Artifacts from RF15-35mm F2.8 L IS USM

[JoshZielinski]

Hello,

For a while now, I've had bad artifacts from my RF15-35mm F2.8 L IS USM when photographing scenes in which light meets a sharp edge beyond which the scene is dark. A recent, particularly severe example is attached from underneath a recess cave. I overexposed the image so you can see how the artifact obscures the texture of the sandstone nearest to the edge. I've encountered this effect on many other similar scenes. It frequently renders the images essentially unusable due to the amount of work it would take make the exposure appear natural.

As far as I can tell, the lens is clean on the front and back, so I'm not sure what would be causing this. Any thoughts?

[rs-eos]

Scenes such as this are not really capturable in a single photo due to the very high dynamic range. If you have a tripod, you could take at least two, maybe three images at different exposures (expose for cave, expose for sky, maybe expose in between). Then blend the images.

Same thing will happen when shooting a portrait against a white background where too much light is added to the background. Details around the edges of the person (e.g. hair) will lose detail and/or be completely blown out.

[JoshZielinski]

That's correct. This image is the final in a bracketed series of five. One balanced exposure, two underexposed, and two overexposed. The other exposures exhibit the same artifact. I featured this exposure since it is most visible. When I merge these photos to make an HDR composite and lighten the shadows to recover detail in the rockface, the artifact becomes obvious.

This overexposed image would be one of two photos used specifically for the rockface in such a composite, so to have the artifact obvious here presents a problem. If I only exposed for the sky and left the rockface as a silhouette, the artifact would not be visible, but that's not the goal. That's why I'm wondering what else might be the problem, or perhaps this is just an innate limitation of the lens.

[SignifDigits]

First, it would be helpful to know shutter speed, f-stops and ISO.  Is that the sun in the upper right corner?  It appears that there is radial prismatic diffraction from that spot in the shot from the aperture blades.  If you are shooting at or about f22 you might want to dial that back and increase shutter speed.  I have certainly experienced diffraction and loss of sharpness due to diffraction at f22 with my version of that lens, but I consider that normal behavior for any lens.

Sorry if I'm missing the point, but I'm not quite 100% certain what I'm looking at and what I'm supposed to be looking for.  To me it looks like a shot with the sun in the upper right hand corner stopped down too much and not behaving atypically, but without more data it's impossible to say.

[JoshZielinski]

I should've thought to add the exposure info. It is:
30 seconds
f/16
ISO 100.

The sun is behind a layer of clouds, but I believe it was behind me, so not visible in this frame. I circled the particular area I was referencing. Rather than a more natural grade from light to dark along the rockface, there are streaks and circles of brightness, which I figure is from the lens.

 

I will try to find another example or see if there's a different way to show it. It's very possible the lens is acting as normal. I just have found the affect very disruptive on quite a few occasions, so I wondered if it wasn't normal.

[johnrmoyer]

I did not notice what software you are using, but metadata seems to say "Adobe Photoshop Lightroom Classic 15.1.1 (Windows)". Below are my guesses, but I hope they might be helpful anyway.

I might guess that the artifacts come from either lack of hood or from lens correction in software. If there is no hood, then light can enter the lens at an angle and reflect. If the hood is optimized for the narrower focal length and not the wider, then there are more likely to be reflections at the wider angle.

Are you doing the HDR composite after lens distortion correction? or before?

If your camera can save DPRAW files, there is a ghosting correction in the DPRAW tools of Canon DPP software and this might be worth trying on each of the raw images before compositing. The 16 bit TIFF files saved from DPP work reasonably well in other software.

I do not have that lens, but use the less expensive EF16-35mm f/4L IS USM. It seems to me that this lens does a little better not wide open and at a little less than the widest angle focal length. This image was not HDR and has less contrast than yours, but I used the "digital lens optimizer" in Canon DPP to sharpen foreground and background details and used the distortion correction and peripheral illumination correction in DPP. I used the DPRAW tool microfocus adjustment to adjust the focus zero front/back with a strength of 8 which seemed to help the edges. https://www.rsok.com/~jrm/2024Feb09_SaltPlainsNWR/2024feb09_reflections_IMG_8919c.html 

Reflections in marsh at Salt Plains National Wildlife Refuge in Alfalfa County, Oklahoma, United States on February 9, 2024 ; EOS R5 ; F Number 4.5 ; ISO 100 ; Shutter speed 1/400 ; Focal Length 20mm ; EF16-35mm f/4L IS USM

100% crop showing contrast (unsharp masking appears excessive at 100%)

100% crop in DPP to show contrast at edges

[rs-eos]

Double-checking your shutter of 30 seconds. Were you using an ND filter? Or was this supposed to be 1/30 seconds?

[SignifDigits]

Good point on the hood John.  I hadn't thought about that.  To me it appears radial which is consistent with aperture blade "stars" of diffraction rather than a "glass element" concern.  But it could be one or both I suppose.  It's snowing out so I can't go out and take a pic at f16 to try to reproduce the effect and do a hood on and off comparison.

Did you say what Camera?

What was your zoom - looks like 35mm to me, but that's just a guess.  IS on or off?  AF or MF?  If AF settings?  Tripod or hand-held?  At 1/30 you IS and IBIS systems might be at work.  Do you see the same thing at faster shutter speeds?  Any and all information you can provide could be useful.  Thanks.

[johnrmoyer]

At F/16, there will be a lot of small aperture diffraction blur, spreading across several pixels with your camera. At F/7.1 the small aperture diffraction blur is not likely to be noticed. The "digital lens optimizer" in Canon DPP or the "capture sharpening" in Rawtherapee free software will do a Richardson/Lucy deconvolution to remove some of the diffraction blur. At ISO 100, this works very well because there is little noise.

If you are using the widest angle, then a hood intended for a fixed focal length of 15 mm might fix it if you can find one that will fit.

https://en.wikipedia.org/wiki/Richardson%E2%80%93Lucy_deconvolution 

https://snapshot.asia.canon/en/article/lens-faq-4-what-is-ghosting-and-flaring 

https://www.cambridgeincolour.com/tutorials/diffraction-photography.htm 

https://snapshot.asia.canon/en/article/eos-5d-mark-iv-tips-3-handy-new-dpp-features-for-correcting-d... 

[JoshZielinski]

Yep, it's 30 seconds. This was taken shortly before sunset. I had no ND filter on the lens.