Talk:Image stabilization

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Film (Rated C-class)
WikiProject icon This article is within the scope of WikiProject Film. If you would like to participate, please visit the project page, where you can join the project and see lists of open tasks and regional and topical task forces. To use this banner, please refer to the documentation. To improve this article, please refer to the guidelines.
C-Class article C  This article has been rated as C-Class on the project's quality scale.
Checklist icon
Taskforce icon
This article is supported by the Filmmaking task force.
Note icon
This article needs an image (preferably free) related to the subject, such as a picture of the set or a film poster. Please ensure that non-free content guidelines are properly observed.


Can we get some information about the effectiveness of optical/mechanical stabilization (and maybe even digital stabilization)? Right now I can't really tell if this technology is just there to sell more cameras or if it markedly improves the quality of the pictures. Mazin07CT 01:21, 29 November 2006 (UTC)

I don't have personal experience with stabilization technology, as my camera is a bit of an oldie, but according to this 2004 Shutterbug article [1], it is very real and makes a marked difference. Eikonografos 22:01, 11 February 2007 (UTC)
Personal experience is that if it is available (Slow shutter speeds and/or no flash makes it unavailable), you can literally swing the camera like a sword and take a picture, and it will come out perfectly fine. 21:28, 15 February 2007 (UTC)
This is a complete over-exageration. Swinging the camera 'like a sword' will create a blurry image at all but the fastest shutter speeds. If the lens isn't pointed at all in the same direction during the shutter release, IS can't magically move the ENTIRE lens back. Ctrlfreak13 05:19, 18 May 2007 (UTC)
Personal experience is that it does work effectively. A 2001 comparison of Nikon and Canon IS lenses here concludes "Both dramatically improved our chances of getting acceptably sharp handheld pictures at slower shutter speeds" Chrisjohnson 22:17, 16 May 2007 (UTC)

The number "3-4 stops" seems to come from manufacturers marketing materials, not from actual benchmarks/tests. Independent tests usually show something between 1.5-3 stops. --Hkultala (talk) 06:37, 30 July 2008 (UTC)

My testing of Pentax K20D gives 4 stops with very high percentage, and can even give more than 6 (!) stops. This is with the latest firmware. Without SR/IS I don't want to shoot under the "limit", so I'm not a human tripod.

-- (talk) 08:33, 2 May 2009 (UTC)

Added a video with software image stabilization. Thumbnail is missing, but meh. --Oggencoder (talk) 16:31, 19 September 2008 (UTC)

Image Stabilization drawbacks[edit]

Are there any drawbacks to using the image stabilization feature of a camera?

Some(many/most/all?)digital cameras require you turn off IS when the camera is static, such as when using a tripod. Otherwise, it can blur the image.
IS uses more power, so you may experience shorter battery life when IS is turned on. Ctrlfreak13 13:29, 19 May 2007 (UTC)

Optical image stabilization as Canon-specific term[edit]

Optical image stabilization is certainly not specific to Canon; nor is the use of OIS as an abbreviation. I don't know if the precise description recently added applies to other manufacturers, though, so I've tried to rewrite in a way that preserves the information without being misleading. --Steve Pucci | talk 21:43, 27 October 2006 (UTC)

Merging of Vibration Reduction into this page[edit]

The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section.

Any opposed to this merger please state argument below and sign name. Suggestions as to how this merger should be executed are also welcome, again, please sign your name. (Ctrlfreak13 22:42, 6 February 2007 (UTC))

VR is truly just a Nikon brand name for a type of image stabilization, so I don't see why it should not be merged with the image stabilazation page. We should create a sub-catergory for it within the image stabilization page, though. The Nature Guy 21:41, 2 May 2007 (UTC)
Support merging. VR is specifically optical image stabilization as we've defined it here, so it should go into that section (in fact there is already a reference to it there). If there are differences between Nikon's and Canon's implementation that are significant I suggest we call those out in the sentence that mentions VR; otherwise I think the more general info in Vibration Reduction should just get merged into the OIS section. --Steve Pucci | talk 04:58, 14 May 2007 (UTC)

As of 15 May 2007 (EST), Vibration Reduction (VR) redirects to Image stabilization. The previous content of the VR page shold still be accessible by accessing the page by overriding the redirect. Ctrlfreak13 03:38, 16 May 2007 (UTC)

