7.9 gyro stab - is lens data required?

I've seen some posts and videos where people are suggesting that for lenses which the camera cannot read metadata from, then you should enter the lens data (e.g. focal length) manually into the slate pages in order to get the best performance out of the new stabilisation feature. Is this true or just inspired guesswork? I do know that some third party stabilisation devices (e.g. SteadXP) do require the lens metadata in order to work optimally, so maybe there is some truth in this?

Other than Capn Hook's post a couple of days back, we're having to go by guesswork, trial and error in order to work out how to get the best out of this new feature. So a definitive answer by a BM staffer about this would be very helpful please.

[CaptainHook]

Yes, it needs a focal length. At the moment if there is none provided in metadata it will assume 18mm.

Can that data be added to metadata in post?

What happen if you use a speedbooster does th x64 or x71 or whatever the factor is need to be included?
My BMPCC4K with x64 speedbooster on a 50mm lens the lens data says 32mm.

Furthermore, why not lens stabilisation + gyro roll stabilisation. Gyro roll stabilisation should not need focal length.

I think that is very difficult to read and store lens stabilisation action, lens contact active or deactivate feature or command iris or focus motors, simple data impulse, send back stabilisation data may far from ability of lenses pin. If not, I will be happy to be wrong.


Inviato dal mio iPhone utilizzando Tapatalk

Yes, OIS + gyro roll stabilization would be awesome.

But I'd would also like to have an option in Resolve within the software stabilizer for setting the strength of the roll stabilization independently of the rest.

Yes, it needs a focal length. At the moment if there is none provided in metadata it will assume 18mm.

Excellent, thanks for confirming. So it seems that to get the best out of it, we should:

• shoot at between 45˚to 90˚shutter angle (assuming 24fps) in order to avoid issues with motion blur artefacts at 180˚
• if using a manual lens, enter the focal length into the slate metadata

This very much mirrors the requirements of the SteadXP device which I tried to use but found too fiddly, plus it required jumping out of Resolve into the SteadXP software to do the fixes. So it fell into disuse for me. Having this feature native to the camera and Resolve is really great.

Suggestion:it might be useful to add the lens focal length to the list of editable options in the Resolve media page so that we could edit the focal length data in post in case it didn't get recorded during the shoot.

Yes, it needs a focal length. At the moment if there is none provided in metadata it will assume 18mm.

When you use the manual entry method using the Lens Data menu of the camera, for a 28mm lens, it's better to write only the number ''28' or ''28mm''?

If you have a full-frame 28mm lens, does it require converting it using the camera crop ratio and type ''44mm''?

Yes, it needs a focal length. At the moment if there is none provided in metadata it will assume 18mm.

If you manually put in lens focal length will that work? For example, using a Cinema Lens that's 35mm and you put that into the lens metadata field of the slate metadata will that help? Or does the lens need to feed the camera that metadata itself?

If the later is the case, then for PL that reinforces my push for /i Technology and LDS in the PL Cinema Lenses.

Yes, it needs a focal length. At the moment if there is none provided in metadata it will assume 18mm.

Is it possible to enter focal length in the metadata in Resolve?

When you use the manual entry method using the Lens Data menu of the camera, for a 28mm lens, it's better to write only the number ''28' or ''28mm''?

Hopefully it parses 28mm to 28 as I always use mm. However easy to switch to 28 if that’s needed.

If you have a full-frame 28mm lens, does it require converting it using the camera crop ratio and type ''44mm''?[/quote]

Gadzooks, man! Maybe we can convert any combination of factors to derive an ‘effective’ focal length of a prime lens, but what are mere mortals going to do?

When CaptainHook mentioned the default of value 18, I’m praying all we must input is the actual lens focal length used for the shot, not the effective focal length.

Now you have me wondering: What do we enter manually for a zoom lens that zooms during a shot? If everything was electronic, the camera could know that we just zoomed from 14mm to 20mm, but even a genius isn’t going to be able to enter a zoomed shot manually ahead of time when shooting live events. In my use case, I may zoom more than once in a clip.

Because all my lenses are manual, does this mean my gyro stabilization is useless if my Tokina Cinema ATX 11-20mm lens is zoomed in and out during a take? Sadly I think I know the answer and it’s not pretty.

