Which focal length gives best blur of backgorund detail

[wideopeniris]

<p>I did some sums, and I also tried out a few lenses and tried to work out the effect of focal length on background blur. My conclusions seem to be at odds with the conventional wisdom so I wanted to check that I havent got it wrong.<br>

So most people say that longer lenses are better for selective focus and blurring of background detail. I think this is wrong _IF_ you keep the subject size the same in the frame at all focal lengths and for the same Fstop. This means that you stand farther from the subject on a longer lens.<br>

When I tried this with a 50mm and 200mm lens there was (almost) no difference in the blur of an object a little way behind the subject.<br>

I have given a lot more detail <a href="http://wideopeniris.blogspot.com/2011/01/which-lens-blurs-background-best.html">here</a> where my tone is more definite - of course I could be completely off the mark and am happy to ammend my thinking if its seems that way!<br>

I hope this isnt too controversial (ducks..)</p>

<p>Kevin.</p>

<p> </p>

[Matt Laur]

<blockquote>

<p>I hope this isnt too controversial</p>

</blockquote>

<p>Nah, it just involves lens design and behavior, depth of field, and may degrade into a discussion of qualitative bokeh issues. It couldn't possibly stir up any trouble. :-)</p>

[Andrew Garrard]

<p>Kevin: You're correct, depth of field, and blur on a background object <i>that's only slightly out of focus</i>, change very little if you keep the same (relative) aperture and subject framing and increase the distance between lens and subject. However, the farther the background is from the subject, the more the magnification of that background will change with the focal length of the lens.<br />

<br />

For example, take a photo of a subject with a 14mm lens at f/2.8, from 1m away. Get the same framing with a 135mm lens at f/2.8 (which requires you to be about 9.6m away). On a background 5m from the subject, the circle of confusion (projection of the lens past the subject) is 25mm in both cases - but the view of these circles is four times bigger through the 135mm lens, so the background is "four times as blurry".<br />

<br />

See my doodle <a href="http://www.zen60163.zen.co.uk/Private/Bokeh.png">here</a>. There's a spreadsheet <a href="http://www.zen60163.zen.co.uk/Private/DepthOfField.xls">here</a> from when I was working this out for my own satisfaction, if you're interested.</p>

[Andrew Garrard]

<p>Follow-up, having read your blog entry. It's true that relative aperture tends to have more of an effect than focal length, and that format (with a focal length change and the same relative aperture) also has an effect (I stirred up trouble by talking about "35mm equivalent aperture and ISO" recently, so I'm staying well out of it this time). Nonetheless, all else being equal, focal length really does matter - and how much depends on the relationship between the lens, subject and background. Correspondingly, objects in front of the focal plane will be less blurred by a longer lens. If you're not seeing it in your experiment, you need to try more experiments with some variables changed. I'd feel awfully silly to have bought a 200mm f/2 if I could get the background to go away in the same way with my 50mm f/1.8.<br />

<br />

You can reasonably say that, under some circumstances, the difference in the background isn't as much as you might have been led to believe, though. I hope that helps.</p>

[wideopeniris]

<p>I guess its a matter of subject. I'm referring to portraits, and mine are often head and torso or head and shoulders. My subject distance is usually about 3m. On small cameras the magnification is definately small. On LF its getting a bit bigger, but still of the order of 1:10. In your 14mm example, magnifaction is only 1:2 and I see your argument completely (havent had a chance yet to look at your spreasdsheet).<br>

The 200 F2 - you paid for aperature mostly at a focal length that gave you the right magnification. No need to feel silly! (I might manage envy except that I have a 420 f3.5 to one day mount on my 5x4 that cost nearly nothing...dont mention the 915 f6...)<br>

However, I suspect you will find my experiment works as I said at least for things near the subject. Now you mewntion the change of magnification of the OOF object, I need to go and try a more elaborate experiment..I'll try to press the shutter and post some pictures this time!<br>

In my raytrace, I adjusted the field angle to maintain the magnification and may have inadvertaintly cancelled out the effect you mention. Thats always the risk with fancy software!<br>

