Most recent cameras claim to have a 5-axis IBIS (x, y, rotate, pitch, and yaw).
It has been argued that the sensors cannot pitch and yaw as that would move the sensor out of the image plane. @thom writes in his Z books that only lens VR can pitch and yaw.
Is 5-axis IBIS only a marketing ploy, or are there sensors that can pitch and yaw?
I believe OM/Olympus have sensors with pitch and yaw, nut Nikon Z has only lateral sensor movement, which is coupled to yaw in some lenses.
Olympus' and Nikon's marketing material describes their IBIS as a 5-axis movement. Both can combine lens VR with IBIS to get better results.
Therefore I am skeptical whether Olympus IBIS can do pitch and yaw. Wouldn't that cause the edges to blur?
i dont think it moves that much. and the auto focus if set only fires when the sensor is stable.
I guess these movements are
A) a half magnitude lower than "sideways smearing), so there is not that much movement /off plane/defocus happening
B) mayby with ratio of 50-66%, of full correction, you still get more hitrate than with no stabilisation at all, and so it is done and marketed. Halfway defocus is better than full smearline.
On Pentax forum, it was observed that a yaw and pitch translates to x/y sensor movements. A 5-axis IBIS may mean that yaw and pitch are detected, but are corrected by x/y movements.
Consider that a very slight tilt on the lens mount or on an adapter will already deteriorate IQ.
There's definitely some sort of warpy sh!t going on at wide angles in video, I've heard it mentioned that IBIS is responsible for that, but no real concrete evidence that this is the true reason for it.
Anyway, I shoot a lot of wide/ultrawide still shots myself, my last two 2023 camera body purchases have been sans IBIS, I know this is only anecdotal, but they look to take sharper shots to me.
AFAIK the corner issue with wide angles in video and IBIS is caused by the sensor not being centered during recording but moving in the X/Y direction.
Right. This has been discussed many times before. The sensor has three degrees of freedom only: it can shift along its plane, and rotate there. Pitch and yaw are detected and partially corrected but full correction is impossible. One can probably argue that what remains uncorrected is “small.”
Can you point me to the relevant discussions. Somehow I skipped it.
Here, for example: www.dpreview.com/forums/thread/4700392 .
Could this be why some cameras seem to have a lot of lenses with less sharpness at the edge, because the sensor is not quite aligned correctly?
I have seen something like this with my X-T5. Take a picture - go pixel peeping and the edge seems slightly less sharp. Take another picture, and it's perfect. But nothing I can prove (I'd need to be much more methodical - and to be honest, it's not that bad).
Alan
That can be due to the lens geometry and subject positioning and depth of field combination
I remember some discussions about making curved sensors to mitigate that. It was fun. 🤭
Expectations from IBIS are more reasonable when you accept the fact that it mostly corrects gross camera shake and will likely add fine jitter, even when working as designed. The CIPA standard for stabilization that most manufacturers target is one for very low image magnification, not for looking at an image zoomed in on a monitor so that the image is one to two meters wide, as people tend to do when inspecting stability. People who need to crop hard or display large to show fine details still need to rely on higher shutter speeds or mechanical lens/body stabilization instead of IBIS.
Many lenses have different magnification from center to corner, and such a lens can not possibly have pitch and yaw corrected perfectly by IBIS no matter how good the detection and attempt at correction. I would only expect "roll" and "X/Y" corrections to get along well with such a lens, and with pitch/yaw IBIS correction, you can not expect very tight correction both in the corners and in the center at the same time.
Many people are very unhappy with Canon's decision to yoke added "roll" and "X/Y" correction together with OIS, because many people only need "pitch/yaw" (narrow angles of view and subjects that are many focal lengths away from the camera, like many wildlife or sports situations). With Canon's IBIS R bodies, you can't turn off added IBIS without also turning off OIS. I went from using the Canon 90D to the R5 for a while for bird photography, and I found that even though the R5 has half the pixel density of the 90D, the pixel-level stability was better in the 90D, in terms of consistency, and the added IBIS was doing more harm than good in my typical shutter speed range. "X/Y" correction is mainly useful with high magnification that comes from proximity, rather than from focal length, and if it isn't needed, it can only cause errors..
Thanks!
The sensor doesn't have to pitch and raw to correct the pitch and yaw. In fact, the extra two axes in '5-axis' aren't pitch and yaw, they are X and Y - and Thom has, I'm afraid, got the wrong end of the stick.
The axes don't refer to the axes in which the sensor moves, but the axes in which movement can be detected and corrected. With IBIS the sensor only moves x, y and rotation. Let's take classical 2-axis IS first.
There are rotation sensors (called 'gyros') in X and Y direction. These detect rotational movement, which is the most common and destructive. In lens IS the rotational movement is corrected directly. With IBIS a rotation of x degrees around the optical axis causes the image to move at the focal plane by tan x * FL (focal length). This is why IBIS needs to know the FL of the lens to work. Once the displacement at the focal plane is calculated, the sensor is simply moved the required amount to correct it.
3-axis adds a gyro in the Z direction, around the lens axis. To correct it the sensor is rotated to correct. In-lens IS can't correct rotation.
5-axis adds linear movement sensors (accelerometers) in the x and y direction. Linear movements are more benign than pitch and yaw for most work. However the image displacement that they cause is inversely proportional to subject distance, so they can cause issues when the subject is close. Correction needs knowing the subject distance, approximated by focus distance. In-lens IS could make this correction, but so far as I know, none does.
Very small. Plus, it can be argued that in-lens correction works by intentionally decentring the lens and will introduce aberrations.
The real limitation of IBIS for pitch and yaw is how far the sensor has to move with a long lens - which is why lens stabilisation is just about universal for long lenses.
There are six degrees of freedom, as illustrated by the movements of a ship:
Three of these movements are rotations, which corresponds to the movements of a camera on a fixed tripod whose position is being adjusted by moving the tripod head: tilt, pan and rotation between landscape and portrait mode.
The other three of these movements are translations, which correspond to bodily moving the camera through space while it remains pointing in the same direction (as on a gimbal, for example) - fore/aft, left/right, and up/down.
Camera-based stabilization, be it in the lens, via a moving sensor, or by a gimbal, can only correct for rotations. It cannot correct for translations, which is why so many people end up frustrated with their attempts to shoot video while walking - although the camera is stabilized, the up/down bobbing that happens while you walk is often plainly visible. It's also why it's difficult for stabilization to handle very close subjects like macro photographs - at close distances even small translational movements of the camera appear as fairly large excursions of the image in the frame.
Full stabilization in translation basically requires a tripod or similar fixed mount. For video, you often want to stabilize two translation axes while moving in a third, which requires more specialized equipment such as sliders, dolly tracks, cranes, or a full steadicam rig.
