Working with Tilt/Shift Lenses (2022) | Hacker News Reader

Working with Tilt/Shift Lenses (2022)

phillipreeve.net

89 pointsluu46 comments

Well, that's a pleasant surprise, didn't expect to see this to show up on HN. I really love the photography style they feature, and it's often an inspiration for my own shooting.

My opinionated take: if you're going for architecture or ultrawide, just buy a 9mm lens and crop. My Laowa 9mm is my favorite lens by far. If you want AF, there's the new 10mm that recently came out.

Why not T/S? Correcting vignette is time intensive (unless the lens has an EXIF chip, but I don't have experience with that); image quality is not as good (I found the Laowa 15mm shift lens to just be straight up worse even when accounting for different focal length); they're heavier and take more fiddling.

Why not perspective-correct a 12mm? I found that it's impossible to compose shots well (corners end up cut off, composing the shot is difficult), and there were situations where 12mm just wasn't wide enough.

One of my favorite contemporary photographers, David Schalliol, has an “isolated buildings” series, and he uses a tilt shift lens to achieve parallel lines for the buildings, to striking effect.

http://davidschalliol.com/photography

Many (30ish) years ago, I subscribed (or my parents did on my behalf, I was 9) to Model Railroader magazine. A significant portion of that hobby seemed to be focused on this sort of photography and the construction or use of these lenses.

I haven’t thought about it much since then (or much at the time, since I was a lot less interested in the detailed modeling and a lot more interested in the control systems and coming up with complex track layouts), but the pictures those guys took were always pretty cool.

Tilt-Shift lenses are great and I own one but ... just like many things in photography now-a-days, computation can do much of this.

No, it's not perfect just like an iPhone doesn't take as a good a picture as a high-end Canon/Nikon/Sony. But, AFAIK, all modern phones have perspective correction built in to their photo editor controls. I use them quite often to do the same things (in relation to perspective) as seen in the article. Again, they aren't as nice but 99.9% of the time the images I personally take are only going to be seen on a small screen. And even on a large screen (my 65" TV) they still look great at a glance.

There's two features I wish were more common.

1. Changing the size of the image. Say you take a 1000x1500 image, in adjusting the perspective to make the verticals not converge you'd really need the image to be 1000x2000. No app I know if will do this. As you adjust the perspective there's no option to increase the size of the image to fit the result so stuff just gets cropped. Ideally it would let me opt into auto-increasing the size of the image to fit the plane of the original after its tilted. (note: I get I could do this in photoshop by pulling the image into a larger document. I'm just annoyed I have to go that far and wish I could do it in the phone in the default photo editor as I just the perspective.)

2. Let me change the perspective live while I'm taking the picture. IIRC I found apps that do this but they want $10-$20 a month. Here's hoping Apple and Google add it to the default camera apps. This could be even better because given computational photography the phone can take 10-20 shots at 1000fps and then merge them for higher res in the areas where there's less censor pixels per arc angle.

Without compositing images, or small aperture with artificial blur, computation can’t help you with the tilt effect like the one of the store sign reading Arche, can it? The geometry of that shot is interesting to me.

I like the compositing idea though. Could that get us 50 % of the effect with 5 % of the bulk and cost?

> Tilt-Shift lenses are great and I own one but ... just like many things in photography now-a-days, computation can do much of this.

No it can't. You can not reliably tilt the plane of focus, something that precisely the tilt-shift lens does.

> No, it's not perfect just like an iPhone doesn't take as a good a picture as a high-end Canon/Nikon/Sony.

iPhone actually does very middle of the road pictures. You can never beat a Canon/Nikon/Sony simply because you do not have the sensor size. I am not even going to try to bring metering discussion here.

> And even on a large screen (my 65" TV) they still look great at a glance.

Depends on the viewing distance. Put ink on paper, and then we will talk.

> Let me change the perspective live while I'm taking the picture. IIRC I found apps that do this but they want $10-$20 a month. Here's hoping Apple and Google add it to the default camera apps. This could be even better because given computational photography the phone can take 10-20 shots at 1000fps and then merge them for higher res in the areas where there's less censor pixels per arc angle.

