Tilt-Shift Dark Edge Issue? Read On...
Since last month, I`ve been using my new Canon TS-E 24mm f/3.5 L II (tilt-shift lens) for most of my photography and I can only say; it is a truly remarkable lens and the finest lens I have ever owned, in fact, amongst the finest lenses ever made by Canon. That said, this post is not a lens review; it is about the limitations of most modern DSLRs designs when using a wide-angle tilt-shift lens shifted up more than 10mm. Note; in this case, a Canon 5D Mark II, 5D and a 1Ds Mark II was used for testing and whether or not other digital camera brands have more suitable mirror box design (those that have mirrors) is not concluded here. I would also like to include the notion that this is about the Canon TS-E 24mm f/3.5 L II (introduced in 2009) that is capable of 12mm shift, and not the first version (introduced in 1991), which has a shift capacity of 10mm.
On one of my recent mountain trips, I was using shift to make three exposures that would later be stitched together into a very high resolution photograph. When done correctly, one can stitch multiple exposures together seamlessly, just like in a well-crafted panorama. I did this by placing the camera in vertical orientation and used shift (up 12mm/left side of photograph, no shift for center and down 12mm/right side) that resulted in an approx. 4:3 format (my preferred format) when stitched. Although composing demands careful attention and is more time consuming than when seeing the whole composition directly in the viewfinder, it is easy to do and can be done quite fluidly with practice and visualization. After I had processed all three exposures equally in the RAW-converter and converted them to 16-bit TIFF-files, I first tried to manually stitch them together (which proved to be a bad idea due to parallax) by using three separate layers at different opacity in order to "sandwich" common points (rocks, mountain edges etc.) for correct and seamless result before merging them into one, single layer. It was during this process that I happened to discover a dark edge on one of the images! *when using panorama/stitching software like PTGui, Photomerge, Microsft ICE etc., any dark edge issues will automatically be blended out and will not affect the final result. So, don`t worry.
Ranten Peak (1419m./4655ft.), Norefjell mountain, Norway.
Confused, I went through many of the other RAW-files to see if there were more with this dark edge and I noticed that in all exposures where I had used 12mm upwards shift, which is max. shift on this particular tilt-shift lens, there was a dark edge running along the side which happened to be nearest the center of the image circle! I had used no filters or filter holders (not even a polarizer), but even if I had, any physical vignetting would then appear on the opposite side of the image; the side nearest edge of the image circle. After going through all the raw-files from the latest mountain trip I found that when shifted down 12mm, or when using no shift at all, all images were fine and I therefore ruled out any sensor anomalies. The images where I had shifted 10mm up were also fine. The disturbing dark edge only appeared in the images where I had shifted up 12mm. To make a long story short, I still hadn`t gotten anywhere after having made extensive searches on the internet about this issue (with no results), testing with my back-up camera, a Canon 5D (also full frame), returned my lens and gotten a new copy (which also showed the same disturbing results), and having sent some emails to photographers that I know are experienced tilt-shift lens users, as well as great photographers, hoping that they would provide some answers to what causes this dark edge to appear. This turned out to be an obscure issue that some had never noticed while some never used more than 10mm shift. Darwin Wigget (www.darwinwiggett.com) was kind enough to do his own test with a 1Ds Mark III/same type of lens, and he did also find the same dark edge appearing in his results, when using 12mm upwards shift. That, at least, proved that there was nothing uniquely wrong with my own two copies of the lens, but it did not stop my fingers from further scratching my head. A day, and a lot of scratching, later...another email landed in my in-box, this time from Tim Parkin (On Landscape: www.onlandscape.co.uk). And, it all became clear, as a mountain stream, to me!
Tilt-shift mirror box occlusion
Tim Parkin explained that the dark edge issue was caused by mirror box occlusion in most modern DSLRs (not with mirror-less or film cameras) and happens when a lens is shifted so much that the light coming in through the lens is occluded by the edge of the mirror box. He further explained that the longer focal length, the less amount of occlusion and the reason wide-angles are affected the most is because the lens element is so close to the lens flange and also that the effect is least noticable when lens is fully stopped down. I was so relieved to hear that nothing was wrong with my equipment. Now, if you have just started using the Canon TS-E 24mm f/3.5 L II lens at 12mm upwards shift and is asking yourself: "What IS that dark edge doing there??", these brilliant illlustrations made by Tim Parkin "says a thousand words".
