Image Stabilization experiment just for fun

Experiment:

The IS capabilities of newer cameras is nothing short of mind blowing. Just using the old standby rule, without any IS, you should shoot with a shutter speed of at least 1/(focal length) to avoid camera shake-induced blur when shooting hand held. Based on that guideline, with my 24-105mm RF lens, shooting at 105mm, sans IS, I should be able to get good sharpness handheld with any shutter speed above 1/105th. So, for this experiment, I'm going to shoot a static subject at 1/125th handheld (trying to stay as fixed as possible, hold breath, elbows in, etc).

Now, according the the marketing propaganda, the combination of the R5 IBIS and lens IS of the RF 24-105L should yield 8 stops of image stabilization. This means, if I can get a sharp image at 1/125th, I should be able to get a similarly sharp image at 8 stops slower shutter speed, or 2 seconds.

I shot the below series, first at 1/125th. Then I shot 8 stops slower at 2 seconds. Note I was about 8 feet from subject at 105mm, hand held and not braced against anything. For Each shutter speed I'm showing two images - one is full size (scaled to allow upload of course) and the second image in zoomed in about 500%. The full size image doesn't look too bad compared to the 1/125th, but when I zoom in to 500% you can see some shake blur, though not terrible (I probably could have done better had I braced against something like a wall.

The next two images were shot at 7-stops slower or 1 second. Then the last two images are 6-stops slower. At this shutter speed (.5 sec), with IBIS and Lens IS on, I feel like the sharpness is pretty darned close to shooting at 1/125th.

Conclusion: I'm impressed with the IBIS/IS capabilities of this camera / lens combo. I also ran the 2 second exposure through Topaz sharpen, (see last two images) and the results were amazing (I think).


Shutter speed 1/125th Not cropped in.




1/125th cropped in about 500%




2 second shutter speed. Not terrible in my opinion for completely hand held (could probably do better with practice and better technique)




2 second exposure cropped in. Here you can see pretty clearly some camera shake. Might be able to do better with practice.




1 second (7 stops slower than 1/125th). Not perfect but better.





1 second exposure cropped in 500%. The camera shake blur is not as noticeable.




1/2 second exposure. This is still 6 full stops slower than 1/125th and I'm pretty happy with this result.




1/2 second exposure cropped in 500%. This is pretty amazing to me for such a slow shutter and hand held by an old guy!



Ok, Now, I took the 2 second exposure and ran it through Topaz AI Sharpen. Pretty amazed at the result:


This is a hand held 2 second exposure after running it through Topaz AI sharpen. Maybe I'm just easily impressed, but I was impressed.




The same 2 second exposure first run through Topaz and then cropped.

I agree, pretty darn amazing!

Rudy

Impressive results for sure.

Im also pretty sure that the listed rating of 8 stops of stabilization applies at shorter focal lengths, where the IBIS makes more of a difference. As you increase the FL, the effect is diminished. Meaning almost getting 8 full stops at the long end is even more impressive.

Gam3r01 said:

Impressive results for sure.

Im also pretty sure that the listed rating of 8 stops of stabilization applies at shorter focal lengths, where the IBIS makes more of a difference. As you increase the FL, the effect is diminished. Meaning almost getting 8 full stops at the long end is even more impressive.

Click to expand...

You're probably right about that. I felt like I got so-so results at 8 stops, ok results at 7 stops and pretty amazing results at 6 stops. It is true that any movement (like from my beating heart) gets magnified the longer the zoom is.

Basil Fawlty said:

Gam3r01 said:

Impressive results for sure.

Im also pretty sure that the listed rating of 8 stops of stabilization applies at shorter focal lengths, where the IBIS makes more of a difference. As you increase the FL, the effect is diminished. Meaning almost getting 8 full stops at the long end is even more impressive.

Click to expand...

You're probably right about that. I felt like I got so-so results at 8 stops, ok results at 7 stops and pretty amazing results at 6 stops. It is true that any movement (like from my beating heart) gets magnified the longer the zoom is.

Click to expand...

That's an interesting test. I've used that same system for improving a few family photos where I had a little blur due to camera shake when the shutter speed dropped indoors.

It may not always be perfect but it's great to see an improvement

Interesting experiment, with good controls. Looks like 6 stops improvement is the magic number for your setup. if you had to go lower, possibly shooting 5 or more images and then choosing the sharpest one, you could get a stop or two more.

