Camera Labs

In the course of my investigations in the new Sigma 120-400mm OS (here) I came across the effect of IS/OS/VR (image stabilization).

As everything connected to image stabilization and shake is a highly statistical game I collected 300 comparable shots in the course of three days with the a.m. lens and my trusty old Sigma 400mm f/5.6 (without any kind of stabilization). Now I come back with some insights about shake in general (which I present here) and the efficiency of the OS on the new lens (in my review of said lens).

I evaluated shots of a testchart at 100% magnification and classify sharpness with 0-5 stars. 0 stars equalling unusable, 5 starts meaning sharp, no blur to be seen. At 50% magnification everything above 2 stars is ok. Full-screen (1920x1200) looks acceptable at 2 stars, but clearly unsharp at 1 star. Give you the examples in a later post.

Let's concentrate on my personal shake statistics handholding, freestanding without the artifical support of IS/OS/VR at 400mm focal length. 139 testshots fall into this category distributed as follows:
27 shots at 1/60 sec, 62 shots at 1/125 sec, 50 shots at 1/250 sec

We'll be back soon with the distribution of stars across these shots!
Stay tuned...

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

Now here's the low-down:
The table contains the number of shots at the shutter speed of 1/250, 1/125 and 1/60 that are of sharpness 0..5 (star rating) plus the sum of shots at that speed.

In absolute terms:
sharpness|.5....4....3....2....1....0....sum
-------------|------------------------------------
1/250sec.|..9..19...11...4....5....2......50 shots
1/125sec.|..2....2....5..14..20...20.....62 shots
1/60sec...|..0....1....1....1....6...18.....27 shots

Relative as % of total shots at a given speed:
sharpness|....5.......4.......3.......2.......1.......0.......sum
-------------|------------------------------------
1/250sec.|.18%..38%..22%.....8%..10%....4%.....100%
1/125sec.|...3%....3%....8%...22%..32%..32%.....100%
1/60sec...|...0%....4%....4%.....4%..22%..67%.....100%

That means that at 1/250 sec 9 out of 50 (=18%) shots are perfectly sharp and 9+19+11=39 out of 50 (=78%) shots are good enough for 50% magnification. At 1/125sec only 2+2+5=9 out of 62 (=15%) shots fall into this category. Now - as you would expect - the situation deteriorates drastically with a shutter speed at 1/60sec: only 2 out of 27 (=7%) shots are sharp enough for 50% reproduction. You can also see that 2/3 of my shots are absolutely unusable at 1/60 sec

So if you want critically sharp shots at 400mm handheld, freestanding, w/o the help of a IS/OS/VR, you better go to 1/500sec. If you're viewing at 50% the 1/250 sec has a pretty high hit-rate.

Now, these numbers may be better or worse for you depending on your individual "shakebility". If you take the rule-of-thumb from film days, that the recommended shutter speed should be 1/focal-length, you could either translate that for a 400mm lens to 1/400 sec or - if you take the crop-factor (1.5-1.6x) into consideration - the safe shutter speed would calculate as 1/600 sec. So I consider myself a typical male in this department as 1/500 sec would suit me fine

If you take the promises of lens-manufacturers at face value, an advertised efficiency of 2-3 stops of their IS/OS/VR should result in a safe shutter speed (with high keeper-rate) of 1/125 sec or even 1/60 sec when shooting at 400mm length.

Well, let's see what the OS of the new Sigma 120-400mm can do for me!

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

A related thread (safe shutter speed with no vibration reducion) can be seen here.

_________________
Nikon D90 + MB-D80 + SB-600 + AF-S VR 18-200mm f/3.5-5.6 + AF-S 35mm f/1.8 + AF 85mm f/1.8 | My photos on Flickr.

Yep, Luis, you were the inventor of the shaker-challenge Thanks for that!