Some of my portraits are in my portfolio here, taken on 5x4 at f2.5 or f4 and you can easily see the dramatic effect of format size. If you print these photos at a decent size the effect is eye popping. Theres no F0.7 lenses on offer, so this is the only way to get this effect at any cost.<br>

Whether its a good effect is up to the viewer. I just find it a bit different and interesting, not necessarily better than the usual arrangement.<br>

Of course there are other advantages of a human interaction type associated with an huge and interesting looking camera that take 10 minutes to set up for one shot. I dont have a problem making my subjects relax. Usually the opposite - they get bored after 3 or 4 frames!</p>

<p>Kevin</p>

 

[Andrew Garrard]

<p>Hi Kevin.<br />

<br />

Actually, I got the 200 f/2 because of the LoCA that a 135 f/2 was giving me. But losing the background a bit more helps. Also, I can hand-hold it. :-) I can do a 500mm f/4, but you've got me with the 915mm until I get a Dobsonian telescope - and I'm not getting one of those for the bokeh... I'm interested that you found a 420mm f/3.5 with 5x4 coverage, though - I may some day get a view camera, and I've seen very few large physical apertures.<br />

<br />

To the point: I definitely agree with you that DoF. and things near the subject (relative to the camera), don't change much with lens focal length. In fact, I came up with my doodle and spreadsheet because someone was trying to tell me there was <i>no</i> difference in DoF with focal length - there <i>is</i> a difference, but the effect is usually very small. It's the relative enlargement of different parts of the scene that matters, and if they're all close together, you don't see much change.<br />

<br />

Don't get me started on the effects of format on DoF. I accidentally raised stink in the Nikon forum when trying to talk about "35mm equivalent aperture and ISO" for crop sensors. :-) I'm not actually sure what happened to Kubrick's f/0.7 lens, but there's a reason I wasn't too excited at a 25mm f/0.95 for micro 4/3. Depending on the subject, I like my shallow DoF - although sometimes because I can't control the background of candids (not so 5x4-friendly). Should I take the large format plunge - so far 645 and a borrowed Rolleiflex have been my largest - I'll make a point of playing with shallow DoF. Tilt helps, of course (I do have a 35mm T/S).</p>

[Matt Laur]

<p>The main issue (with juggling those lenses and the change in working distance that comes with change in FL for the same subject size in the frame) is the change in angle of view and perspective. I reach for that longer FL when I want to reduce how much background I'm seeing, not necessarily out of some concern about throwing things more or less out of focus, per se. Changes in useful DoF along those lines are pretty insignificant, but changes in <em>what you see</em> because of the angle of view can make a big difference. And, of course, one lens may render that background in a very different way than another - not quantity of blur but quality.</p>

[sjmurray]

<p>Here's a link to an interesting overview of depth of field.<br>

http://toothwalker.org/optics/dof.htm</p>

[dave_redmann]

<p><em>So most people say that longer lenses are better for selective focus and blurring of background detail. I think this is wrong _IF_ you keep the subject size the same in the frame at all focal lengths and for the same Fstop. This means that you stand farther from the subject on a longer lens.<br /></em><br />Yes, correct. There are clueless people who will state otherwise, but they usually ingnore the fact that, with the longer lens, you have to increase the camera-to-subject distance to keep the size / framing of the main subject the same. (Some of these same clueless people will tell you that they get great background blur with their tiny-sensor compacts at f/2.8--what it 'enough' blur is partially subjective, and the absolute amount of blur depends on factors such as subject-to-background distance, but <em>all else being equal</em>, a compact digital at f/2.8 has about the same depth of field as an "APS-C" DSLR at f/9, or a 4x5 at f/54!)</p>

<p>Of course, when you change focal length and move the camera to reestablish the same framing of the main subject, you affect how much of the background is in the frame--the wider the lens, the more of the background you get in the frame.</p>