No, you are wrong. When you are shifting, you actually move over the "sensor" in the focus circle, not merely do a cropping gizmo like apps do.

If you want to really understand why view cameras still whip ass and tilt shift lenses are relevant, try taking photograph of a tall building and see how the edges are not parallel and if you try to correct that in software, you will see distortions.

I'm sorry but for 90% of picture consumption today, i.e. on phones, the original commenter is right. iPhone photos are good enough to look at on your phone. I'd guess that something like 90% of photos today are a) viewed on a phone and b) looked at for less than three seconds each.

You don’t get a TS lens because you’re trying to make a discardable commodity image on a discardable commodity phone… yes, 90% of the images humanity takes aren’t worth viewing, much less keeping, but that doesn’t mean iPhones are “good enough” for images that are important, and in the coming decades we’ll really regret all of our visual legacy having been recorded with lossy compression through crap lenses on ludicrously small chips.

(context: high end tilt-shift lenses are used for architecture photos to make buildings and complexes look all square everywhere and not tapered skywards due to perspective effect. The stakes are higher and spending more effort than snapping and editing on an iPhone is justified)

You can’t do the ‘tilt’ part computationally (except by going all-in and computing a depth map), but simple perspective corrections do a good job of emulating ‘shift’. Shift is typically used for subjects at infinity (such as buildings) where small changes to the plane of focus are irrelevant.

Computational shift is a quick and dirty fix for a random picture with the – auspiciously, should I say – «misaligned» perspective that can be salvaged without the detriment to the composition.

Oftentimes that is not the case, though, and results in a crop anywhere between 25 and 40% completely ruining the composition.

Both have their merits, and I occasionally use the computational shift to fix the perspective in my phone camera snapshots with a varying degree of success, but if the circumstances are conducive, I will bring and use my full frame DSLR with the trusted manual TS-E 24mm lens. It is much more fun that way, too.

Tilt, however, is a completely different cattle of fish…

I know the author poo-poos on it, but I thought the miniature effect was cool and I'd buy a tilt-shift lens for my old medium-format camera solely for that purpose. I've never seen a phone pull off an effect like that, and even if it could, sometimes it's just more fun to do it the hard way.

It's very simple to create a similar effect in post.

For example blur the image where the blur radius is high at the top and bottom of the image and zero at the center.

Of course this does not take depth into account but simulated the effect well.

There are a lot of tutorials about this subject.

> A “shift panorama” is a very simple panorama as you don’t need any accessories like a rotating plate and you won’t be running into stitching errors

I have a panorama head (Manfrotto 303SPH). It's enormous, weights around 3kg and... I never use it.

In my experience, Lightroom does an extremely good job at stitching panoramas from handheld shots, as long as there is reasonable overlap between shots. Seems to work perfectly every time.

Googled that head and I really don't get the point. Just get an L bracket and be done with it, surely.

It depends. For some applications, sure.

But especially when you're doing multi-row panoramas with long exposure times, preparation is key, and so is execution time.

I often do multi-row HDR panos. With brackets of 3 exposures each, the time for a single frame can quickly add up to 45s or more. In a 3 row, 5 column pano that adds up to a total exposure time of 11.25 minutes - net. This doesn't include the time between frames needed to pan to the next column (or worse, next row and 1st column), making sure you get enough overlap, align everything and finally tighten the tripod head knobs/clamps again.

Depending on the season, the usable blue hour may be as short as 15 minutes. That mean's you essentially get one shot at this.

So any gear that allows you to do that reliably and quickly, for example a panoramic rotator head with indexed degree stops [1], helps immensly with getting good results.

This is an example of a 2x5 "pano" with a 3x exposure bracket (20s, 6s, 2s) I took:

https://commons.wikimedia.org/wiki/File:Federal_Palace_of_Sw... (Warning: Full res image is huge)

With 20s as the longest exposure, the chance of a car or bus driving right through my shot was huge, so I had to retake many of the frames, some even several times.

[1] For example the Nodal Ninja RD16-II Advanced Panoramic Rotator

My comment was not intended as a dig against that head. It's fantastic at what it does, and does it perfectly well. It's especially useful for zoomable images: take hundreds of shots with a tele lens and assemble them in one picture that one can navigate ala Google Maps. For that kind of image, handheld won't work because one has to make sure not to miss a spot.

