Discussion in 'Open Discussion' started by drd1135, Aug 30, 2016.
Stabilisation is good…but only up to a point.
Interesting indeed. It would be nice to see more work as a proof of those theories.
Unsurprising. IS gives us 'good enough'. If you're after technical perfection, dampened shutter (electronic, mirror lock up, etc) plus tripod is required.
Had a quick read but just got up,its early here in the UK,a couple of points i have found you need to do your own tests as to what works and what doesn't,i had a Tamron 150-600 on a Nikon D7100 and the OIS worked great on a tripod,it also worked well hand held all the way to 1/4000th.(i was looking for negative impact on IQ)
The other is a possibly more simple question by how much does IS effect lock on speed with AF.
I have had some very bad experience with Fujifilm's OIS, sometimes it really ruins the IQ at higher shutter speeds. But I have never experienced such problems with Micro 4/3 gear.
In no way can you achieve one hundred percent certainty about something like this when handholding a camera - you can never predict the intensity of micro movements. Stabilization has always been an approximate game, don't we all know this? That said, it's gotten very good.
oh and um...
During a recent trip to Cambodia and Laos my wife and I took up the opportunity to travel on the Bamboo Railway as it is soon to vanish due to "progress and upgrades". This system uses a set of wheels on a bamboo platform with a petrol motor which can all be taken apart in a few minutes. Obviously the tracks these run on were from the colonial days of the French and are warped and out of alignment, but still used by tourists and locals.
We were on the railway for about an hour which was great fun. I used the Pen-F with IS on the video and I didn't expect much but was astounded how the camera stabilised the motion of the vehicle into a virtually smooth ride. A really severe test that as far as I'm concerned passed with flying colours.
I've noticed strange results sometimes with IBIS. Most common is slight double imaging over some parts of the image. It's not a big deal and it only happens occasionally. Overall I think IBIS is a massive advantage.
In reality, image stabilisation works just fine for the majority of users.
Given the computational power of these cameras, it would be easy to have various options available like automatically turning off the IS as high speeds. Of course, that might clutter the Olympus menu system.
And it she's nothing to help subject movement
It might just be the difference in how you hold the lens & camera as to how well the IS compensation performs.
Olys do have an SIS Auto setting... I have assumed that's what it does, but I don't think it's documented.
For me the limits of IS are more practical than technical. Beyond saving a couple stops of shutter speed on a telephoto, so I can use 1/80 instead of 1/400, I don't see many practical applications. Anything living or moving will want relatively high shutter speeds. Anything that demands long exposure (nightscapes, water, etc) will want longer shutter speeds and a tripod. So you have this narrow window of scenarios that can tolerate something like 1/2 - 1/20 shutter speeds.
When I can get relatively sharp images hand held at 1/15 to 1/2 second, IS works! It also helps with composition and ensuring I'm focused sharply on the correct subject matter in the frame when using long lenses (150mm up to 420mm).
I think there are a number of flaws in his paper.
First, he says "There is one fundamental problem here: the camera determines the necessary compensatory motion with some degree of delay, because you need to establish a pattern to the motion before you can compensate for it." Implicitly here is is saying that vibration is periodic. i.e., has a "pattern" like a sound wave or a radio wave. At best, you can say he leaves this unproven. More likely, I think it is factually wrong.
The CIPA specification (CD-X011) for IS testing is silent on this topic. The words "random" and "periodic" do not appear in the document. It does talk about "frequency" when discussing how they instrumented actual cameras to determine what their test waveforms should be but I don't think it implies that the motions are periodic in the way Ming implies. More likely, they are referring to how often the camera moves or, possibly, frequencies as determined from a Fourier transform of the motion.
Incidentally, non-periodic motion can still be compensated. A simple example would be a passenger train that is starting to roll with a boy running along side talking to his girlfriend through a window. A camera can sense movement like this and compensate with an equal and opposite motion of the sensor or the image on the sensor. I have what is to me an unbelievably sharp picture of sea lions on a rock shot with a GX8 and PL 100-400mm where the Zodiac inflatable boat was pitching so much that I couldn't even frame the composition. And that was for sure non-periodic motion.
Later, Ming says: "There’s one more limitation with the nature of stabilisation: there’s a limit to the resolution over which the compensating elements can move; the heavier the parts being moved, the more difficult it is to move them precisely in small increments." This is flatly wrong. Machines move seriously heavy hardware precisely in small increments all the time. I think he is conflating this with Newton's second law: Force = Mass x Acceleration. More mass => more force. Faster movement => more force. In the limit Newton says that the correction motion cannot happen instantaneously because it would require an infinite accelerating force to produce such infinitely fast motion. Said another way, the corrective motion is doomed by Newton to be behind the game. How far behind? The manufacturers are not telling.
Another interesting thing in the CIPA specification is this statement: "Camera shake has components in six directions ... However ..., yaw and pitch are dominant and the other axes can be practically negligible." My next statement will draw violent objections, I know: The CIPA results would say that all of the multi-axis stabilization systems that are advertised are simply marketing department talk and do not in fact improve real-world stability. A corollary to this is that since the CIPA spec is the only standard testing spec, and the CIPA spec uses only two-axis motion, then there is no standard way to measure and compare the performance of systems claiming to stabilize in more than two axes.
What kind of images are you taking at those shutter speeds? What are the subjects?
Any kind of ordinary landscape or urban shooting around dusk is a pretty useful situation. The light gets nice around then, but if I'm out hiking (or biking home from work), I'm certainly not going to have a tripod with me.
So yeah, between stabilizing longer lenses and broadly improving the quality envelope I can get in general use photography, IBIS is something I ideally want on any new camera...
Air Conditioning Ice Room, Vancouver Orpheum Theatre by Graham Moore, on Flickr
Orpheum Theatre, Vancouver by Graham Moore, on Flickr
Orpheum Theatre Lobby by Graham Moore, on Flickr
Orpheum Theatre Original Round Door Safe by Graham Moore, on Flickr
I'm usually comfortable enough with my camera's performance to let the ISO float within a reasonable range, but when I have the presence of mind to work at base ISO for better quality, it's nice knowing I can get predictably sharp results around 1/13s, even with the GX7.
I'm using water as an example, too, because I don't think the really long exposure cotton-candy look is always preferable. Moderately slow shutter speeds in that 1/4 to 1/20s range still communicate some of the dynamic of rushing water, which I think is visually interesting.
Separate names with a comma.