<p>Another thing, it gets hard to make really fast (wide f-stop) lenses that are appreciably long or wide-angle. And long lenses with wide maximum f-stops are impractical. The fastest 35mm-style lenses are all 50mm's: the Canon rangefinder 50mm f/0.95, the Leica rangefinder 50mm f/0.95 and 50mm f/1, and the Canon EF 50mm f/1. The fastest 500mm is the Sigma 200-500mm f/2.8--and it weighs about 35 lb (16 kg) and costs $29,000. A 500mm f/2.8 must have a front element at least 7 inches (179mm) in diameter.</p>

<p>And as already stated, for most tastes, there is a preferred perspective. IMOPO, for adult human portraits, you don't want to be closer to the subject that about 6 ft (1.8 m), or father than about 10 ft (3.0 m), <em>unless you are deliberately trying to make the person look different from the way they normally appear to you</em>. At a subject-to-camera distance of, say, 7 ft (2 m), using common, middle-of-the-road standards for print circle of confusion, with a full-frame DSLR or 35mm camera with a 100mm lens, assuming an 8x10-inch (roughly A4) print, if we take the depth of a human face to be about 3 inches (76 mm), then you need to stop down to about f/4 to keep it all in focus. I am not a fan of portraits with sharp eyes but other parts of the face, or really even the head, blurred. But your taste may differ.</p>

[wideopeniris]

<p>Andrew: the 420 3.5 is a projection lens. It is a triplet in the style of a cooke lens. It covers 10x8 (!) with ease, being about 42 degrees at 420 giving an image circle of 300mm diameter. But over this angle its a bit soft at the edges as you might expect. But over 4x5 its plenty sharp enough. Also its coated which means its actually quite contrasty (I have an agfa apotar for 6x6 which is a triplet and at f8 its sharp and by far the most contrasty lenses I own - so dont sniff at triplets..)<br>

Only problem is that its 130mm diameter and the back focal length is large (250mm). It also has no diaphragm. So its going to be a bit of a project to make , in effect, a custom camera with focal plane shutter and modifying the lens to add waterhouse stops or something. At the moment its on the back burner...<br>

The 915 f6 is just crazy. Its a white elephant, though an interesting one. An APO-Tessar type design with evil blade shutter and weighs some 7kg originaly for high altitude aerial photography on 18x18in film.</p>

 

[Andrew Garrard]

<p>Dave: you can't ignore the change of background magnification when discussing the amount by which the background appears blurred as focal length changes - that (for a fixed image format) is the entire reason that "clueless people" claim that longer lenses lose the background better. Claiming a difference in depth of field is another matter, as we've discussed - the small difference between the focal plane and DoF relative to the subject-to-lens distance makes the magnification effect negligible most of the time.<br />

<br/>

Format (and the same field of view) makes a significant difference, but you need to include the magnification of the image when working out by how much - DoF at the actual size of a crop sensor is much smaller than when you produce a print from that sensor which is the same size as a 5x4 slide. I was recently involved in a really long rant about DoF in a "DX lens numerology thread" in the Nikon forum, and I don't really want to revisit it. Mainly, with the lens in the same place, it's the physical aperture (entrance pupil) size that determines DoF by format, and this changes with focal length and relative aperture (f-number).<br />

<br />

I don't believe there are many (designed-for) LF lenses with the physical aperture of the biggest 35mm or medium format lenses. A 200mm f/1.8, 400mm f/2.8 or 600mm f/4 (or 300mm f/2, or 1200mm f/5.6, to include exotica) is pretty good at losing the background no matter what you compare it to. Sure, a 600mm f/1.8 on a 5x4 would lose the background better than a 200mm f/1.8 on 35mm (with the same field of view) - but there aren't many 600mm f/1.8s out there; there aren't even many 600mm f/5.6 lenses which might be comparable. I do use my 200mm f/2 for (candid) portraits; before that, I used a 135mm f/2. There aren't many 400mm f/2 lenses out there for me to fit on a 5x4 either (although I'll admit a 400mm f/5.6 - which does exist - on 5x4 has comparable DoF and field of view to a 135 f/2 on 35mm). Capabilities of camera systems aside, I'll agree that ultra-narrow DoF is not always desirable.<br />

