Jump to content
jonpais

Motion Cadence

Recommended Posts

EOSHD Pro Color for Sony cameras EOSHD Pro LOG for Sony CamerasEOSHD C-LOG and Film Profiles for All Canon DSLRs

Rolling shutter artifacts are present, yes, but unavoidable. The settings are posted and were double checked off of a checklist for each cam, which was mentioned already.  The shots can be verified by anyone,  using the method described in the posts above.   Kye, if you look at the pic at the start of the video, all the cams are on the same head, panning at the same rate.  The rate of pan gradually subsides in conjunction with the pendelum. Static shots with no pan are provided as well.

 Yes, the addition, subtraction, and/or the interpolation of frames seems to be the most logical culprit.  We can, however, only speculate about it's cause without further tests.   ( for example, what if each frame was filled with even more ever changing detail, like foliage, or running water? Would there be even worse hiccups as the imaging devices in question would have even more information to process?) 

 

Share this post


Link to post
Share on other sites

This is Sony's demonstration of their global shutter sensors versus rolling shutter sensors (presumably also Sony ones ;) )

 

This one is a comparison between two board cameras with global and rolling shutter sensors.

 

RED's Motion Mount can be quite informing about how this stuff plays with the same camera as it has enables their cameras to have variable ND but also gives them two global shutter modes, which they refer to as Square and Soft. 

So in demos like this - which is a repeatable pan under motion control - you are able to see the effect on the same camera between ND only (i.e. rolling) and the two different types of global shutter.

And from the same user, the effect when filming motion from a locked off camera.

 

Share this post


Link to post
Share on other sites

@jonpais, I recently submitted a broadcast file via the Universal Records system and you have to pass numerous quality checks. One of them, is motion cadence. 

You can't submit a video that has 4:1 cadence, which has 4 progressive frame of video, and every 5th frame is a repeat of the 4th frame. (This is more to do with people using content shot at a lower frame rate (like 24fps), and converting it to 30fps incorrectly). They are preventing the video motion from looking unusual or low quality with this check. 

I'm in NO WAY WHATSOEVER a camera technician or scientist but I'll explain my understanding. Apart form the obvious frame rate / shutter speed / global shutter / codecs influence on motion cadence, all cameras capture and interpret frames differently. Some cameras may capture in a "false progressive" (interlaced frames which mimics progressive video), frame capture that is made up of some progressive frames or some interlaced, or cameras that have a truly progressive frame capture. 

Now there's motion blur. Motion cadence on modern TV's is destroyed because they are defaulted to a special motion setting (like TruMotion), which manufactures say makes the images look clearer and smoother. The setting when turned "On" makes the image feel like there's a bunch of new frames thrown in, eradicating motion blur and giving us the dreaded "soap opera effect". 

With all of this considered, an attractive, cinematic motion cadence is the ability to capture full progressive frames smoothly with intricate motion blur (as long as your frame rate is half your shutter and if it's global then BONUS!). To do this, you must start with the right codec first (the best of which, to my experience, is uncompressed RAW converted to ProRes).  

It's no coincidence that expensive cinema cameras (and BM cams) have this motion cadence (very high shutter readout, RAW and ProRes, high bit rate) and consumer cameras do not at all (besides the 5D Mk 3 Magic Lantern). 

An external factor I find is the lens. A lens with electronic contacts with I.S and all that jazz, usually overly sharp and contrasty, hurt motion cadence. I feel it's because they heighten the "electronic" factors (sharpening, aliasing, moire, noise, jittery look). Lenses with a smooth rendering, regarded "creamy", very often manual glass and high T-stop really smooths out the motion of the image and hides the electronic factors that make our images digital looking. 

So in a nutshell, for beautiful motion cadence, shoot in RAW at 24p and 1/48s with a very fast rolling shutter /global shutter, with manual "proper" cine lenses and finish in ProRes. Come think of it, I now find my post pointless, as I thought we all knew that anyway!! 

 

Share this post


Link to post
Share on other sites
47 minutes ago, Oliver Daniel said:

@jonpais, I recently submitted a broadcast file via the Universal Records system and you have to pass numerous quality checks. One of them, is motion cadence. 