For me IBIS is kind of a moot point for still images, as I shoot moving trains in low light and the shutter speeds even with wide angle lenses for relatively slow-moving freights still have to be around 1/160 - 1/200 second. For me fast, sharp f1.4 primes on my APS-C camera are the only solution, coupled with higher ISOs and DxO Deep prime - lens IS and IBIS won't help me. IBIS would be helpful for video in stabilizing the composition of the image without a gimbal.

Thanks for the experiment.

I’ve done similar experiments out in the field, with the RF 24-105 f4 L, and have gotten similar results. In more controlled/supportive environments, 2 seconds seems to be about my limit as well.

Out in the field, in less controlled/supportive environments, I’ve been able to get 1/2 second handheld if I put my R5 in electronic shutter and high speed continuous. If I take a 10-20 image burst, and control my breathing/posture as much as possible, I will get a handful that will match up to my heartbeat and I’ll get a sharp image. Not the best method, but works in a pinch (and if I don’t have a tripod).

I’ll only say this on this forum, (because it will start a war in other places) but I was getting only very slightly better results with my OM-1 and the Oly 8-25 f4 (didn’t have other lenses) I had at the time. And when I say “slightly better,” I mean maybe only 1-3 more Oly frames that were sharp…I still had a fair number with camera shake.

I also did this with the Oly LIVE ND feature. I never got consistent results, but there were a few times where LIVE ND feature improved my “hand-hold-ability” by one more stop than without using LIVE-ND. I imagine this feature is doing something similar (but more complex) to my crude ES high speed burst method with the R5 described above.

(Disclaimer, I’m sure I could get 4+ seconds handheld with a sync-IS Oly/OM lens…like the Oly 12-100 f4 PRO….but there are only 3 of the Oly Sync-IS lenses to my knowledge. I’ve gotten 5 seconds with the Oly 7-14 PRO at 7mm in the past).

Ultimately, by getting very similar results with the R5 as my OM-1 (sans the Sync-IS lenses, like the Oly 12-100mm f4 PRO) persuaded me to shift fully away from the OM-1.

Thank you for this interesting and practical comparison.

Just two remarks to it:

- the 8 stop of IBIS+OIS efficiency is rated with RF 24-105/4 @ 105mm

- the CIPA standard defines sharp image as if the common post card size print is being evaluated from approx 45 cm of viewing distance. So that is pretty loose criteria. Viewing the image on a monitor at 100% magnification is MUCH more strict.

In the real life I am consistently getting some 5 stops of IBIS+OIS efficiency when observing images at 100% magnification.

Larry Rexley said:

Interesting experiment, with good controls. Looks like 6 stops improvement is the magic number for your setup. if you had to go lower, possibly shooting 5 or more images and then choosing the sharpest one, you could get a stop or two more.

For me IBIS is kind of a moot point for still images, as I shoot moving trains in low light and the shutter speeds even with wide angle lenses for relatively slow-moving freights still have to be around 1/160 - 1/200 second. For me fast, sharp f1.4 primes on my APS-C camera are the only solution, coupled with higher ISOs and DxO Deep prime - lens IS and IBIS won't help me. IBIS would be helpful for video in stabilizing the composition of the image without a gimbal.

Click to expand...

Agreed, image stabilization has limitations for moving subjects (not that it's completely useless). For example, I shoot a lot of birds in flight and set my Sigma 150-600 lens to "OS" Mode 2, which corrects for vertical movement but does nothing for horizontal movement. Since I pan horizontally (usually), I don't want the stabilization to try and correct for my panning. My little experiment is pretty much meaningless for anything but still subjects. For this experiment, 6 stops was indeed the sweet spot, but I feel like I could have gotten ok results at 7 stops with a little more practice at my camera holding technique. Even at 8 stops, I got acceptable results by using Topaz AI sharpen.

Here's 6 seconds handheld with a Canon 10D at 135mm. The long exposure was in order to capture the moving throat. This photo was made before I got Topaz Sharpen AI.



This is my longest exposure handheld that is sharp. I was prone on the ground leaning on my elbows.

P.S. Why does Fawlty Towers have only 12 episodes? Seems like it could have gone on for several seasons.

--
Victor Engel

Victor Engel said:

Here's 6 seconds handheld with a Canon 10D at 135mm. The long exposure was in order to capture the moving throat. This photo was made before I got Topaz Sharpen AI.

This is my longest exposure handheld that is sharp. I was prone on the ground leaning on my elbows.

P.S. Why does Fawlty Towers have only 12 episodes? Seems like it could have gone on for several seasons.

Click to expand...

Pretty impressive for hand held!