https://blog.kasson.com/the-last-word/sony-a7ii-ibis-and-sloppy-technique/
Jim
--
Posted as a regular forum member.
https://blog.kasson.com
Last edited:
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
A few questions about IBIS
- Thread starter olyflyer
- Start date
-
- Tags
- nikon z6 nikon z7 olympus e-3
Jim
--
Posted as a regular forum member.
https://blog.kasson.com
Leonard Shepherd
Forum Pro
Thanks for the link - clarifying up to 5 stops with Z mount lenses.
Whether this is for the 3 current lenses, or includes maybe the 70-200 Z is for the future.
"Less" with F mount lenses does not clarify whether up to 3 stops with lenses without in lens VR, or some other amount.
Why is it true that IBIS provides better results for some users than others. There are two main reasons.
The first is about the range of frequencies that the IBIS system is designed to stabilize. It is limited on both the upper and the lower end. The reasons for the upper-end limitation have to do with power, the speed of the voice-coil actuator (VCA) motors and the update frequency of the servomotor control system that controls the VCAs. As the upper frequency at which the IBIS system is designed to be effective increases, the VCAs need to move faster and they get heftier and consume more power. If you've looked at the moving head control systems in a desktop hard disk (don't try this if you want to use the disk afterward), you've seen some reasonably hefty VCAs. In order to make effective use of fast VCAs, the servo controller needs to update faster. That means it runs the control loop faster, stressing the processor more, and, if the processor is beefed up to handle that, consuming more power. So the system becomes ineffective (or even makes things worse) above some forcing function (that's the user's motions) frequency. The lower frequency is limited by the desire to limit IBIS travel. What if the lower frequency were zero? Then the user would pick up the camera, the IBIS would travel to its stop, and it would stay there forever. That would be pretty useless. So the system is designed to slowly center the sensor, limiting the lower frequencies at which it is effective.
The second is that a ham-handed user can apply forcing functions that can cause the IBIS to reach its travel limit sooner than a steady user. This becomes more of a problem as the lens focal length gets longer. I have successfully used IBIS on a 280 mm lens, but if I were sloppier, that wouldn't be the case, and I sure couldn't do it on a moving water ski boat.
Jim
Leonard Shepherd
Forum Pro
Broadening the topic out a little, Nikon IBIS with the 24-70 S seems extremely effective compared to in lens VR in my 24-70 VR.
The first Nikon in lens VR from around 15 years ago was "OK".
Nikon said of VR II it handled low frequencie vibration better.
We are now seeing in lens VR with up to 4.5 stops gain in some newer lens designs.
Nikon used to say in lens VR sampled for vibration 1,000 times a second. This is not necessarily the same as the voice coil motors activating for camera shake 1,000 times a second.
Assuming 1,000 corrections a second; based on a fastest shutter fully open time around an assumed 1/300 (a little faster than a 1/250 flash sync speed0 allows time for 3 VR corrections at 1/250, 6 corrections at 1/125 and so on; with more VR gain at slower shutter speeds.
At 1/500 the shutter movement time remains the same at around 1/300. The 1/500 exposure is achieved by using a narrow shutter slit width. The reduced slit width would allow time for only 1.5 VR vibration corrections. This implies less VR benefit at faster shutter speeds.
Nikon (toward the end of the current F lens brochure) provides CIPA VR gain at different shutter speeds for the 200-500 at 500mm - which are synonymous with the above assumption i.e negligible to nil gain above 1/500 and proportionately more gain to around 1/8.
Edit - the only information I can find on in lens VR CIPA performance at different shutter speeds is on page 40 of the lens brochure - https://www.nikon.co.uk/imported/im...ept/brochure_nikon_nikkor_en-GB--original.pdf
End of edit.
An issue with CIPA is that it is a "laboratory result" using lenses secured to a vibrating anvil.
Real world hand holding, as Jim has pointed out, can be different - and does vary from photographer to photographer.
Fast forward to Roger Circla's tear down of the Z7. This shows a much larger IBIS VR unit than that in a Nikon lens VR lens.
While IBIS corrects in more directions than in lens VR, being larger and taking advantage of recent technology it may vibrate and correct at perhaps 2,000 times a second.
Nikon has said nothing.
Jims tests with IBIS (with Sony) indicate good VR correction at 1/500 shutter speed, though less so at 1/1,000.
The Nikon in lens VR figures of the 200-500 show little gain at 1/500
Nikon IBIS is likely to be as effective as Sony.
It seems probable the correction frequency is (my estimate) twice as fast with IBIS as the best in lens VR.
If correct it would have a positive benefit at 1/500 shutter speeds, in line with Jim's IBIS measurements.
This opens the door for even faster IBIS correction in the future with a clear benefit at 1/1000 shutter speeds.
Last edited:
olyflyer
Forum Pro
To me this is logical and this is also my interpretation of number of stops when we talk about IBIS/VR. After all, the benefits should be the same, or very near the same for everyone. 5 stops is five stops (with some small variations) for everyone.
Peter3579
Leading Member
To enable ibis with an Ai (etc) lens you have to input the maximum aperture and focal length into the non cpu lens data. It then shows on the rear screen that it is enabled along with the focal lenght you have input.
does this mean ibis has different levels of stabilisation at different focal lengths, or is it just reminding you which lens you have mounted?
does this mean ibis has different levels of stabilisation at different focal lengths, or is it just reminding you which lens you have mounted?