How is that done? The redirect seems to take place automatically, without means for interruption. The carefully crafted VR text has effectively disappeared. This doesn't look like a merger so much as a mass deletion of content. Correction, please? Hertz1888 04:19, 16 May 2007 (UTC)
I believe Ctrlfreak13 meant that the text is available to editors who want to refine the merge. One way to override the redirect is to click on a link like Vibration Reduction. When it redirects to this page, there will be a phrase at the top of the article that says
(Redirected from Vibration Reduction)
Click on the Vibration Reduction link in the phrase, which goes to the original page here. Once you are on that page, you can either edit it (the text won't be visible on the page but you can still see it if you edit the page), to recover the raw text for merging, or use the history tab, to go back before the redirect was done. --Steve Pucci | talk 14:08, 16 May 2007 (UTC)
Steve, thank you for the guidance, but... what you are telling me amounts to saying the job of redirecting/merging has been left less than half done. In fact, no merging has been performed at all. Until such time as someone gets around to incorporating ALL the information on the VR page -- which I and others worked hard to develop and polish -- into the IS page, this information is lost to the reader. Wouldn't it make sense to immediately move this information in its entirety to the IS page? Wouldn't it make even MORE sense to undo the redirect and simply link from VR, where the topic is treated in detail, to the IS page, and vice-versa? Hertz1888 15:17, 16 May 2007 (UTC)
No problem, and I appreciate that it's not very nice to see your work evaporate. I confess I didn't really look to see what had been merged already; I did note that there were a couple of edits to this article from Ctrlfreak13 so I assumed he/she had done the merge to their satisfaction. I agree that if the merge is not complete, the redirect should not be there yet, and I would support restoring the original VR article until we have consensus that the merge is complete. --Steve Pucci | talk 15:36, 16 May 2007 (UTC)
I have just migrated some more information from the old VR page to the IS page, I hope this helps and of course feel free to migrate more and do a general clean-up of the IS article. A lot of the information on the VR page, however, was already present on the IS page as the technologies are the same, just under different names. I am working on adding more references to some of the content here. Ctrlfreak13 16:51, 16 May 2007 (UTC)
Thank you, Ctrlfreak13. I will gladly work on the migration and cleanup, but am unable to tackle it immediately. I hope no one will mind very much if we put the redirect on hold in the meantime, so that the existing material stays available to the Wiki user. With that in mind I have temporarily reverted the VR page to the status quo ante. Hertz1888 22:14, 16 May 2007 (UTC)
Hertz1888, I feel like now we may be ready to reinstate the redirect, I have read through the VR page and believe that all the information there is now well represented in the IS page; I do agree that my initial actions were taken too soon and appreciate your assistance in merging content over. Just wondering where you think the merger's status currently stands. Ctrlfreak13 13:37, 19 May 2007 (UTC)
Greetings, Ctrlfreak13. I see only one small issue remaining. I believe Nikon's more vigorous "Active Mode" may be in addition to, instead of identical to, the vertical-only mode that comes into play in panning. This question can be dealt with later, subsequent to a merger. As far as I am concerned, you may proceed to push the button (or shutter release). Thank you for the collaboration. In the end, the teamwork looks to have been a big success. Hertz1888 14:15, 19 May 2007 (UTC)
Okay, I've now put the redirect back and will look more into Nikon's Active Mode. Thank you again for your help with the merger. Ctrlfreak13 16:12, 19 May 2007 (UTC)
I've taken care of it. We should be all set, at least w/r/t Nikon. Does Canon offer a similar capability, perhaps under a different name? Hertz1888
Yea I saw, I edited a couple of things, you made too much of an emphasis on auto-panning detection, which Canon does not offer and instead is done by flipping a switch on the lens (Canon dubs this as 'IS Mode 2'). Canon doesn't offer a mode similar to Nikon's Active Mode, although they could very well have their IS auto-detect larger movements and act accordingly instead, as they do when tripod-mounted (except kinda the opposite purpose). If you can't tell I'm a photo-buff :) Ctrlfreak13 21:46, 19 May 2007 (UTC)
How could I ever have guessed? Yes, and a knowledgeable one who's able to maintain his focus. The article has just about doubled in size in a month, and become strong, informative, and notably less dry. I enjoyed working on it with you. All the best. Hertz1888 22:06, 19 May 2007 (UTC)

Final result was the merge of Vibration Reduction into Image stabilization. -- Ctrlfreak13 01:48, 21 May 2007 (UTC)

The above discussion is preserved as an archive. Please do not modify it. Subsequent comments should be made in a new section.

