The reciprocal-of-the-focal-length rule for hand holding

[StewartR]

I'm sure most people have heard of this old rule of thumb. Basically it says that, when using your camera hand held (and without image stabilisation!), you should keep the shutter speed at least as fast as the reciprocal of the focal length in order to avoid the effects of camera shake on your images. For example, with a 50mm lens you should shoot at 1/50th or faster; with a 200mm lens you should shoot at 1/200th or faster; and so on.

Obviously it's just an approximation and different people will get different results. I have a slow resting pulse and steady hands, and I expect to get away with a shutter speed that is one stop slower than the rule suggests. My wife has less steady hands and needs to shoot one stop faster than the rule. But still, it's a useful starting point. If you have a 200mm lens, you're more likely to find the threshold of acceptability around 1/200th than around, say, 1/10th or 1/4000th.

But here's my concern. It's an old rule which was derived in the days of 35mm film SLRs. So how does it apply in the digital age? Specifically:
* Does the sensor size make a difference? I've seen people recommend that it's the full-frame-equivalent focal length which should be used, and that sounds reasonable, but I haven't seen any science to underpin that recommendation.
* Does the pixel density, or the number of pixels, make a difference? I've seen people recommend adopting a more conservative rule for the D800 camera, but not for other DSLRs which have higher pixel densities, which seems odd to me.

So - Who can help answer the questions?

I don't want opinions. Opinions are worthless here unless they have some scientific underpinning. I want FACTS. Either the results of experimentation, or analysis based on the underlying physics of the situation. If you haven't done that, perhaps you've seen references online to other people who have?

 

[ChrisHeathcote]

I haven't explored the theory, however U wouldn't, have thought sensor size would make a difference. My logic is as follows. At longer lengths the tiniest of moves at the body can equate to a larger movement at the end if the lens. The magnification is more sensitive to this. The reason I don't think sensor size matters is that the magnification doesn't change between crop and full frame, the only thing that changes is the field of view. The crop sensor cuts off the outer portion of the frame making it appear as though a longer lens was used.

Apologies for the lack of scientific fact

 

[Slaphead]

The reciprocal rule (or guideline really) was originally a way of getting an acceptable sharpness for a given standard print size - I can't remember the actual print size. If any enlargement is to be done then a faster shutter speed would be more appropriate. Obviously this would vary from person to person - somebody with the shakes would need a higher shutter speed, and somebody with sniper steadyness would be able to use a lower shutter speed.

This translates into the digital world with SS = 1 x FL for full frame, and SS = 1.5 x FL for crop, but only if that given print size is to be used. However with digital we tend not to just print to a given size, but we often crop the hell out of an image, pixel peep, etc which means that this rule has become somewhat out of date. So what is the rule with digital? well the answer is whatever it needs to be to get the shot given the conditions?

There are a few ways of looking at this, but ultimately pixel pitch (pixel density) plays the most significant role these days as, whether it be right or wrong, a lot of people judge sharpness when the the image is on the computer and zoomed in to 100%. Let's take the D800 (36mp FF) and the D7000 (16mp crop) as an example here as they have, for all intents and purposes, the same pixel pitch. A photo taken with the same settings on each of these cameras should yield pretty much the same sharpness at 100%. Now let's take the D4 (16mp FF) and the D800 (36mp FF). Now assuming the same camera shake the D4 will have a sharper image when viewed at 100% than the D800. To acheive the same level of sharpness when viewed at 100% the D800 would need a shutter speed of approximately 1.5 times that of the D4.

OK, viewing at 100% is not the whole story, and probably not really how sharpness should be judged in all situations. Taking the above example you would notice that on screen the D800's image would be visually larger than that of the D4's, so if we were to zoom the D800's image to 66% thus making the the two images visually the same size you should no longer be able to see much, if any, difference in sharpness.

This leads us to the printed image. Assuming the resolution of the camera with the least resolution is enough to allow a print at the printers resolution for a given size, then there will be no difference between the sharpeness of prints from each of the different cameras. Any extra resolution provided by the other camera, which would show any blurring due to movement, would be lost due to the printers resolution.