Leonard Shepherd
Forum Pro
Short answer - yes - with less benefit on longer focal lengths.
As "camera shake unsharpness" increases with focal length more powerful VR correction is needed to get 4 stops (CIPA basis) with lenses like the 500 PF compared to the 24-70 f2.8 VR.
With in lens VR Nikon say they tailor the VR performance to the individual lens. The most common differences are likely to be vfocal length, weight of elements that are moved in the VR group and the price point of the lens.
IBIS has a fixed benefit and thus becomes less effective with longer focal lengths. It is likely to have less advantage with the 70-200 S (when it comes) at 200mm than the 24-70 S at 70mm.
Leonard Shepherd
Forum Pro
IMPORTANT - the limited information Nikon provides (for the 200-500 at 500mm) for in lens VR indicate no more than 2 stops gain at 1/250, with 4.5 stops not available until 1/6 using CIPA.
The implication of CIPA is shots are as sharp at around 2 seconds with VR as without VR at 1/6 for the 200-500 at 500mm.
In the real world who can get a sharp shot hand holding a 500mm at 1/8 without VR? ;-)
I consider CIPA is often not a reliable indicator of real world VR advantage, and useful only as a comparison tool between different lenses and different brands.
My advice, in line with Jim's guidance, is do your own testing the way you shoot to see what your safe VR shutter speed is using whatever sharpness standard you consider reasonable.
Digressing slightly without VR a super fit body builder might get shots sharp at 1/250 with a lens at 500mm whereas an infirm person with shaky hands might need 1/1000 for a similar level of sharpness.
Things like taking a lot of falling over water, a lot of caffeine or being tired affect how steady most can hand hold a camera and lens.
--
Leonard Shepherd
In lots of ways good photography is much more about how equipment is used rather than the equipment being used.
Last edited:
Peter3579
Leading Member
Thank you. I understand that a fixed amount of ibis will become less effective at longer focal lengths, but that still doesn't explain why I need to input the focal length into the non cpu lens data, as the ibis implementation is constant irrespective of focal length.
Re. the 70-200 S, I will be surprised if it does not have in lens VR.
olyflyer
Forum Pro
I think the reason is that Nikon is using different algorithm for different focal lengths and the focal length is used for calculations inside the camera to give us the most efficient IBIS they could come up with.
It is logical, and also wrong. See above posts by me.
Not true, especially for unusually unsteady shooters.
Jim
Peter3579
Leading Member
Have Nikon ever confirmed this?
I see no evidence that Nikon is using a different algorithm at all. It is necessary for the IBIS system to know the focal length so that it knows how far to move the sensor for a given amount of pitch and yaw. The sensor cannot directly induce compensatory pitch and yaw.
Jim
It's not exactly that it's less effective, but that it runs up against the limits of its travel for less yaw and pitch. Before that, it's just as effective.
It is necessary for the IBIS system to know the focal length so that it knows how far to move the sensor for a given amount of pitch and yaw. The sensor cannot directly induce compensatory pitch and yaw.
Jim
Anonymous Pi
Leading Member
- Messages
- 915
- Reaction score
- 1,030
Possibly a good place to remind everyone that Logic is not concerned with the truth value of the premises or conclusion, but only their relationship to each other.
I don't have any "position" on this question, or any data to support one side or the other, but I am curious how you determined that, "bad technique delivers fewer stops," as opposed to, "bad technique requires more stops than deliverable"?
Bingo.
By testing. There's a link in this thread to one such test.
But there are also engineering reasons why bad technique should result in less-effective stabilization: the easiest one to understand is the limited travel of the IBIS system.
Jim
Anonymous Pi
Leading Member
- Messages
- 915
- Reaction score
- 1,030
I haven't had a chance to look at the links yet, but I meant, specifically, how testing determined one possibility vs. the other. (Perhaps the links address this.)
For example, if one one "runs out" of available travel, is one getting fewer than 5 stops, or does one actually need more than the 5 stops (possibly) provided, and how can one know which?
It seems obvious that it might very well be "less effective" at producing good images with poor technique, but it's not clear to me (again, not having followed up the links yet) how one would determine if it provided fewer stops of stabilization or actually required more than it was able to provide. In the first case, if it provided, say, only 3 when 5 are required, it is 2 stops short, in the second case where it required, say 7, but was only able to provide 5, it will be 2 stops short. How does one distinguish between these cases?
If the system pushes the sensor to the stops, the improvement is then zero.
Take a look at my methods, look at my other IBIS testing -- extra points if you find the posts using the shake table to generate the forcing function.
Then ask questions. Please give me some of your background in control systems, so I'll be able to pitch my answers to your level of understanding of the principles involved.
I consider the "stops" criterion pretty crude. As I've pointed out above, how many stops of improvement you get depends on what you define as acceptable sharpness.
Jim
Anonymous Pi
Leading Member
- Messages
- 915
- Reaction score
- 1,030
Excellent point that had not occurred to me (or that I missed earlier). So, basically, in greatly simplified form, if you need >5, you get 0.
Q.E.D.
Leonard Shepherd
Forum Pro
I presume the camera also needs to know the lens widest aperture to correctly calculate exposure.
There is a lot not made public about the Z6 & 7, such as the new Matrix metering.
As there is no prism and no mirror, unlike a Nikon DSLR, matrix exposure is likely to be calculated entirely from information recorded on the sensor.
Similar threads
Keyboard shortcuts
- f
- Forum