But for simple panoramas, handheld + Lightroom is fine.

I agree.

A 30% overlap allows be to basically never have an automatic pano merge fail in LR. Though I rarely shoot panos full handheld, usually still from a tripod with at least a ballhead.

Even better, LR allows you to merge a HDR pano in a single action - which has become an important part of my workflow, because it works so well, and results in a nearly RAW quality DNG that can be edited non-destructively.

With one exception: Multi-row panos with 3 rows, where the top row is mostly sky. Even with lots of overlap, LR usually can't figure this out. But the workflow using tools like Hugin or PTGui is so involved (and requires baking the RAW files first) that I usually just avoid this situation. Besides the fact that it's often not the most interesting composition.

45TSE is probably one of my top 3 fav portrait lenses, the glass is really something special on that lens, un-tilted the clarity is just phenomenal. Not sure what it is, but I love shooting with it, and it's fun to tilt in portraiture! :) https://s.h4x.club/DOuJJKJp

The miniature effect you can achieve with tilt/shift lenses is very cool, but I've always wondered: why does the result look miniature? Is it that it has the same visual problems that a cheap-ish lens focused on a miniature would have (e.g. very shallow depth of field)?

(My brain wants to know what my eyes seem to intuit...)

I think it’s to do with focal length relative to the object - the length of the camera and size of the aperture relative to the size of the subject. Less about the cost of the lens, and more about its diameter and the camera’s length, I think? Macro lenses are often long and skinny (but yes, expensive)

The tilt lens gives you a plane of focus that is not parallel to the lens, which can mimic the narrow depth of field of a truly enormous camera, or (more convincingly, a tiny subject)

The “shift” part is a separate movement axis (parallel) and corrects perspective, giving straight lines instead of a vanishing point.

But I have no real idea what I’m talking about, so I hope some optics expert chimes in and explains it — maybe even with a diagram if we’re lucky. Optics are neat.

I first did tilt-shift photography as a teenager before I even knew what it was called, when I inherited a crown graphic from my grandfather, and I noted that I could change the plane of the lens, and could see on the ground glass viewfinder what the effect was. Had great fun running around Italy with a backpack full of Polaroids and a far more precious supply of 4x5 colour positives, taking architectural photographs. Had less fun developing them all at home later - the colour positive process is not easy.

I’ve since tried tilt shift with my DSLR but it just ain’t the same - it gives good enough results, but not much that you can’t do in post - whereas looking at some of the prints from the crown graphic, the quality and the… I don’t know, quality… of the images is different - better.

I guess my view is that it’s worthwhile with wet photography, but not with digital.

A digital adapter back (to something like EF or RF mount) for a 4x5 only runs a few hundred dollars. Best of both worlds.

EF is not really practical, the depth of the mount limit too much the movements you can make.

Maybe it's viable with a high f number as the problem of the body shadowing the sensor will be less.

Quite surprised to see something of this genre discussed here without any mention of the obvious benefits in industrial process.

By normalizing the spatial representation of the captured image, it is possible to perform quantitative analysis without any additional processing overhead or associated temporal delay.

Presumably sharp focus and a lack of distortion would grant similar benefits when training machine learning for visual applications.

Source: Own multiple books on the early era of photography when tilt-shift was a normal feature ("view cameras") and have designed a fair number of autonomous optical systems. The best book I have, for the benefit of others, is View Camera Technique (7th Edition).

A stellar article by Bastian Kratzke as usual.

Using tilt to increase depth of field is incredibly useful in macro photography. For example this picture of sushi by Ming Thein [1] using the Nikkor PC-E 85mm lens. It's particularly useful when the subject is mostly flat, like a watch.

I personally have a Laowa 15mm f/4.5 Shift lens, with which I took a picture of the Hallidie Building [2].

[1] https://www.flickr.com/photos/mingthein/6912586530/

[2] https://en.wikipedia.org/wiki/File:Hallidie_Building_in_San_...

