for tint (m-g). However, with masks saved as presets by aperture, results again very closely matched those achieved with an LCC applied using LR’s flat frame correction option.

Example results below show: an example radial mask; an uncorrected image taken with my 35mm L-88 at F/5.6, the same image corrected with first a radial mask, and then an LCC using LR's flat-field correction.

Based on these results, I then explored the extension of this approach to stitched panoramas, which I'll describe in another entry.

-John

SKAD35_L88_at_f11.PNG
ApoDigitar35Uncorrected.jpg
ApoDigitar35MaskCorrected.jpg

ApoDigitar35LCC_Corrected.jpg
 

John Leathwick

Well-known member
Once I got single image masks working satisfactorily, I then pondered the problem of dealing with stitched panoramas.

In this case stitching a pair of left-right shifted LCC’s is not possible, as the stitching algorithm has no useful image detail for alignment. My solution was to photograph an evenly illuminated external wall of our house, leaning two umbrellas against the wall to provide some detail for the stitching software. Once imported and stitched, I then used the same approach as I had with the LCC’s, manually adjusting an inverted radial filter until I had even brightness and colour balance across the entire stitched image. As I generally shoot images for stitching at F/11, I made presets for each lens at its maximum useable shift left and right, i.e., +/-9 mm for the Apo-Digitar 35mm, +/- 10 and 20mm for the Digitar 47mm, and 7.5mm for the Apo Componon 60mm.

Again, the radial filter for the 60mm stitch was the most straightforward, requiring just correction of exposure and colour temperature. The Digitar 47mm stitch with 10mm of shift was also straight forward, but when the Digitar 47mm was shifted +/- 20mm, I needed to add a linear mask at either end of the stitched image, the right-hand one to correct a magenta cast, and the left-hand one to correct a green cast. I’m not fully confident about this latter setup, but I can only see myself using this degree of shift on rare occasions – on the one real world image that I have tried it on, it has worked very well, requiring almost no further work other than use of a curves layer in PS to set the black and white point and balance the shadows and highlights. The 35mm panorama mask was a little more straight forward, given that shift with the GFX is relatively limited to +/- 9mm.

Images below show my umbrella image with the Digitar 47mm, an uncorrected panorama with +/- 10mm of shift, and versions corrected with a radial filter applied to the stitched image, and with LCCs applied to the images prior to stitching.

I'm really interested to hear if other people have explore this work around, and if so, what other tricks could be added to my approach.

-John

Digitar47FlatsShifted-10mm.jpg
Digitar47_10mm_uncorrected.jpg
Digitar47_10mm_radialmasked.jpg
Digitar47_10mm_LCCs.jpg
 
Last edited:

dchew

Well-known member
Gosh John, It might be worth buying a Capture One perpetual license for $300 just so you could use its LCC process. I find it to be darn good. I realize you will have to export a tiff to get the file into LR when you are done. That may be a non-starter for you, but I do it all the time. I don't like C1's stitching process, so I export the LCC-adjusted file as a tiff into PS for the stitching. Then it goes to LR for local adjustments, fine tuning and catalog / storage.

Dave
 
Last edited:

John Leathwick

Well-known member
Gosh John, It might be worth buying a Capture One perpetual license for $300 just so you could use its LCC process. I find it to be darn good. I realize you will have to export a tiff to get the file into LR when you are done. That may be a non-starter for you, but I do it all the time. I don't like C1's stitching process, so I export the LCC-adjusted file as a tiff into PS for the stitching. Then it goes to LR for local adjustments, fine tuning and catalog / storage.

Dave
Thanks for the suggestion, Dave. I've considered that as an option, but given the success of the radial masks that I've built, and which only took a few hours in total, I'm quite comfortable with the prospect of sticking with them as my primary tool. The big advantages I see are their non-destructive nature, meaning that the original raw image is always preserved, and avoiding having to export each image as a larger, uncompressed file that then needs to be re-imported.

The biggest drawback that I've found so far is LR's trick of reshaping a mask when applying say a landscape preset to a portrait image - it reshapes it according to the proportional changes in the horizontal and vertical image dimensions. Although it's straight forward to reshape the mask back to circular shape, I've now saved 'portrait' versions for my more frequently used masks that get around the problem. As a result, for the majority of my images, the relevant mask can be simply applied with one click in an easily reversible process.

-John
 

dougpeterson

Workshop Member
It’s hard to imagine giving up the C1 LCC workflow in favor of this. But different strokes for different folks - definitely a creative workaround.
 

rdeloe

Well-known member
It’s hard to imagine giving up the C1 LCC workflow in favor of this. But different strokes for different folks - definitely a creative workaround.