There are too many variables these days to say that setting the shutter speed to 1 x (or 1.5 x) focal length is still as applicable as it was back in the film days. Also higher resolution cameras do benefit from a higher shutter speed than what the rule would suggest. However one thing that can be said is that the larger your intended output size, or the more you crop, then the higher the shutter speed should be. That said FF in general allows for a slower shutter speed than crop frame of the same resolution for a given output size.

IMO the guideline still works, but only as a minimum starting point.

The other rule I use is 2 beers = 1 stop of steadyness, but you can only take this rule so far

 

[ecoleman]

No scientific facts, but I agree with the logic in post 2. Logic says that the sensor size shouldn't make a difference.

 

[desantnik]

I don't think you can get a truly mathematical scientific analysis of this because the key factor is how steady your hands are.

Maybe you could take a vast sample across many hundreds of users to get an average but even then you have to have a benchmark for "acceptably sharp" that is consistent across your test.

However, I will add another factor to your potential test parameters, the weight of the camera+lens.... holding a 300 f4 on a 350D is easier than holding a 300 2.8 on a 7D with a grip.

 

[Brian_of_Bozeat]

I think you hit the nail on the head when you said that what works for you doesn't work for your wife... The trick is to find what works for you with your camera. It's not a rule it's a guide.

 

[artyman]

Whilst a good rule of thumb we do have the luxury of IS these days, assistance undreamed of years ago, however it depends person to person, the same reason some folk can hit a target better than others they keep it still. As regards sensor resolution, say fro arguments sake you are wavering 1/10mm and the pixel is 2/10mm the movement stays within the pixel which can only record one level of info so no movement is shown. If the pixel is 1/20mm then the adjoining pixel may get that info so blurriness can occur, my experience is that all things being equal my higher MP camera is less forgiving of camera shake, and that is without cropping.

 

[ernesto]

I haven't explored the theory, however U wouldn't, have thought sensor size would make a difference. My logic is as follows. At longer lengths the tiniest of moves at the body can equate to a larger movement at the end if the lens. The magnification is more sensitive to this. The reason I don't think sensor size matters is that the magnification doesn't change between crop and full frame, the only thing that changes is the field of view. The crop sensor cuts off the outer portion of the frame making it appear as though a longer lens was used.

Apologies for the lack of scientific fact

Agree. It is about magnification and movement so as long as using an equivalent length to compare the sensor makes no difference.

 

[KIPAX]

I don't want opinions. Opinions are worthless here unless they have some scientific underpinning. I want FACTS. Either the results of experimentation, or analysis based on the underlying physics of the situation. If you haven't done that, perhaps you've seen references online to other people who have?

OK so as everyone else posting has totally ignopred your last paragarpah.. i will as well

Your taking a rule of thumb.. which you yourself are aware is just a guide that may or may not work for some or others.. then you want scientific underpinning and facts ? ...The two dont go together.. you may as well ask for the exact heart rate of a cartoon character

 

[Phil V]

OK so as everyone else posting has totally ignopred your last paragarpah.. i will as well

Your taking a rule of thumb.. which you yourself are aware is just a guide that may or may not work for some or others.. then you want scientific underpinning and facts ? ...The two dont go together.. you may as well ask for the exact heart rate of a cartoon character

What he said, and just to add, sensor size matters, all other points being equal.

Try the experiment yourself, hold a pencil in your fingertips at the blunt end, place the pointy end within a small frame, say 1" square, see how much it's apparently moving. Now retry the experiment inside a 6" square frame, see how little it's now apparently moving.

That's your proof that sensor size makes a difference.

But it's still a rule of thumb, some people are very steady and some aren't, sharpness is relatively subjective and if you want to make a massive print, use a sturdy tripod.

 

[swanseamale47]

The question cant really be answered by any scientific facts, theres too many variables, whats acceptable, what size print at what viewing distance , how steady your hands are etc etc.
The so called rule was only ever intended as a guide, I'd take it all a bit less seriously.

 

[HoppyUK]

The science is quite straightforward.

It is all about magnification, and the standard of sharpness assumed is the same as for depth of field calcs. The fact that the numbers translate directly to use on a full-frame camera is pure coincidence, and any change to that must be adjusted by the crop factor. By the same token, cropping a full-frame image has the same impact as shooting on a smaller format camera in the first place, eg cropping down to 50% of image area has the same effect on camera shake as shooting with 1.4x longer focal length.