Can we have a White Paper or a decent entry in the Resolve Manual and/or the Pocket manual to explain this new feature thoroughly?

Can we have a White Paper or a decent entry in the Resolve Manual and/or the Pocket manual to explain this new feature thoroughly?

Yes please, this is definitely needed to help cut out the guesswork in how to get the best out of this feature. My take so far after much experimenting and advice from Captain Hook is this:

• Turn off OIS in lens and/or camera, whichever is applicable
• Use a high shutter speed, e.g. at 24fps use a 90˚shutter angle - or even 45˚ if close to the target - to avoid motion artefacts
• For manual lenses, enter the focal length into the slate metadata otherwise it will default to 18mm (my Oly 12-100 does this automatically for the full zoom range)
• Keep the camera movement relative to the target as slow as possible to avoid motion artefacts
• Alternativley, keep as much distance from the target as possible to avoid motion artefacts
• Shooting close while moving past targets with large vertical elements which fill the shot (doors, windows, etc.) can introduce weird bending artefacts in the verticals (maybe related to rolling shutter?)
• The smaller the focal length, the better the outcome
• It's very effective at removing hand jitter, less effective for removing walking jitter
• The more you can minimise hand jitter and shake, the less the crop

So while it can be very effective in some scenarios as Grant Petty's demo video shows, we could definitely do with a 'best practices' guide. The problem currently is that without knowing best practices, you really can't tell in advance whether the shot will gyro stabilise well in post or not be usable.

Good summary, but I have never shot a subject at 45 degree shutter except for a test. It would seem that 45 degrees would make any subject movement look off. Or do you think a “180 degree shutter angle is best for natural-looking movement” might be an urban myth passed down through the generations without question?

180 degrees has always been considered "normal", to satisfy cinematic expectations of motion blur. If you want to see an example of 45 degrees used for effect, look at the Omaha beach scenes in 'Saving Private Ryan'. There's a very distinct staccato or even strobe effect to it, quite unlike "normal" dreamy cinematic motion.

But you can shoot as low as 11 degrees and, depending on the distance and speed of the movement, get away with it. Up close, it could be disorientating.

Haven't played with gyro, but I think people are adding a bit of motion blur to shots recorded at 45, so it looks "normal" in the end.....

Good summary, but I have never shot a subject at 45 degree shutter except for a test. It would seem that 45 degrees would make any subject movement look off. Or do you think a “180 degree shutter angle is best for natural-looking movement” might be an urban myth passed down through the generations without question?

No, not at all! 180˚is the default for good reason! 45˚is pretty horrible to look at, you'd have to add motion blur in post to calm it down, at which point it all starts getting a bit more trouble than it's worth. Maybe that's why the feature was launched with so little fanfare. Getting it stable and natural-looking without any artefacts seems to need the alignment of many factors. It's interesting and fun to play with but so far as I can see it's no substitute yet for a steadycam or gimbal in my admittedly limited experience/expertise. Maybe further tweaks by the dev team will widen its usable range over time.

Thanks, folks. Will be interesting to hear how people test pushing the borders of what is acceptable. In the example from Grant, there’s modest movement of the camera and the subject. With slower movements of subjects, the fast shutter may not be an issue, but people are eventually going to determine the motion limits for a given shutter angle. BMD’s recommendation is helpful. Subject motion is one thing, but unintended camera motion while a subject is essentially stationary is another.

Gadzooks, man! Maybe we can convert any combination of factors to derive an ‘effective’ focal length of a prime lens, but what are mere mortals going to do?

When CaptainHook mentioned the default of value 18, I’m praying all we must input is the actual lens focal length used for the shot, not the effective focal length.

I will wait for the answer from the person I have quoted, which has the insider info.

Note that although it doesn't work with lens stabilization, it does work with gimbal stabilization. With careful movement and a gimbal, you should be able to get extremely smooth "gliding" motion.

question remains: what focal lens to type in when using speedboosters?