<br />

That said: Kevin, cool piece of kit, although I'm curious about its resolution (if it was designed for projecting large format, maybe - but if we're talking a cinema projector, even IMAX film isn't that big). I'll be interested to know how you get on with it. As for the 915 f/6... okay, that's exotic. I've seen a 1200mm f/6.3 cine lens, and if it's as preposterously huge as that (probably worse, given the coverage) then I'm sure you'll have fun trying to use it. Just leave the lens caps on if it ends up in a yard sale, or you'll burn a hole half way to China. :-)</p>

[Andrew Garrard]

<p>Just to soften my tone a bit... I agree that it ought to be possible to make a 150mm f/2 lens, that would not be huge, with 5x4 coverage - and that 50mm f/0.7 lenses (with similar DoF and FoV) for 35mm are few and far between. However, largeformatphotography.info mentions barely any any f/4 lenses, and the 150mm lenses listed for 5x4 are all f/5.6 or slower; not so challenging to match in 35mm terms. It's true that getting the same narrow DoF out of an ultrawide is a bit tricky on 35mm, though (I know of no f/1.8 15mm to match the f/5.6 47mm super-angulon). Matching large format DoF in 35mm is much easier in long (I'd better not say "telephoto" in the context of a large-format discussion) lenses.<br />

<br />

As someone with limited view camera experience - I want to get into it at some point - the existence of reasonably fast LF lenses is new to me. Kevin: may I ask the origin of your f/2.5 lens, assuming it's a bit less exotic than the other glass we've been discussing?</p>

[Andrew Garrard]

<p>I owe an apology, before any LF enthusiast bites my head off for inaccuracy. I'd been using a 3x "35mm equivalent focal length" factor for 5x4, out of some poorly-remembered approximation. Using the diagonal gives a conversion of roughly 3.76x, so the 35mm equivalent (in DoF terms) of a 400mm f/5.6 5x4 lens is something like a 105mm f/1.5, which I'm not sure exists in the 35mm world (although a 135mm f/1.8 does), not, as I claimed, 135mm f/2. On the other hand, the 5x4 equivalent of a 135mm f/2 would be a 500mm f/7.5, which is faster than I can find listed for that focal length in 5x4 format.<br />

<br />

Just avoiding an accidental bias toward 35mm - it was incompetence and lack of LF expertise, not a deliberate attempt to make 35mm look good.</p>

[wideopeniris]

<p>I actually use a scale factor of 4:1 for 35mm vs 5x4. You almost always crop the edges off a 24x36mm frame when printing in 35mm, and so the short edge defines the ratio - i.e 96mm/24mm - exactly 4:1.<br>

There are three moderately common lenses on 5x4 - The Schneider 150mm f2.8 Xenotar, The Kodak Aero-Ektar 178mm F2.5 (aerial camera lens for 5x5inch) and the Dallmeyer Pentac 8inch (210mm) f2.9.<br>

I cant afford the Schneider. I have an Aero and a pentac, and the pentac is my favourite. The aero is the sharpest wide open and is slightly faster, buts its huge and heavy. Also its suffers from mechanical vignetting at the edge of the field giving 'cats-eyes' in the blur of point lights in the background and giving slightly harsh bokeh. The Pentac is has less vignetting and by f4, where I use both lenses most often, its blur disk is smooth and round right to the edges. The sharpness is perfectly adequate for portraits and its more contrasty than the Aero (but not much - I'm lucky as my pentac is single coated).<br>

In addition there are plenty of tessars at F3.5 and almost any focal length. Also there are a few f2.7 and 2.8 tessars. But the 3.5s are much sharper wide open. The field curvature on the tessars makes the edge of coverage soft. Then finally there are some ektars that are double gausses that have f4.5 down to about 120mm for something wider.There are many 4.5 lenses out there like (hi cost) heliars etc. Mostly speed lenses for press use in more or less 'standard' focal lengths.<br>