 

[desantnik]

My comment about determining the average user and weight of the camera+lens remain though Hoppy

 

[HoppyUK]

On the related issue of standards of sharpness and pixel density etc, the 'rule' as we know it refers to acceptable* levels of sharpness on a standard size print viewed from a standard distance - same as depth of field, eg 10in print viewed from a distance equal to the diagonal, ie 12in (or any larger size print with viewing distance adjusted).

As such, pixel density has no significant influence, as that standard is already substantially exceeded with very modest pixel numbers**. However, digital practise throws everything out of the window when we view at much larger effective print sizes on screen, and from much closer than the standard distance.

Nikon D800 is a case in point, and if you are to realise the full resolution potential of 36mp, then you must a) print large, and b) view close. That takes everything so far beyond the 'standard' that the 'rule' becomes irrelevant, which is why Nikon suggests only the best lenses, at optimum apertures, with camera on a tripod and mirror lock-up engaged.

* the standard assumes a circle of confusion of 0.2mm in a 10in print viewed from 12in
** to meet that basic standard only requires 1.1mp

 

[HoppyUK]

My comment about determining the average user and weight of the camera+lens remain though Hoppy

Absolutely. This whole business is awash with huge variables. Different people, different hand-holding techniques and circumstances, different cameras, weights and sizes. The theory only holds good in an 'all things being equal' situation that rarely applies in practise.

And overlaying everything is the rule only affects resolution potential, whereas what we call sharpness is a combination of both resolution (the fineness of detail) and contrast (how clearly those details are shown). In fact, it is image contrast, not resolution, that has the greater influence on perceived sharpness. Example - when we see an image on here and think it looks very sharp, we cannot be referring to resolution or pixel numbers because they're way low. What we're seeing is high image contrast, or 'punch' and that's down to the lens, not the sensor (basic MTF theory).

 

[HoppyUK]

Further to the above, some comments on the variables.

Hand-holding technique is a big one. All the weight should be taken by your left hand, cupped under the camera/lens around the balance point. Right hand does nothing but steady the camera and operate controls. Left elbow should be pressed against the side of your chest, forming a supporting triangle. Firm but comfortable - any muscle stress translates directly into more shake. Stance is ideally with feet 18in or so apart, and usually turned at a slight angle (to the right) to the subject. Squeeze the shutter release gently, don't stab at it. With these techniques, most people can reliably match the standard rule.

Things can be further improved by leaning against a wall, or made much worse if you're maybe down on one knee and leaning forward to get the right angle. A bit of weight in the camera/lens helps, up to a point and within reason, and the opposite of that would be a light weight compact or CSC held at arm's length - not good, though image stabilisation helps a great deal.

When you're around the hand-holding limit, there's safety in numbers. If you shoot several frames, there's a good chance one will be significantly sharper than the others.

 

[KIPAX]

Hand-holding technique is a big one..

As i know my 70-200 will prob never go on a monopod or tripod.. the first thing I did was take the tripod collar off.. makes it much more comfortable to hold and much firmer for me.. takes a bit of wieght off and doesn't hurt your hand after hrs at an event

 

[arclight]

On the related issue of standards of sharpness and pixel density etc, ........................

Nikon D800 is a case in point, and if you are to realise the full resolution potential of 36mp, then you must a) print large, and b) view close.

Or crop heavily, which compares to printing large

That takes everything so far beyond the 'standard' that the 'rule' becomes irrelevant, which is why Nikon suggests only the best lenses, at optimum apertures, with camera on a tripod and mirror lock-up engaged.

Nikon do indeed suggest all these things with the D800, Richard.

* the standard assumes a circle of confusion of 0.2mm in a 10in print viewed from 12in
** to meet that basic standard only requires 1.1mp

 

[artyman]

Lens shake is directly proportional to the rarity of the bird you are shooting

 

[Teflon-Mike]

Go back to 1st principles.

Car traveling at 30mph / 50Km-h... 13m per second It's about 5m long, 2m wide and 1.5m tall.

It's conveniently going round a race track for me; so I am set up at the side of the track, the camera on a tripod...

Lets ignore lens lengths, and subject to camera distances for the sake of simplicity, for a second.