Engineering choices: In body versus in-lens[edit]

Hi: There's a bit in the article but I'm not certain the performance benefits are addressed. This may require a bit of research to answer as it's probably an abstruse design philosophy difference but Leica, Nikon and Canon all put vibration elimination in the lenses while a company like Pentax has opted to put it into the camera body.

What are the advantages and disadvantages from an engineering point of view? Weight is covered in the article but what about performance and longevity of the mechanism? The article also says that in-lens vibration eliminators are fine-tuned for the lens but how much of a difference in performance and reliability is there? How about in their zoom lenses?

On the other hand, Pentax aims for consumer rather than pro SLRs and may sell image stabilization as being on all the time even when it doesn't need to be (ie below 100 mm); or for hand-held flashless night photography but these are marketing rather than technical reasons.

Of course image stabilisation is useful for focal length below 100mm. What are you talking about? It may be a marketing reason to claim that image stabilisation is not needed in a fast fifty, but boy it helps at 1/20. ClassA42 (talk) 05:06, 31 August 2010 (UTC)

I don't know the answers but I am curious. —Preceding unsigned comment added by (talk) 21:00, 13 January 2008 (UTC)

I would also guess that A) VR is more or less useless unless you are over 85mm at least and B) in lens may be faster and more mechanically robust than accelerating and decelerating the FF sensor of pro cameras. How it VR plays out in macro photography I can't guess since I never do that. —Preceding unsigned comment added by (talk) 15:35, 30 June 2008 (UTC)

In lens versus in body[edit]

We'd have to hear from engineers working on the problem but full frame sensors aren't massless and the frame of electromagnets used to jiggle the sensor about may have acceptable performance in mid-sized and small photo sensors but not have the responsiveness needed in full frame.

I wonder if the lens moves in the optical systems weigths more or less than a sensor? Doesn't Sony already have a stabilized full frame sensor camera? -- (talk) 08:49, 2 May 2009 (UTC)
Yes, the A850/900 have a stabilized full frame sensor. ClassA42 (talk) 05:04, 31 August 2010 (UTC)

In Lens versus In Camera[edit]

  • In Lens works with film backs. No support for film/alternative backs with In Camera
  • In Lens gives three to four stops while In Camera gives two stops at regular focal length and not Macro/Microscopic photography.
Regular lens-based systems are not fit for macro/micro photography either. Only special "hybrid" IS lenses are. It is true though, that currently there are such lenses while I'm not aware of a camera-based IS system fit for macro photography. The difference in stabilisation performance you state is not backed up by real world test results. Both systems yield comparable results. ClassA42 (talk) 05:03, 31 August 2010 (UTC)
  • What you see is what you get: In lens can be seen through the viewfinder. Live Preview (the gimmick of the season) is needed for In Camera.
  • Image correction is accomplished at the natural magnification of the lens while In Camera, the IS compensates for a magnified vibration. Additional artifacts may be added. I say may because I don't know for sure as the sensor movement algorithm is unknown to me
  • I don't know about scientific/engineering photography but how well infra red/UV photography works with image movement correction. Most sensors have UV filters built in, which means you can't do certain kinds of spectral analysis photography with In Camera image stabilization.
Sorry, this doesn't make sense at all. No disadvantage for the sensor-based system here. ClassA42 (talk) 05:14, 31 August 2010 (UTC)

Maybe someone can write this up for the article?

  • The reason why there is lens based IS at all is because of film. Had IS been invented in the era of digital photography, all the cameras would have in-body systems as they work with all the lenses, not just selected few.
  • There is no evidence on your claim of lens based systems being more effective. Canon's marketing studies are not evidence. I also have no evidence, only personal tests on sensor based system: 4 stops is easy, even with a high pixel density sensor at pixel level. I have done very limited testing, but at 350mm I've achieved more than 6 (yes, six) stops benefit. IMHO, it is not about the method of stabilization, but the effectiveness of the implementation.

I'm an engineer. You're correct it the matter that implementation matters; however, it's far easier and to implement an in-lens system and to make it more responsive. -Tom —Preceding unsigned comment added by (talk) 14:59, 28 November 2009 (UTC)

  • At least Pentax uses the very same method to recognize camera shaking the lens based systems use too, so your idea about body based system limiting camera use regarding spectral analysis photography is nonsense.
  • Additionally Pentax system corrects movement along all three axix, while the lens based correct only on two axix. When you press the shutter you do induce a little bit of rotation to the body and so far no lens based system corrects this.