The 420 3.5 projection lens is for an epidiascope - the sort of thing used for projecting book pages in a lecture hall. I expect it to be soft at the edges, but a cooke is pretty good in the central 2 thirds.Having projected an image onto a GG I am confident it will at least be interesting. It renders an almost painting like effect with still lifes. I'm inspired now to get it mounted...<br>

Kevin.</p>

 

[Andrew Garrard]

<p>Thanks, Kevin - nice to be learning! I'll reject the "all LF lenses are very slow" theory that I'd reached, in that case. Interesting that eBay prices for the f/2.8 lenses aren't necessarily any higher than the going rate for the slower glass I'd already heard of. Also, interesting to see the 1217mm f/6.3 that came up when I searched eBay for Aero-Ektars - maybe I was misinformed about the other big lens I saw. I'll bear in mind that largeformatphotography.info doesn't, for some reason, seem to mention faster glass when I next refer to it.<br />

<br />

So: I buy that a 210mm f/2.9, or any of the other lenses you mention, on 5x4 will get you a narrower DoF than any reasonable 50mm (I'll go with your "4x" argument for now) lens on 35mm - although it won't necessarily dispose of the background any better than my 200mm f/2, depending on the scene geometry. I have to say that one reason I stick clear of normal lenses on 35mm is that, wide open, the quality tends to be shocking - in sharpness and LoCA, if not bokeh; I have a 50mm f/1.8 for emergencies, but never felt that faster glass justified itself to me. I imagine (I'm beyond my lens design expertise here) that the LF lenses, being slower, might behave better - which comes back to "best", as opposed to "most", blur of background detail. So you might have persuaded me to love normal lenses (for some photos) after all - so long as I use a LF camera with them. Which is fine - I'm happy to have a long lens for candids and be able to post group shots. That's given me some research to do, anyway! Up until now, I'd mostly been thinking of LF for wide lenses, since my DSLR sensor runs out of pixels for big prints when viewed up close, but you might have persuaded me a 5x4 wouldn't be a one-trick pony after all.<br />

<br />

Good luck with the 420mm; I hope you'll report back.</p>

[rob_piontek]

<p>I have a PhD in physics (really) but often I can't decipher people explanations. But neither was I very good in optics! And it was 10 years ago. I read the blog post quickly, and I think what your missing is the magnification factor of the background that makes long lenses seem to have shallower depth of field. I would put it this way... for the same subject framing, DOF is the same for different focal lengths (for the same format at least) if the aperture is the same. But the longer lens will appear to isolate the subject better because the blurred background is larger. Am I wrong?</p>

<p>I just bought a pentax 6x7 with a 150 2.8 to hope to get something like an 85 1.4. Just got the first negs back today and can't wait to scan them tonight. But I think the quest towards shallower depth of field has it's limits. While I like the portrait of the man in the OPs gallery, to me the shallow DOF (and maybe some kind of movement?) is a bit distracting from the subject. I wonder what non-photographers would think of this portrait? Also, on the other side of things, sometimes when your shooting people you might actually want more DOF. Say you have two people or a group. Or maybe the environment is important to the story that goes along with the picture, and you'd like it to be sharp. Or maybe the light is getting low and you'd rather have a 1.4 on 35mm compared to a 5.6 on 5X4 (though I see you have some fast LF lenses, I didn't know they made these). I've noticed this already with my pentax on the first roll!</p>

[wideopeniris]

<p>The picture you are referring to was shot in a room with curtains in the background that have a large amount of small details in a regular pattern. Even when extensively blurred, there is still that distracting harsh looking background. The shot was an experiment, and I would normall have opened the curtains to remove the distractions. If you look at the beard, you can see that it is very sharp and there is no movement in the shot.<br>

If you look at the other shot on photo.net of the lady, you can see that the plane of focus is actually in front of her face, and yet the blurr is not unpleasant - an interesting effect (her right wrist is sharp). This was caused by a (very) small error in the GG position on my camera. I had to shim it by 0.4mm to correct the plane of focus between film and GG. This is the other problem of large aperature lenses.<br>

You are correct I think about the magnification of the blurred objects. So DOF is constant, but the blurring of background objects a long way behind the subject is greater. So I was not entirely correct. However the effect is not as dramatic as increasing format size.<br>