You can't submit a video that has 4:1 cadence, which has 4 progressive frame of video, and every 5th frame is a repeat of the 4th frame. (This is more to do with people using content shot at a lower frame rate (like 24fps), and converting it to 30fps incorrectly). They are preventing the video motion from looking unusual or low quality with this check. 

I'm in NO WAY WHATSOEVER a camera technician or scientist but I'll explain my understanding. Apart form the obvious frame rate / shutter speed / global shutter / codecs influence on motion cadence, all cameras capture and interpret frames differently. Some cameras may capture in a "false progressive" (interlaced frames which mimics progressive video), frame capture that is made up of some progressive frames or some interlaced, or cameras that have a truly progressive frame capture. 

Now there's motion blur. Motion cadence on modern TV's is destroyed because they are defaulted to a special motion setting (like TruMotion), which manufactures say makes the images look clearer and smoother. The setting when turned "On" makes the image feel like there's a bunch of new frames thrown in, eradicating motion blur and giving us the dreaded "soap opera effect". 

With all of this considered, an attractive, cinematic motion cadence is the ability to capture full progressive frames smoothly with intricate motion blur (as long as your frame rate is half your shutter and if it's global then BONUS!). To do this, you must start with the right codec first (the best of which, to my experience, is uncompressed RAW converted to ProRes).  

It's no coincidence that expensive cinema cameras (and BM cams) have this motion cadence (very high shutter readout, RAW and ProRes, high bit rate) and consumer cameras do not at all (besides the 5D Mk 3 Magic Lantern). 

An external factor I find is the lens. A lens with electronic contacts with I.S and all that jazz, usually overly sharp and contrasty, hurt motion cadence. I feel it's because they heighten the "electronic" factors (sharpening, aliasing, moire, noise, jittery look). Lenses with a smooth rendering, regarded "creamy", very often manual glass and high T-stop really smooths out the motion of the image and hides the electronic factors that make our images digital looking. 

So in a nutshell, for beautiful motion cadence, shoot in RAW at 24p and 1/48s with a very fast rolling shutter /global shutter, with manual "proper" cine lenses and finish in ProRes. Come think of it, I now find my post pointless, as I thought we all knew that anyway!! 

 

1) We're assuming the video is shot at 24fps and delivered in 24fps, not speeded up to 30fps.

2) Can you show proof that some cameras are shooting 'false progressive'?

3) Right, televisions and computer monitors can affect motion.

4) I've watched lots of movies most of which I assume were not shot in RAW and never noticed unusual motion artifacts or weird cadence.

5) Can you provide evidence that lenses with electrical contacts with IS and 'all that jazz' affect motion cadence? 

6) Does Universal Records require you to shoot in RAW with cinema lenses with no electrical contacts? If what you are asserting were common knowledge, as fussy as they are, why wouldn’t they make that a requirement?

Your post isn't pointless, but it's full of conjecture.

Share this post


Link to post
Share on other sites
2 hours ago, jonpais said:

Your post isn't pointless, but it's full of conjecture.

Well yes, I did say I'm not a camera technician or scientist. My post is connecting the dots I see to how the motion feels in an image, to how it was captured. There will be far more qualified people to explain, far better than my ramblings. However I trust what my eyes see, and I see a variation of "motion cadence" between cameras. Definitely. 

 

1) We're assuming the video is shot at 24fps and delivered in 24fps, not speeded up to 30fps.

Yes, because that's what film pioneers have made the standard and for good reason. Go up to 30fps, and the motion blur is lost. Higher? SOAP OPERA. 

2) Can you show proof that some cameras are shooting 'false progressive'?

I know the Canon XL2 did or a model like that back in the day, point being though, progressive isn't made equal. I can see that with my eyes.

3) Right, televisions and computer monitors can affect motion.

Certainly. Have you ever used a broadcast Panasonic monitor? The images through it look like a motion cadence dream! Why? I've no idea!

4) I've watched lots of movies most of which I assume were not shot in RAW and never noticed unusual motion artifacts or weird cadence.

Good. I was just saying that RAW>ProRes as a capture codec, yields the most pleasing results for "motion cadence". 

5) Can you provide evidence that lenses with electrical contacts with IS and 'all that jazz' affect motion cadence? 

Not right now, I've just experienced this with my eyes while shooting videos. 