Now the only thing missing are 100% crops from the 0..5 star-rating so that everybody can get a feeling what it means when I classify a shot as "cr*p" or "perfectly sharp". All shots below are 100% crops, taken with the same Sigma 400/5.6 at 1/125 sec!

*****, perfectly shake-free


****, slight traces of shake


***, clear traces of shake


**, no longer good enough for 50% viewing


*, almost lost, even for small prints


0 stars, unusable

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

Thomas, this thread apply just as well to build-in image stabilisation, such as in my A700, right? I'll see what else comes up, thanks!

- Bjorn -

_________________
Street and documentary photographer | Google+ | Twitter | Blog | Tumblr

Leica M9-P (my article on Camera Labs) | Leica D-Lux 5 | 50mm Summilux

Bjorn, this is just a way to do some meaningfull statistics. The most important part is that you need a comparable consistent measure for shake. I'm doing this with my eyes and I can tell you, my head almost hurts. But at least I'm reasonably sure to judge all my blurry pictures alike.

Now, as to your question: With in-body IS or in-lens IS or even without IS.
Once you have taken your "profile" with IS switched off based on a reasonable number of shots with different shutter speeds you can then compare the effect of switching IS back on again and see how your "keeper-rate" improves. This is what I will do next with the new Sigma 120-400mm. See you over there in the review-thread of that lens!
Just give me a little time to put the results in tables. If not tonight then tomorrow. I'll promise!

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

Well, how does the saying go: "Lies, damn lies and statistics"?!
The deeper you dig, the more your head hurts...
I was just concentrating on analysing the 120-400mm data, when it came to me as a shock that the base data above (that is for my shake without IS-support) are quite different depending on which lens I take. The above statistics were derived from the sum of shots from the Sigma 120-400mm at 400mm plus the Sigma 400mm (naturally at 400mm ).

But when I look at my favourite shutter speed of 1/125 sec there is a hugh difference in blur between both lenses and guess which lens comes out on top? Well yes, the lighter, leaner 400mm fixed focal! (I'll come back to this a little later)
There are 23% shots rated at >=3 stars fom the fixed focal vs only 3% from the zoom
From the worst shots (1+0 stars) the fixed-focal had 53%, the zoom had 69%

The picture changes at 1/60 sec shutter speed. There the blur distribution between both lenses is much closer.

At 1/250 sec shutter speed you have 56% >=4 stars with the zoom and 56% >= 4 stars with the fixed-focal with a clear advantage for the latter lens at the 5 star rating.

What do we get out of these statistics?
1. It seems that the 1/125 shutter speed is the "critical speed" at 400mm where the results are most sensitive to influences. At the shorter speed of 1/250 sec things are getting more equal because shorter speeds should eliminate shake more often. The same is true with 1/60 (and should be at even slower shutter speeds): The odds are aginst getting any sensible IQ out of those shots.
2. If this is so (the 1/125 sec being the discerning speed) and the samples are large enough (in this case around 30 shots each) the results could be statistically significant.
3. The only explanation that comes to my mind are:
3.a The leaner fixed-focal lens is easier to grip
3.b the lower weight of the fixed-focal (1300g vs. 1600g) is easier to hold steady and aim (but there is certainly a limit to this logic: my 600g 500mm mirror lens is much harder to hold steady than the 400/5.6!)

Are there any consequences to m review of the 120-400mm?
Yes, there are. If I compare the 120-400 with OS="on" with the statistic that mixes the results from the fixed-focal and the zoom one could say that this is not fair. Ok so for the sake of an apples-to-apples comparison I present here the results for the 1/125 sec shake statistic purely for the 120-400mm lens:

Relative as % of total shots at a given speed:
sharpness|....5.......4.......3.......2.......1.......0.......sum
-------------|-----------------------------------------------------
1/125sec..|...0%....3%....0%...28%..34%..34%.....100% (only 120-400mm)

((any numbers that don't add up are due to rounding errors!))

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

Me too.