-- (talk) 08:46, 2 May 2009 (UTC)

In-lens SLR systems use tilting elements so your information is either imprecise or out of date. —Preceding unsigned comment added by (talk) 21:13, 15 February 2010 (UTC)

A few points:

  • Thrity-five millimetre lenses, notably the Leica M lenses, cover 24x36. If you have a moving sensor, the corners of the sensor must move into the vignetted edge of the lens's image circle. So IS would be feasible for a cropped sensor, but not in a full frame sensor. For a full frame sensor, 35mm lenses would have to be designed with a larger image circle, so the sensor could move, or the IS would have to be built into the lenses, along with the larger image circle. Ie, you could shoehorn a medium-format lens onto your FF DSLR with a moving sensor, but beyond that, don't bother.
What about the Sony A850/900 with a FF sensor and image stabilisation? Seems to work without requiring FF lenses to be over-engineered. ClassA42 (talk) 05:03, 31 August 2010 (UTC)
  • The top manufacturers also tune their IS systems per lens, as they have access to the design specs. This gives these systems both speed and accuracy. —Preceding unsigned comment added by (talk) 00:11, 10 September 2009 (UTC)

What is the validity of the claim that the AF system is not receiving a stabilised image and hence AF function is compromised? Perhaps the sampling frequency is high enough so that shake doesn't play a role? This alleged disadvantage of sensor-based systems seems to be nothing more than a claim without a corresponding reference. Why would the shake be more problematic in "bad light"? If image shake is a problem for AF, should it not be a problem in good light already? A reference supporting the claim should be added. I removed the section for now. Here it is for reference:

"Another disadvantage of moving the sensor instead of the lens is that only the main imaging sensor is moved, but the autofocus sensor is not moved. This means that camera shake can lower the performance of the autofocus system in bad light. This is an issue only with DSLRs which have a dedicated phase-detection autofocus sensor, not an issue with smaller cameras which use the main sensor for contrast-detection autofocus."

ClassA42 (talk) 05:03, 31 August 2010 (UTC)

Logical absurdity of "times less than" and similar phrasing[edit]

In recent years, it has become commonplace to use phrases such as "times less than", or "times smaller than", et cetera, to describe something smaller than a reference magnitude. The word "times" is a multiple of the original magnitude, so that when one does the arithmetic, one obtains a negative magnitude, several times as large as the original, which is nearly always an absurdity.

Instead of saying "one-fifth the size of", one is likely to see "five times less than". Let's say the original dimension is 100mm, and the new dimension is 20mm. Five times 100 mm is 500 mm, so five times less is 100 - 500, which is -400(mm). We're getting into a strange realm, indeed, suggesting the smile of the Cheshire Cat (Lewis Carroll). (For fun -- Is that smile a negative quantity in the alpha channel?)

One might conclude that "five times less than" is a misguided attempt to write at second-grade (US schools) level, assuming that "one-fifth" is either too clumsy, or beyond the understanding of most readers. The USA has quite-serious problems in elementary and secondary education[1], but I don't like to pander to such weaknesses. [1] Charles J. Sykes, _Dumbing Down Our Kids_, St. Martin's Press, New York 1995 ISBN 0-312-13474-6

In this article, I maintained the numerical factors greater than one by elaborating the text somewhat; perhaps I could have been more concise. (Fairly sure that one will note the usage of such expressions as 1/1.5 , 1/1.6, etc. when referring to vidicon equivalents in video cameras.)

See idiom; logical or not, this one appears in thousands of books in English. Dicklyon (talk) 18:39, 22 November 2009 (UTC)

"Times greater than"[edit]

The situation regarding "times greater than" is much less distressing. A three-meter wide door is two times (two meters) wider than a one-meter-wide door; "wider" describes the dimension added to the original, not the multiple of the original. It is three times as wide as the original, of course.

In elementary mathematical terms, we are comparing [n] (times as...) versus [n-1] (times greater than); in this case, 3 vs. 2. As [n] increases, the difference between the two becomes less important in everyday (but not engineering!) terms.