The reason for my question was to determine whether it was worth using a longer lens of similar aperature (with the attendant weight and inconvenience not to mention cost) for use in portraiture. I normally use my fast lenses stopped down to f3.5 or f4, so its like using an f1.0 lens on 35mm. As you say this doesnt allow low light shooting, but these are posed daylight shots. I get out my 35mm camera for candids.</p>

<p>Kevin.</p>

 

[Andrew Garrard]

<p>My claim (in the DX lens numerology thread on the Nikon forum) was that, relative to the final print (if they're the same size), the grain in the large format will be smaller than in the smaller format, if the film is of the same ISO. If we say the large format is shooting at f/4 and we want to compare it to the DoF on an f/1 35mm lens with the same field of view (and we claim 5x4 is four times the linear size of 35mm), the 35mm grain is enlarged 4x more than the 5x4 grain. If you're using ISO 100 film on 35mm, you could use ISO 1600 film on the 5x4 and get, I believe, the same shutter speed as the 35mm camera, and the same grain size in the final print. An f/4 large format lens may have better quality - something about which I'm quite curious to learn more - than an f/1 135-format lens, if you can find one, but of course the 5x4 is a little more unwieldy (albeit not necessarily heavier than my 35mm cameras).<br />

<br />

There's a common claim that "large format cameras have less grain than 35mm" - not if you match the exposure for the same field of view and depth of field. It's just that, effectively, the film is slower when you consider it in the context of the final print. (I'm not saying the film <i>is</i> slower, or that there's anything wrong with the ISO film measurement mechanism - I opened a can of worms by talking about this in the past - just that this is another way of thinking about it. The same applies to the ISO response of crop and full-frame digital sensors.)</p>

[photom]

<p>I am not quite clear what the question is. "Subject: Which focal length gives best blur of backgorund detail"</p>

<p>It sounds like you are trying to find the lens/format combination that will give the most narrow DOF. Best, and most narrow are not the same. People argue all day long as to which lens gives the best bokeh.<br>

<br /><em><strong>IF</strong></em> you want to find the lens/format combination that <em><strong>mathematically</strong></em> gives the shortest (most narrow) DOF, and therefore, the most blurring of objects outside the DOF that is acceptably in focus, then look at this website and plug in lenses that are available for sale. </p>

<p>>>> <a href="http://www.tawbaware.com/maxlyons/calc.htm">http://www.tawbaware.com/maxlyons/calc.htm</a></p>

<p>Start out using the Dimensional Field of View Calculator near the bottom of the page. There is a lot of trial and error involved but you can find the answer. Start out with a real lens that is available such as Canon 85mm, f 1.2. Decide on a real world fixed long dimension for your portrait - say 4 feet. Plug in the 85mm focal length and through trial and error find the subject/lens distance that gives 4 ft. in the long dimension as the answer. Save the subject/lens distance that solved for 4 ft. Then, go to the top program, plug in this distance, and the remaingning variables, and solve for the DOF. Write down the answer to compare with other possible lens/format combinations you wish to compare with.<br>

<br />Next, try another lens that is available that you believe may be a contender. Do the same thing, keeping the portrait long dimension at 4 ft. You can try medium format film dimensions if you wish, and lenses that are available for same. I am not 100% sure what the answer is, but I suspect I just gave you the answer in the above paragraph. However, there may be some 8x10 film lens that is the answer. But you will have to find the circle of confusion and calculate the focal length multiplier for this lens to verify this. When you have found the answer, please post back because I would be interested in knowing.<br>

<br />Please note, this program assumes a standard image ratio of 3:2 for the Dimensional Field of View Calculator. Not my program so I cannot change this. So when you compare 24x36 film to medium format and large format you will have to mathematically "cut" the film to this same ratio before calculating the focal length multiplier, if you want the exact answer.<br /><br /><br />Again, that is assuming I even understood your question.<br>

<br />Good luck,<br>

<br />= Tom<br /><br /><br /><br /></p>