The two Sigma sites I've checked have no mention that I can find of the 400mm prime. I take it that it's the Nikon version of the 400mm f/5.6 HSM APO macro reviewed by PhotoZone here?

It looks like it's much longer than the APO 120-400mm F4.5-5.6 DG OS HSM so I'm left wondering if it has a similar moment of inertia despite being lighter. That might explain your ability to hold it a bit more steadily.

This is a great thread and I can see that a tremendous amount of work has gone into producing statistically meaningful results. I'm still not sure where this is taking you so, once the final tests results are in, I look forward to reading your executive overview. Not that I'm any sort of executive but when all is done I think I'm really going to need that overview.

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

Ooops, I got the weights of both lenses not correct Now they are!
The length of both lenses at 400mm focal length without lens-shade is almost identical. But certainly the weight-distribution is different: the fixed-focal weighs 800g at the front and 500g at the back, the zoom is 780g in front and 820g at the back when extended to 400mm. But that in turn means that the inertia against turning (sideways or up/down) should be roughly the same. That leaves only the additional 300g plus perhaps an easier grip on the leaner fixed-focal and of course statistical deviations as explanation.

As for the consequences of this all: Well, I have to decide toworrow whether to keep this lens or give it back. That will be the ultimate executive summary

-----------
Oh and yes, the 400mm fixed-focal is the discontinued 400mm f/5.6 APO macro.

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

Interesting thread and tests Thomas, thanks!

Looking forward to hearing how the new OS version compares...

I've started to unravel the mysteries of shake-reduction (OS) on the Sigma 120-400mm here. Stay tuned, there is more to come...

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

I just wanted to add a note about the 100% magnification used in the presentation of the images above and the relevance to "everyday use".
Clearly, looking at 100% crops is 100% "pixel-peeping" so a few remarks are in order as to whether this has any practical relevance:

Look at the thumbnails in the flickr-preview of those shots here. These are smaller versions than the original 100% crops: while the original 100% crop has 676x423 pixels the thumbnail is 240x150, that is a reduction to 35% of the original or approx. 1/3 original size. The full image would be then 1500x1000 pixels and would easily fit on any of todays modern monitors.

Now judge for yourself with these thumbnails: Which images don't look optimally sharp (hint: the number X in the name "ShakeX nnnnn" gives the number of stars this shot got from me). Me, I'd say that Shake0 and Shake1 are clearly blurred and Shake2 is a little soft.

So even with the original image shrunk 2/3 to fit your screen the difference in IQ is clearly visible

And believe me: making a good A3(+) print at 33x48 cm from your 12MPix images needs all the sharpness that you can get. There are even people out there who sharpen their "good" photos before printing them because on paper you have at least 3x (linear) the pixel density than on your monitor. That's when you even see the difference between 3-star and 5-star sharpness easily

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

YOU, Thomas....are wild.

we need to call you the research-meister...good job...I skimmed over the bulk of it, wondering where it was all going...and NOW I am going to go read the entire posty posts, and hopefully learn from your expertise.
And, just how does all of this relate to quackage, I wanna know.

patti

_________________
canon 7D, Mark iii 5D, ef 70-300 f/4-5.6, ef 28-135mm canon 100-400mm L IS 4.5, and a little 24-70 L f/2.8...

Yeah, I should have titled this one "Shake, quacki! Shake!"
I'm sure that would have got more views. But as this thread is very head-heavy and nothing for some unadulterated quacky fun I thought it better to avoid any connotations that would have lead more people to read this stuff
But it's nice you came by!

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

well, you know what Thomas..this IS dedication. I can hardly wait to sit down tonight, after work, and mull it over.

You would be surprised how many people, even quack aficionados read thru every thread....


patti

_________________
canon 7D, Mark iii 5D, ef 70-300 f/4-5.6, ef 28-135mm canon 100-400mm L IS 4.5, and a little 24-70 L f/2.8...