6) Does Universal Records require you to shoot in RAW with cinema lenses with no electrical contacts? If what you are asserting were common knowledge, as fussy as they are, why wouldn’t they make that a requirement?

No. ProRes (as the final delivery format) is the requirement. As long as the image passes the list of regulations, you're in.  

 

2 hours ago, jonpais said:

Just for the record, I’m not disputing that IBIS can produce undesirable effects. I experienced that today while out on a shoot. But shaky camera movement is much worse!

Well the Universal Records submission had IBIS footage. How incredible! :glasses:

Here's a screenshot from a recent submission, there's loads of stuff that fall under these categories but give you an idea:

 

Screen Shot 2018-04-22 at 20.22.13.png

Share this post


Link to post
Share on other sites

Highlight and shadow rollof is going to have a huge effect on motion blur. Thins that usually looks stuttering is dark objects like lamp posts against a bright background like sky. If you look closely you can see it has 180* shutter but half of it is so faint it's almost invisible. So what you end up with is a lot higher "shutter speed" than what you originally thought. So you would need to overcompensate with say 1/40 or even 1/30 to keep those parts looking smooth. But then other parts that have lower contrast ends up being too smooth, maybe.

Since old film cameras sometimes had a rotating shutter the leading and trailing "edge" would be somewhat faded and not as abrupt which I guess would lead to a smoother motion. Just like the filter thingy mentioned in the thread.

Share this post


Link to post
Share on other sites

@Oliver Daniel, just a number of points:

  • The days in which cameras shot fake progressive video (by firmware-deinterlacing an interlaced camera signal) are long gone. Those are leftovers from SDTV/SD video/CRT times when progressive scan video and tv standards didn't even officially exist.
  • One cannot judge the motion rendering quality of 24p video unless one has a perfect 24p playback chain where (a) the software video player, (b) the graphics card and (c) the monitor actually plays 24p / has a 24Hz refresh rate, as opposed to playing 24p on 60Hz with oddly repeated frames/motion stutter as is the case on 99.9% of all computers. For true 24p playback, the minimum you'll need is a Decklink card (as opposed to a run-of-the-mill computer graphics card), an NLE that can play back through the Decklink card and a monitor that accepts a 24p signal and natively plays it back at 24 or 48Hz. (Normally, such monitors cost several $10,000, but the cheapest option is ca. $1500 Eizo CG.) 
Quote

Have you ever used a broadcast Panasonic monitor? The images through it look like a motion cadence dream! Why? I've no idea!

My last bullet point hopefully explains this.

So I'm pretty confident that 99-100% of "motion cadence" discussion here and elsewhere is really about (a) rolling shutter and (b) non-native 24p playback. 

Share this post


Link to post
Share on other sites

The monitor / TV refresh rate does seem like it would be a huge factor here.  You mention a 48Hz monitor; seems like a 120Hz monitor would also work (divides into 24 equally).

Why does the video card have to be a Decklink though?  Many video games provide options to lock the framerates at particular values and the consumer GPU's respect those settings.  It seems like all that's needed on the video card side is the playback software outputting the video signal at the correct playback framerate to the card.  Or if running in a window, then the desktop running at the correct framerate (which in Windows seems to be tied to the display's refresh).  My display (LG 4K 31MU97) runs at 60Hz over DisplayPort but gets capped to 24Hz in 4K when connected over HDMI.  I'm going to do some comparisons and see if I can notice smoother playback on 24p content when Windows and my video card (GTX 970) are capped at 24Hz.

Share this post


Link to post
Share on other sites
2 hours ago, tellure said:

Many video games provide options to lock the framerates at particular values and the consumer GPU's respect those settings.  It seems like all that's needed on the video card side is the playback software outputting the video signal at the correct playback framerate to the card.  Or if running in a window, then the desktop running at the correct framerate (which in Windows seems to be tied to the display's refresh).  My display (LG 4K 31MU97) runs at 60Hz over DisplayPort but gets capped to 24Hz in 4K when connected over HDMI.

The problem is that GPU frame rate settings for games only work in DirectX overlay mode. (The same is true for 10bit output, btw.) If you connect a computer monitor to a graphics card, it will sync at 60Hz (or 30Hz if the GPU cannot deliver a higher refresh rate in a high resolution), and you can't do anything about it. AFAIK video applications always use the resolutions and refresh rates configured for the desktop interface unless you use a dedicated video output card like a Decklink. 

