Vignette correction

The lens I took this photo with is a Canon FD 50mm f/1,8 adapted to my DSLM. The photo is a total mess, I just didn’t pay attention at this moment. I focused on the clock at an aperture of 1,8 and then I forgot to go back to 9.0 or 11.0 for the shot. It also appears that I messed with the exposure time, dialing it to 1/16000.

The photo now serves as a training subject for me for correcting a heavy vignette in RawTherapee. Not that I never corrected a vignette in RT. But this photo gives me a hard time. I’d love to see how you do it. Of course I also like to see corrections done with other software than RT.

In addition to the RAW in question I upload another similar one I took with an aperture of either 9 or 11 as a reference for the vignette correction I want to achieve.

The files are licensed Creative Commons, By-Attribution, Share-Alike.

R6_06283.CR3 (9.8 MB)
R6_06278.CR3 (14.5 MB)

my proposal: in addition to the standard vignette correction of RT, I darkened the center of the sky a little bit:


R6_06283.CR3.jpg.out.pp3 (19,1 KB)

2 Likes

Thank you, I will take a look at your .pp3 file.

A quick attempt using Adobe’s lcp file, I will try to make something without it.


R6_06283_rt_01.jpg.out.pp3 (19.5 KB)

1 Like

No lcp and slightly darker:


R6_06283_rt_02.jpg.out.pp3 (19.5 KB)

2 Likes

R6_06283.jpg.out.pp3 (14,3 KB)

1 Like

R6_06283.CR3.xmp (11.5 KB)

Every single development looks good. Play Raw is wonderful, one can learn so much.

1 Like

The easiest (when it’s a major brand/common lens) is to grab the lens profile (possibly lensfun, check Adobe DNG converter, it’s a free software with Adobe’s lens profiles too). and just load it in “profiled lens correction”, it will automatically apply corrections based on lens model/aperture/sensor size. Make sure the profile you use corresponds to your sensor size, as your lens is 35mm, so the vignette a crop sensor will have is much much less extreme than a full frame sensor, and adapting lenses to cameras like Nikon DSLR’s, where you have to use an additional lens element to correct the deeper flange distance, can increase your vignetting. i’m not sure if this applies to FD->EF lenses.

If you aren’t seeing results you like from these generic lens profiles, you can create your own “flat field” image of your specific camera and lens (make sure to keep your UV filter on when doing this, as you want it as accurate as possible).

I usually take an afternoon, once a year or so, and create fresh flat field images for all the cameras I’m currently using, and the lenses I typically use wide open (so portrait lenses and some of my f/2.8 wide lenses like 24mm 2.8, 28mm 2.8, etc) I take a couple sheets of good quality printer paper, and tape them to one of my older IPS monitors (the thicker, and heavier a high-end monitor is, the higher the chance the backlight is consistent across the entire monitor), create a completely white .jpg in MSpaint, set it as the desktop image, and hide all shortcuts/taskbar. and just to make sure the sheet of paper is the same luminance, I use a calibrator (if it’s handy), or a luminance app on my phone to spot check. I’ll also take 2 small desk lamps with 5w LED bulbs, and shine them onto the paper on either side of the camera, but that’s not necessary unless our camera has huge, deep well pixels. Because you want to expose one shutter speed adjustment below highlight clipping. Then grab the camera/lens, set everything including focus to manual, base ISO, try to only put the inner-2/3 of the paper in frame, Make the entire paper as far out of focus as possible (focus to infinity, then pull focus 20 or 30 degrees) and start trying to get your RGB highlight image review to start blinking. Then speed up your shutter speed until you get no clipping (it should always clip from the center out, if you’re getting edges, or irregular shapes clipping, adjust your lighting, or just give up on using the monitor like it’s a backlight for X-rays and just illuminate the sheet of paper with the monitor turned off. Once you’ve got f/1.8 captured, you should be able to match the new aperture with the corresponding increase in shutter speed. But don’t rely on “knowing the triangle”, run it up to clipping, then dial back until you get no blinkies in the image review.

Last couple things… Make sure you are taking unadulterated/molested image files. the correct setting is Raw file, lossless/uncompressed/highest bit depth available. Make sure you have Vignette correction off as well. In “color management” select the most tame option available. For my Nikon D810 I have “flat” color management, which turns off virtually all corrections and adjustments. Obviously a raw file shouldn’t have any permanent corrections, but if you’re using the in-camera clipping warning, that system is using the .jpg your camera created from the raw file. I know most DSLR’s don’t offer electronic shutters, if yours does, make sure you select mechanical shutter. . and then yeah, grab all the available apertures, rename each file in this format BRAND_MODEL_FOCALLENGTH_MAXAPERTURE_LENSMODEL_FLAT_FRAME.RAW(orNEForCR3orwhatever. So my D810 with the 135mm f2 ais is:
Nikon_D810_135mm_F2_ais_flat_frame.NEF

Then go into Rawtherapee preferences, and set the Flat Field folder to whatever the folder you stuffed those image files in. and you can actually select “auto selection” in your default profiles you make, so if your metadata shows a 50mm running wide open, or some fast wide angle zoom lens (they all have ridiculous vignetting wide open), rawtherapee will just automatically apply your own personal corrections for that camera/lens. Same goes for Dark frames, but ehh… theres’s so many variables beyond just ISO/Shutter speed when it comes to latent noise in an image (sensor/DSP temps being the most obvious), that I usually let RT auto select a darkframe, then deselect auto, and choose a couple at random from nearby configs, and see if the image is better or worse. There have definitely been times where all the dark frames I tried, made the noise more exaggerated, or distinct, etc. But Flat field images are much more predictable, as it should basically show your lens performance at a specific image diameter (sensor size) at each aperture. As long as the luminance across the frame is pretty consistent, you will never have to correct vignetting manually again.

Thank you, this is valuable information.

1 Like