I frame on the car, side on, to get nice balanced proportion... 1/4 frame in-front of car, 1/4 frame behind, the car occupying half the width of the frame. So at subject distance, my frame is covering aprox 10m of track

If I were to use a 1 second shutter, then... car would travel 13m along the track... almost three times its own length.... and more track that I have in frame... result, car drives in one side the frame, leaves the other, before shutter closes... I get a streak, not a car.

If I went up to 1/60th.. car is going to move 21cm along the track, accross a frame that at subject range, is aprox 10m wide... so the car would blurr over 21cm of 1000... 2% of frame.

If I go up to 1/125th.....car will move 10cm along the track, blurring just 1% of the frame.

1/250th? 5cm along track, 0.5% of frame....

1/500th: 2.5cm of track, 0.025% of frame

1/1000th: 1.25cm of track, 0.01% of frame.

STOP the car.... leave the camera on the tripod, I can use almost any shutter speed... the car is NOT going to move in relation to the frame, there for no motion blurr.

The degree of 'blurr' is then how much movement there may be in the frame, during exposure time.

The lens length and sensor size, is then to a large degree immeterial.

Its a matter of how much movement you might get between subject and sensor.

So take the camera off the tripod and hand hold... if the car is stationary, its not going to give rise to any motion blurr, its entirely down to how steady your hand is...

So lets turn this around; Lets say, that the camera shakes... what? 0.25mm left to right in your mitt...

You have a 36mm wide 'full-frame' sensor... that's going to be about 0.7%
You have a 24mm wide 'crop' sensor... that's going to be about 1%

WHATEVER you have in the frame..... its ALL going to blurr that much.

So... smaller sensor = more motion blur.
(same amount of shake, during same length exposure)

However much 'shake' you have, the smaller the sensor, the larger that amount of movement is going to be as a % of the frame. And its the same 'ratio' as the equated length of lenses.

So lets go back to the moving car and camera on the tripod; if you have a 200mm lens on a full-frame camera, its going to be smaller in the frame, than on a crop sensor camera.

I suggested framing so that the car filled half the width of the frame. 200mm lens on full-frame camera, we have 36mm wide frame, and the car occupying 16mm of it. Put that same lens on a crop sensor camera; you have 24mm frame, the car still taking up 16mm of it.... but... err... 16mm of 24 is 2/3 the frame not half, and so any motion of the car during exposure is going to blurr about 30% more frame, the viewed image...

So; 200mm lens on full frame camera, rule of thumb suggests 1/200th for hand holding. same 200mm lens on Crop, has effective length as a 280mm on full-frame.

Same subject, same movement, but 'effectively' magnified... no not on the viewing plane, but within the viewed FRAME.... any motion blurr will be magnified by similar proportion.

To wit answer is; the 'when hand holding, never go lower on shutter than your lens length' is not far off for full-frame, and probably 'close enough' on crop but, %'s being %'s smaller sensor IS going to suffer more blurr in the frame for the same amount of shake. And its ONLY a 'rule of thumb', which you have already accepted, depends on how steady a hand the photographer has.

If I was shooting with my old Sigma Film camera; with full-stop shutter increments.... 200mm lens... I would have the choice of 1/125, 1/250, or 1/500. So 1/250th would be 'close enough'... If I used 1.4x converter? 280mm 'effective? 1/250th is still going to be the closest to my lens length to not be so worried about.

I'd be more worried about what I was shooting, and what was most apropriate to the subject & situation.... and if I was shooting that car going around the race track.... I would probably completely IGNORE the rule of thumb anyway...

210mm lens on 35mm camera and like as not I would be shooting down as low as 1/60th.... whilst 'panning' to blur the back-ground from actual camera movement, while limiting the subject 'blur' in the frame tracking, trying to keep it centered in the frame, within 1% or so.... something like this:-

If I was taking a portrait, at close range with 50mm lens, pretty wide to blurr my back-ground? Well, 1/60ths the closest I have... might get away with 1/30th, but, with close subject, large in the frame, and critical focus, I would probably incline towards using 1/125, to be on the safe side, and in good sunlight, probably need a shutter that high, or higher anyway, to avoid over exposure, at wider, back-ground blurring aperture....

Unfortunately, as you might be able to guess... portraiture is not really my forte! But you get the idea.

Like the f16-Sunny 'Rule' or the 'rule of thirds', its a base-line starting point; to work from.