A pro photographer friend of mine stopped using Tilt/Shift lenses for architecture, he says the image quality is noticeably better if you correct the perspective in photoshop/lightroom afterwards.

Depends. Extremely good lenses like the Fujifilm GF 30mm f/5.6 will definitely outperform correcting perspective in Photoshop. Of course, that costs many thousands of dollars.

But comparing older Canon TS-E 24mm f/3.5, or the Samyang/Rokinon 24mm f/3.5 against any decent wide angle lens nowadays is a wash.

> But comparing older Canon TS-E 24mm f/3.5, or the Samyang/Rokinon 24mm f/3.5 against any decent wide angle lens nowadays is a wash.

Edge sharpness and barrel distortion? ;)

Barrel distortion is automatically corrected for any given lens by Lightroom.

There’s no logical reason why a TS lens should be sharper at the edge of its (moveable) image circle than an ultrawide should be at the edge of its fixed image circle. They both offer around the same angle of view, just at different focal lengths.

I used to own the TS-E 24mm II. It was a great lens, and using it was a process that made you slow down and consider things.

But I think your pro photographer friend is right. Modern DSLRs and a really good wide angle (say 9mm or 12mm) will let you correct in post for any shift you want.

Tilt is still not possible even in post, but I found that to be rarely useful at 24mm. For macro it would probably be useful all the time.

Isn't that the other way around, shift being fancy cropping and tilt being perspective correction?

Shift shifts which part of the image you are implicitly cropping to. (Every photo is a crop: you’re taking a circular image plane and cropping a rectangle out of it.). Tilt tilts the focal plane.

I occasionally wish someone would make a camera with a circular sensor that captures the entire usable field of view from the lens. If nothing else, a professional camera like this could skip the extra side grip and save some bulk because there would be no need to ever hold the camera sideways.

> I occasionally wish someone would make a camera with a circular sensor that captures the entire usable field of view from the lens.

I've had the same thought, though I wanted just a 1:1 square camera sensor. It could perhaps be "oversquare", so that the corners are slightly rounded off. That wouldn't be so wasteful, since a sensor needs some shaded pixels to measure noise anyway.

I was just watching a YouTube channel that Tilt Shifts video games. Here's one for Elden Ring:

https://youtu.be/dZ9RU7pznTs

A tilt/shift lens is great… but it’s nowhere near as flexible as a field or monorail view camera.

Luckily they tended to be built to last and there’s load of them out there, if you know where to look.

The Canon RF backfocus distance is short enough that you can get a tilt/shift adapter that lets you just chuck a medium format lens on the camera. Pretty slick.

If you're interested, "Little Big World" is a very pleasant YouTube channel who does tiltshift videos of global cities set to classical music.

The miniature effect is just one thing tilt/shift lenses can be used for but isn’t their primary use.

The key thing about all TS-lenses I know of is they only support manual focus; while this may not be a problem for architecture or landscape photography I'm too casual to try to focus a portrait manually (Gregory Heisler talks about doing this in his book "50 portraits" and I love his work).

Rumour has it that Canon's RF-mount tilt-shift lenses will support autofocus though...

It's not that bad if you stop down to f8 or so and understand how the focus aligns, or you can use liveview and zebra stripes

When stopping down you also can use the distance markings on the focus ring effectively (it's easier to do with sufficient depth of field, and with a DSLR you can practice as much as you like). Over time you can do this intuitively, like how a professional focus puller does on a film set.

Another intuitive skill that can be easily practiced on a DSLR is metering by eye. Take a couple test photos to dial in the right exposure for the conditions then use that as a baseline when making adjustments for a reflective subject, areas in shadow, etc. With a modern mirrorless camera and EVF this skill may not be that necessary though.

I am generally pretty fast with lenses that were designed to do manual focus, like Zeiss Otus and Milvus lenses for Canon (For example). Lenses that come with autofocus does not have a long enough throw to be able to perform manual focus quickly and accurately, but it is definitely doable.

Just watched Poor Things[0], up for 11 oscars including best pic.

Anyway, they used some crazy fisheye lens effects, and also some tilt shift.

[0] : https://www.rottentomatoes.com/m/poor_things

Reply on news.ycombinator.com