kfh
lvl.2
Denmark
Offline
|
GRandyB Posted at 5-29 15:38
As you are aware, it only takes a slight breeze to cause wind noise. This is a clip from it being mounted with a 3 suction cup mount on the side of a car.
This is a clip
It's pretty easy to see wind has nothing to do with the unstable footage.
Sorry, but I see (nearly) no jitters at all! I see lot’s of YouTube delivery problems, but…
I’m not trying to tell you, how you have to do your recordings in the following. Just pointing out a few facts, that have helped me get around a few problems in my time.
The Youtube quality is lousy, even on my connection 350/350 megabit full duplex fibre channel (typical YouTube connection speed 48-96 megabit or more). Whether I kick the video onto my big UHD TV (fairly new), show it on my old iPad Air 2 or a newer iPad 10.5 Pro, everything runs as smooth as “a babys bottom”. On my completely new Apple Mac Mini with M1 chip, I can just about make out some, very small, vertical jitters on my 4k Dell monitor (DCI-P3 color space) placed around 50cm/half a yard from my eyes, but then I have to concentrate (and only the latter half of the video has stabilised sufficiently on YouTube - the first half has all kinds of “weird adjustments” going on.
Quality cannot really be judged on YouTube. The rendering quality in my end is abysmal, so bad, that it in itself introduces jittering “outlines” in the top of i.e. the trees. Also ask yourself: How many look at your works at a distance of half a yard, from a highly tuned, DCI P3 pro monitor with 2160p resolution in top tuned colors. Any 1080p 'hobby' video can show tiny jitters periodically in this case. It would be near wonder, if it didn't happen at all at "double size" display!
A lot of people experience all sorts of bad quality - including jitters in any and all directions as well as speed - using YouTube. It even happens to me, but very seldom since the infrastructure to/from my country, and to the distribution point to my connection is really high quality all the way from the source. The video I got from YouTube had these data in the ‘frozen’ video in statistics for nerds (note, to avoid my own shadow, I had to take the image from an angle, and then ‘de-skew’ afterwards):
YouTube statistics for nerds
I noted, that I get 1080p/30fps from YouTube. Is that your original material? Did you perform any treatment - like downsampling from 4k etc. - that involved rendering to final format? At least use 60 (or 50 fps in 90-95% of the world outside the US and other 60Hz mains countries).
Do you have a longer sequence of your original recording, that you could make available for download from OneNote or Google Drive?
When I have really challenging recordings, I max everything out (adjusted for the mains frequency in the zone, I'm at at the moment). In my case, it is 2160p/50 fps, but in the US I would choose 60 fps (in order to minimise any chance of unwanted ‘side effects’ in areas, where mains is 60Hz) with maximum available quality (100 megabit on Pocket 2).
What SDXC card do you use? In theory, you could get away with a lesser quality, but cards certified as V30 (minimum sustained 30MB/sec) will often, if not consistently perform better, than many cards only specified as U3 (30MB/sec, but not necessarily sustained, as required for video - I have a few bad examples from the time before V30 became standard). The standards mess is seen in full and frightening galore: https://en.wikipedia.org/wiki/SD_card
If I see anything awry in POST, I activate stabilisation (DaVinci Resolve 17 (free) or iMovie (free) or FCPX (paid). Where possible, I activate “Tripod” mode ‘rock solid view’, but seldom available on quick moving, low flying taxis driven by former fighter pilots (it seems ;-) in inner city Rome with plenty of cobble stone streets toward the Vatican (used my old Osmo Pocket on vacation in Rome in April 2019 with veeeery good results).
At 100 megabit/sec, you need an absolute minimum of 12.5 megabyte/sec, and cards with higher “umpffhh” will leave room for camera handling overhead. Cards, that are juuust about good enough, may introduce visible artefacts too; sometimes also visible as jitters (a periodic dropped frame or a few missing lines in 30 fps is often far more visible, than a dropped frame in 60 fps). Some jitters - especially the vertical type - are really a kind of rolling shutter resulting in a “jello effect”.
I experienced an especially bad example, when filming a ride through the streets of Singapore handheld Panasonic GX80 camera at 4k/25fps placed in the front on an unsuspended tricycle on a ride with Tricycle Uncle. Certainly more leisurely, than your example, alas… It was (sea?)sickening to look at; images went from stretched to compressed and vice versa depending on the camera moving/angling up or down. Now imagine a sensor, that can read the content of each frame much faster (60fps instead of 30fps), then the “jello effect” becomes more like a jitter, and if the camera can read the complete sensor in “one go”, you have what is called global shutter.
In most cameras, even if exposure (of each line/section?) is only 1/2000 sec - the sensor is read piecemeal from top to bottom. Line after line, taking - let’s say 30 ms in all reading from top to bottom line at 30fps. At 60 fps, the camera cannot use more than max. 16ms (typically a bit less) to read the content of the sensor lines from top to bottom. Result: less jello/jitter, than at 30 fps. In your example: Try 60 fps.
There are filters, that try to counter act the effect in POST, but on a fast moving motor cycle, the problem is, that the top lines of the sensor are read at a larger distance from a far away object, than the bottom line in the same frame. You see that as tilting tree-trunks or telegraph poles etc. in a video from a motor cycle, car, bus or train. Especially on the super fast moving Maglev train from Pudong Airport in Shanghai to the city, when scheduled to run at 430+ kph to please the tourists, nothing outside the windows within a short distance is rendered as vertical, when filming.
Personally I would also experiment with manual focus or locked focus in your case (tend to lessen “jitters” caused by ‘jittery autofocus adjustments’ caused by ‘step motors’ moving lens groups in tiny steps to alter focus - haven’t tried, if focus locking is possible on the Pocket 2 - haven’t felt the need - yet! ;-)
Hope at least some of the many words can be of use to your or any other person working their way through this text ;-)
Regards
|
|