Like John, I use Lightroom -- not C1. I only use LCCs for one lens. Buying C1 and using it to develop only those files, and then having to bring them into LR as TIFF where I've now lost the RAW functionality, is not a solution for me.

It's frustrating that Adobe created this functionality and then half-assed its execution. I managed to get by with it in 2024, but with recent updates it's actually getting worse.

For someone like me, John's solution is a nice backup strategy -- and may even become the better choice for my work if Adobe makes flat field correction worse!
 

tenmangu81

Well-known member
The issue raised by John is also relevant for people (like me ;) ) who haven't buy an X2D yet, and who would like to buy a XCD28P. I would be more than happy if LR could correct lens colour cast/vignetting of this lens using a workaround. Thank you John for your tricks !
 

glaiben

Active member
I like this workaround for LCC's, given I use a CFV-100C and prefer to do all my post in LrC. I understand the inversion of the radial gradient. What are you changing to modify the vignetting? Exposure? Also, what about the color cast - curves?

Thanks (and sorry for what are likely, very basic questions).

...gregg
 

John Leathwick

Well-known member
It’s hard to imagine giving up the C1 LCC workflow in favor of this. But different strokes for different folks - definitely a creative workaround.

Doug, I can't imagine that either - "if it's not broke then don't fix" sounds like it applies for C1, but sadly in my experience it doesn't apply to Lightroom.

Yesterday I repeated a couple of my stitched mask shots using some taller objects to ensure that I was on an evenly illuminated part of my target wall, and that improved my masks even further. I then went back to some earlier images where I had both my original raws and LR flat-field corrected images, restitching the originals and applying my updated masks to the stitched panorama. The storage advantages are considerable: the original raw is about 10% smaller than the dng file produced by LR, but the need for duplicate images plus LCC images goes completely - so about a 2/3 reduction in storage requirement if you keep the lcc and original. Finally, the workflow was so much simpler - clicking a preset for the stitched image, rather than individually correcting the component images before stitching. Importantly, the final mask-based outcome was indistinguishable from the more tedious and space demanding lcc-based route.

So as an LR user, I can't imaging going back to LCCs even if Adobe substantially fixed the current issues with their 'flat-field correction' - having done the setup work, masks are just easier all round.

-John
 

John Leathwick

Well-known member
I like this workaround for LCC's, given I use a CFV-100C and prefer to do all my post in LrC. I understand the inversion of the radial gradient. What are you changing to modify the vignetting? Exposure? Also, what about the color cast - curves?

Thanks (and sorry for what are likely, very basic questions).

...gregg
Greg, if you look at the very first image I posted you will see that in the most extreme case I'm modifying exposure, colour balance and tint.

-John
 

dougpeterson

Workshop Member
If I could offer the archive oriented advice that you should still shoot an even field image even if you don’t use that currently in your workflow.

Storage is cheap and the future is hard to predict. If you move away from adobe, or adobe moves away from you, then having these corrections only in adobe-specific metadata will limit your ability to reproduce the end product. If you have an even field reference capture you can redo even fielding any number of ways.

Just my two cents.
 

rdeloe

Well-known member
If I could offer the archive oriented advice that you should still shoot an even field image even if you don’t use that currently in your workflow.

Storage is cheap and the future is hard to predict. If you move away from adobe, or adobe moves away from you, then having these corrections only in adobe-specific metadata will limit your ability to reproduce the end product. If you have an even field reference capture you can redo even fielding any number of ways.

Just my two cents.

That's a solid recommendation Doug.
 

rdeloe

Well-known member
I'm with Doug on this one. LCC's are great for dust removal (well, masking), and that's difficult to manage any other way.

Huh, interesting. I've never seen that benefit.

The LR process could be so much better. They just need to hire a good co-op student and put them on the task. ;)
 

John Black

Active member
C1's LCC is pretty powerful IMO. The LCC itself can modulate fall-off, color and dust. It's slider based too, because some times the corrected file looks too correct. Paired with the Lens Correction palette, that adds distortion correction which factors in the lens movements, C1 is a pretty stout app in this area.

So long as I'm taking good notes in the field, I don't bother doing the LCC's when on site. Once back at home and after have I figured out which shots are the "keepers", if an LCC is needed, then I'll recreate the settings on the camera and take the LCC from home. Saves alot of hassle in the field.
 
Top