lines in the blurred part of background

[rglick1]

So I've got these images and the background out of focus parts have lines, anyone seen this before? The camera is a Nikon D850 and the lens is a Nikon 300mm f4. The exposures are labeled on the photos.

[Bjørn Rørslett]

VR on?

[rglick1]

Yes the VR is on, but look at this one. same camera and lens with VR on

[Bjørn Rørslett]

The VR feature can add a treacherous degree of unpredictable double lines on occasion. This is something the makers' praise of vibration reduction *never* mentions.

Shooting in sunshine at fast shutter speeds would normally not need any VR on at all, provided a decent tripod was used.

[richardHaw]

yes. I read somewhere that using VR on shutter speeds faster than 1/200s is detrimental to your picture

it has something to do with the frequency of electronics or something like that. I forgot where I read that, maybe at Thom's site?

I talked to a lens designer and they're happy NOT to incorporate VR for a few reasons, maybe this is one?

by the way, is this the 300/4 PF? doesn't the fresnel lens create artifacts of its own?

[Øivind Tøien]

The VR sampling frequency is 1000Hz (according to a VR article by Thom Hogan), which means that the highest signal frequency that can be sampled is 500 Hz  - 1/500 s according to the Nyquist Theorem. However we do not know which countermeasures Nikon has put in place to avoid that VR goes haywire beyond that shutter speed. In my own experience the VR on the 300 PF works well up to 1/500 sec. Beyond that it would not help keeping the exposure steady but I find that the Sports Mode is a great help for aiming though the viewfinder and keeping focus point on the target; I have hardly seen any detrimental effects up to 1/1000 sec. At 1/1250 sec, I have seen signs of background artifacts, although nothing horrible. at 1/1600 second I would likely want to turn it off. This is all handheld with the D7100 and lens collar mounted, other bodies could perhaps respond differently both technically and with respect to physical interactions.

I do not think I have ever seen bokeh effects that can be ascribed to the phase fresnel lens element in any normal captures, however if one really go for it one can create a nice phase fresnel ghost in IR as in this false color rendition:


NIKON D40X IR-720nm, AFS 300mm f/4 PF VR  @ f/4, 1/400, ISO 100

[rglick1]

I guess I'll do some experimenting with VR and shutter speeds. Thanks for all the help.

[richardHaw]

If I get the chance again to talk to the designers, maybe I can show them the pictures here? 

[Øivind Tøien]

If I get the chance again to talk to the designers, maybe I can show them the pictures here? 

It would be very interesting with some background information on this directly from the primary source!

[richardHaw]

we asked before about VR and both designers mentioned if I recall that they're on the fence with regards to the VR because there are trade offs. the answers were very careful at least this is my impression.

[Akira]

Øivind, your image is amazing!

If VR would cause negative effect at faster shutter speed, Nikon should implement the custom setting to set the highest shutter speed limit to deactivate VR automatically in order to avoid such kind of ill effect.

[Øivind Tøien]

Thanks Akira, it took some work to get the right ghosting. 

Customers would likely start complaining if the viewfinder image suddenly got shaky when shutter speed was set above 1/1000 sec, so I am not sure they could take that kind of approach. To keep speculating, some kind of predictive algorithm (extrapolation in time) could perhaps be used when sampling time gets too short.

[Frank Fremerey]

I guess I'll do some experimenting with VR and shutter speeds. Thanks for all the help.

very helpful thread indeed. I shoot a lot with that combo 300pf/D850. Depending on the granularity and distance of the background and the aperture used I witness similar "nervous bokeh" on occasion.

I will now also try to experiment with aperture, VR, exposure time to see whether the bokeh change is visible. Very big and very small structures are no problem also a background that is far away, enough to blur out completely, is good. But a stone beach with medium size stones (or a meadow) and a bird in the foreground cause trouble because the is just not enough blur to get rid of the repeating structure, that interferes with my recorder

[Akira]

Customers would likely start complaining if the viewfinder image suddenly got shaky when shutter speed was set above 1/1000 sec, so I am not sure they could take that kind of approach. To keep speculating, some kind of predictive algorithm (extrapolation in time) could perhaps be used when sampling time gets too short.

Oh, you are right.  I forgot that the image stabilization is not only for reducing camera blur but also for comfortable framing.   

[Ilkka Nissilä]

I'm sure it would be possible to have VR ON during composing/focusing and OFF during the actual exposure, it should be just a matter of programming. Whether Nikon think this is a sensible thing to do is another matter. 

I find that VR SPORT with the 300/4 PF does work nicely also for faster shutter speeds (typically 1/500s to 1/1250s in my case with this lens); I would turn VR OFF at 1/1600s or faster.  Hand-held of this very lightweight tele is much easier in VR SPORT mode than VR OFF. But VR OFF may result in sharper pics (and perhaps fewer artifacts) in some circumstances.