Tilt-shift mirror box occlusion wide aperture
There you have it. Nothing is wrong with the lens; it is an issue caused by modern DSLR mirror box design when using a tilt-shift lens at 12mm upward shift, or in situations mentioned above. In practice, this issue should not be a problem as long as you don`t shift (up) more than 10mm with a wide-angle tilt-shift lens on a DSLR. If you for some reason need to use max. shift up (11-12mm), account for this issue or make at least two exposures with different shift amount and stitch them to overlap the dark edge. Let me also remind you that using software like PTGui, Microsoft ICE or the Photomerge in Photoshop, any mirror box occlusion effect will automatically be blended out. Due to the fact that this issue is not related to the remarkable image quality or functionality of this particular lens, I would highly recommend the Canon TS-E 24mm f/3.5 L II to anyone who is planning to buy a tilt-shift lens with optical performance that is amongst the finest in the whole Canon lens selection.
*oh, I almost forgot; in-camera settings like contrast and saturation, even white balance and color mode, affects the histogram readings. The histogram is based on JPEG-representation EVEN if you shoot RAW! That should explain why I use these in-camera settings. Read this post from my old blog: Seven Swords (2nd paragraph), for more info about this.
After notes: thank you so much, Darwin Wiggett for doing his own tests, QT Luong, Rainer Mirau for the feedback on this issue, and a special thanks goes to Tim Parkin for the technical enlightenment and the provided illustrations! Do check out their astonishing work:
and thank you for visiting my blog and for reading this particular post.
Well, the intention was certainly not to scare anyone from buying this lens (24mm); it is an extraordinary lens in many ways, both regarding image quality and practicality, and I have never regretted purchasing this lens. All lenses have their (often several) flaws, but this particular issue is not directly an issue of the lens itself, rather the way DSLRs are designed and built. I can still testify after almost one year of use, that this is the best lens I`ve ever owned.
Thanks a lot for sharing your thoughts, Wesley, and may you have a wonderful winter time!
Thanks for this post. I actually found it via Darwin Wiggett's tilt-shift eBook. I read this before I test-drove the Canon 24 mm TS-E II and am ver glad I did. These lenses are incredible, but the workflow is complex. I <a href="http://www.wesleypicotte.com/blog/ebook-review-the-tilt-shift-lens-advantage-by-oopoomoo?utm_source=Oopoomoo%20Review&utm_medium=Blog%20Comment&utm_content=Oopoomoo%20Review&utm_campaign=Commenting" target="_new">reviewed The Tilt-Shift Advantage eBook</a> if anyone's interested in checking it out. Darwin and his partner/co-author Sam are great photographers are did a fantastic job on the eBook.
Thanks again for the post - I have the same setup (5DII) and your explanation is really helpful. My understanding is that this is not as much an issue with the 45mm or 90mm, for what it's worth.
Thank you very much, Roger, for your feedback!
I knew there were more photographers, like me, scratching their heads due to this particular issue!
As I said, it is not a big deal, but it IS nice to know what causes it. For a while, it felt like just having bought a brand new, expensive car while at the same time have the strong feeling of something being wrong without being able to put my finger on it.
I am very happy to hear that this post eliminated further scratching your head:-)
And yes, it is a fantastic lens!
Shifting in portrait orientation (often with added tilt) for stitching is my most common use of the 24 MKII. I have also noticed the dark edge and scratched my head wondering why... Never had the time to investigate further, but tried two different lenses and both the 5D mkII and 1Ds, so I was confident it was not the lens. You have enlightened me, no reason to think more about it :-) As you mention, never been a problem when stitching using photomerge.
No comments posted.
Recent PostsFeatured Fine Art Print for April: Hellestø 9 Featured Fine Art Print for March: Hellestø 3 Photography Is Not Just Photography Dørålen in Rondane Nasjonalpark: Awaken in Monochrome 10 Favorite Photographs from 2012 Featured Fine Art Print for December: Sastrugi VI Monochrome Rondvasshøgdi Revealed Monochrome Featured Fine Art Print for November: Autumn Lily From My Archives 10: Trees By The Sea Michael Kenna`s Hokkaido