Nikevich (talk) 16:18, 21 November 2009 (UTC)

sony "Optical Steadyshot" is digital[edit]

[2]"Optical SteadyShot™ digital image stabilization compensates for camera shake and helps prevent blur by capturing two photos at high shutter speeds and combining them into a single clear image.³ " —Preceding unsigned comment added by (talk) 23:38, 4 January 2010 (UTC)

[3] Super SteadyShot® Optical Image Stabilization System: The Super SteadyShot® Optical Image Stabilization System from Sony reduces blur caused by camera shake and vibration, so your images stay crisp and clear. This optical stabilization system achieves an even higher level of smoothness without the image degradation that can occur with some digital stabilization systems.-- (talk) 02:48, 3 June 2010 (UTC)

Images used in article[edit]

I do not know how to upload a new photo to Wikipedia so I apologize for this, but the image comparison of IS Off/IS On is awfully large and I was able to optimize it without any readily apparent artifacts. The optimized version is available here: (talk) 06:42, 7 March 2010 (UTC)

Not needed, but thanks anyway. I have reduced the reproduction size in the article a bit. Easy to adjust. Hertz1888 (talk) 08:30, 7 March 2010 (UTC)

Video and sensor based stabilization[edit]

"With the increasing popularity of video in DSLRs, it is worth noting that sensor-shift stabilization does not function when recording video. The sensor must lock in place during video recording. Lens-based stabilization systems don't suffer from this drawback." see "The Pentax K-7's body-based image stabilization can be used when recording movies" and —Preceding unsigned comment added by (talk) 16:37, 22 June 2010 (UTC)

Photo request[edit]

We need a picture of the correction mechanism. I found one on YouTube once -- it looks like it uses a pair of voice coils to move a group of lenses. Also, it would be good to show a ray diagram of that lens group to show how it is shifting the image without refocusing it or adding aberration. Resources section of this page shows the IS group of a 200mm prime lens -- it looks like a negative asphere, an air gap, then a concave surface of a positive meniscus cemented to a biconcave element all near the stop after it. I suppose that means if you shift the group down, an on-axis ray will be steered up by the first element, up a little more by the surface after the air space, then refocused... —Ben FrantzDale (talk) 12:29, 11 May 2012 (UTC)

Image Stabilization - other than in photography[edit]

I suspect that there are many types of image stabilization other than in photography. I will be beginning a stub on image stabilization in the color digital laser printing process shortly (my first article!). What other types of image stabilization might exist that are not directly related to photography? 4ces (talk) 04:22, 10 June 2012 (UTC)

Artists drawing in charcoal use 'fixative' (hairspray) to stop it smudging ! Not related to this ...
I think one control hidden away inside an industrial printing-press is too specialised a topic for Wikipedia ?
I only found one reference to it (HP3500) in the Googleverse
Image stabilization control
This controls the variation of the image density caused by an environmental change or deterioration of the photosensitive drum, toner, etc.
-- (talk) 09:47, 7 March 2013 (UTC)

I've toned-down the anti-digital-stabilization rhetoric[edit]

I've toned-down the anti-digital-stabilization rhetoric and deleted this as unclear:

  • The main disadvantage with this solution is that motion blur from shake under medium to low light conditions will be impossible to correct by the post processing images stabilization system, so in real life this will only work well for short exposure times (daylight).

(was that originally referring to the video technique 2 paras above ?) and added

  • Others are now also using digital signal processing (DSP) to reduce blur, for example by sub-dividing the exposure into several shorter exposures in rapid succession, discarding blurred ones, re-aligning the sharpest sub-exposures and adding them back together again.

with citation from Sony spec. "reduced subject blur by combining six successive frames into a single image with less noise"

Feel free to add other links - I'm sure Sony aren't the only ones to find a way.

There's also a mode to reduce subject movemement, as well, by selecting the sharpest sub-exposure for each region of the image, like Focus stacking, but it's harder to explain.

-- (talk) 06:36, 24 June 2012 (UTC)

Lens assembly based stabilization[edit]

Nokia Lumia 920 has an optical image stabilization mechanism that I think is not covered in the article. It is based on moving the entire optical instead of a single lens. Should this be added to the article? And if so, should it be a new subsection in the Optical image stabilization, or under the Lens-based?

Nokia's white paper on the subject Fintux (talk) 19:34, 31 January 2013 (UTC)

See also Sony's BOSS - 'Balanced Optical Steady Shot'
Shouldn't that page be merged here ?
Video is not that different - Sony SS is also used in stills ?
-- (talk) 09:36, 7 March 2013 (UTC)

Adverse effect on Bokeh ?[edit]

There are several people on internet forums claiming IS harms Bokeh .
I doubt it !
I've started a section on Talk:Bokeh#Image_Stabilization_and_Bokeh. -- (talk) 09:51, 7 March 2013 (UTC)