I do not see how that would help anything,it would be just gimbal doing exactly the same than gyro stabilisation, if the gimbal does its job properly, the camera gyro has nothing left to do

A gimbal does not correct motion on the so-called "z axis". Namely, up and down bobbing. Gyro correction, pending more information, may be able to do something about it.

Shooting a 45 degree shutter is not a casually made decision. Unless there were another valid reason you'd really only want to make it knowing that gyro stabilization was going to be involved.

Has anyone experimented with double frame rate at 360 degree shutter? Once the stabilization is complete one can run the footage through ffmpeg and extract every other frame to recover the 24p@180 cadence. Or get really crazy and shoot at 72@360 (resolution permitting) and reduce down to 24P@120 which wouldn't be horrible as far as motion blur is concerned, only now Resolve is working with 3x the information.

Good Luck

… shoot at 72@360 (resolution permitting) and reduce down to 24P@120 which wouldn't be horrible as far as motion blur is concerned, only now Resolve is working with 3x the information…

The camera will write 3x the gyro data to the file. Is this data embedded with the intraframe video data or a separate file? If the gyro data is applied to the frame by the camera, then Resolve will only have a third of the gyro data at 24fps. Clarify?

BMD gyro-corrected samples I've seen -- youtube, FWIW -- do appear to correct for z-axis motion. I don't think anyone expects a "perfect" result. It might be better to wait for actual results, or an official word, than speculate further on what BMD's implementation might or might not accomplish, at least where expectations don't violate the laws of physics.

Didn't watch the video, but it's no secret that Z-axis correction on a gimbal requires an arm or the kind of arrangement DJI is using on the 4D. If there's some other magic, it has yet to hit market....

… shoot at 72@360 (resolution permitting) and reduce down to 24P@120 which wouldn't be horrible as far as motion blur is concerned, only now Resolve is working with 3x the information…

The camera will write 3x the gyro data to the file. Is this data embedded with the intraframe video data or a separate file? If the gyro data is applied to the frame by the camera, then Resolve will only have a third of the gyro data at 24fps. Clarify?

I have no idea how the data is embedded, it could be 1 step per frame or 1000, the point is that Resolve will have 3x the visual information which will necessitate less interpolation between frames.

Good Luck

See pages 64-65:

https://documents.blackmagicdesign.com/ ... 6572410000

The BMPCC6K Pro viewfinder has a smooth option which steady hand held movement so some sort of gyro stablisation is already in existance back in 2021 when the camera was first released, just not recorded in the footage.

I am new to his Forum so forgive me if this has been answered.

I have updated my BMPC6K Pro camera to the newest 7.9 firmware, and I have the new Resolve 18, but my clips don't show an option for Gyro stabilisation in the drop-down menu?


Any help with this would be appreciated thanks.

I am new to his Forum so forgive me if this has been answered.

I have updated my BMPC6K Pro camera to the newest 7.9 firmware, and I have the new Resolve 18, but my clips don't show an option for Gyro stabilisation in the drop-down menu?


Any help with this would be appreciated thanks.

Have a read here
viewtopic.php?f=2&t=162827

and here
viewtopic.php?f=2&t=163113

and here
viewtopic.php?f=36&t=162940

and here
viewtopic.php?f=36&t=162758

Hi

I’m also curious to what focal length to input in the meta data when using speed boosters and manual lenses. Example: 35mm lens combined with a speed booster .71

From testing my most logical approach is Focal length 25mm. But I get more aggressive stabilization when input 35mm… which seem unlogic to me. It’s not better, but it crops in more.

Would be nice if BM could give some pointers regarding this.

Hi

I’m also curious to what focal length to input in the meta data when using speed boosters and manual lenses. Example: 35mm lens combined with a speed booster .71

From testing my most logical approach is Focal length 25mm. But I get more aggressive stabilization when input 35mm… which seem unlogic to me. It’s not better, but it crops in more.

Would be nice if BM could give some pointers regarding this.

In my tests I seem to get much better results accounting for the Speedbooster.. While the focal length doesn’t change.. the image circle does.. so it makes sense that accounting for this difference yields less crop..

Effective focal length still changes, because light rays are bent.

The longer the focal length the more same real-world transform affects motion in relative screen space so 35mm will get more screen space stabilization than 25mm.