Share this post


Link to post
Share on other sites

 But.......

I use a decklink, a 24 Hz monitor, and resolve studio set as specified. There appears to be jumps in motion both ahead, and then back, most noticable on the s8 and somewhat on the 1dc.  

 

 

Share this post


Link to post
Share on other sites
2 hours ago, sam said:

 But.......

I use a decklink, a 24 Hz monitor, and resolve studio set as specified. There appears to be jumps in motion both ahead, and then back, most noticable on the s8 and somewhat on the 1dc.  

 

 

Then your computer might be too slow and dropping frames during playback.

Share this post


Link to post
Share on other sites

Get a proper application to play video files in (for example MPC-HC) and a good renderer (like madVR). If you still have problems getting smooth playback you are doing something wrong.

You do not need a special graphics card or decklink to get proper framerates. You just need one powerful enough, which varies depending on what you do.

Share this post


Link to post
Share on other sites
2 hours ago, sam said:

 But.......

I use a decklink, a 24 Hz monitor, and resolve studio set as specified. There appears to be jumps in motion both ahead, and then back, most noticable on the s8 and somewhat on the 1dc.  

That indicates a too slow hard drive (for the 1dc material, that stuff needs speed)  that results in dropped frames. No idea about the s8 but jumps in motion are usually dropped frames.

Share this post


Link to post
Share on other sites
21 minutes ago, no_connection said:

You do not need a special graphics card or decklink to get proper framerates. You just need one powerful enough, which varies depending on what you do.

Then please tell me which mainstream graphics card can send the normal desktop via a 24p signal to a computer display. (Most computer displays do not even sync with 24p).

We're running in circles here.

8 hours ago, tellure said:

The monitor / TV refresh rate does seem like it would be a huge factor here.  You mention a 48Hz monitor; seems like a 120Hz monitor would also work (divides into 24 equally).

In theory, yes. In practice, these are high-speed gaming displays with TN panels whose color reproduction is not good enough for video. 120Hz isn't really available any longer, and only few of them have a 240 Hz refresh rate: https://www.blurbusters.com/faq/120hz-monitors/

 

 

Share this post


Link to post
Share on other sites

If your monitor can't display 24p and goes black does not mean the graphics card can't output it, but it usually means it won't output it unless you force it to, since the monitor reports that it can't display it anyway.
You can make custom resolution in nVidia control panel and I think AMD have same option. I even think Intel allows making it.

Share this post


Link to post
Share on other sites

Thanks to those who mentioned slow drive...etc...not the kind of hiccup being discussed... Seperate Ssd's for each: os/app, scratch, media, and renders.  8Gb video ram.  Cantsin is correct about a lot of this. But...I'll try again.   Straight out of cam on the timeline in resolve there appears to be drift. And not the kind from 23.976 vs 24. This is visible in the file I posted. Nothing whatsoever to do with stutteringng playback from an inadequate system.

 Follow?

As mentioned previously, Its most likely from whatever method of interpolation/compression each cam is using but this is speculation.  The point was to show side x side examples using an industry standard workflow with all settings posted so the small differences could be observed within this context. Everyone could then draw their own conclusions about the movement that each camera with said settings produces.  

 

 

 

Share this post


Link to post
Share on other sites
On 4/17/2018 at 2:36 PM, sam said:

Not sure I follow.   I own all these cams and tried to be fair to each  and all.   I also designed the test so anyone can replicate it themselves on the cheap. Take a 6ft. length of rope. Measure 2 marks on a wall 6ft wide. Fasten a weight, rock, etc.. to the rope and hang it from something like c stand or a ladder.  Set your camera to the same settings and dictance as described in the yt video.  (I would choose something large and round at the end of the rope so you can easily see whats happening).  Now let gravity do its thing.   

Sorry, I wasn't being clear, no criticism of you or your test was intended. I meant that you can see the effects of the rolling shutters pretty clearly in your test and at least for me, your results go beyond subjective judgements about motion cadence and into almost defect or definitely artifact territory. I think tests like yours are great, I'm just also (in combination with tests like yours) curious about tests that give the viewer the ability to judge more ambiguous qualities and see what they prefer.  I feel a lot of things like motion cadence are a mixture of actual effects (distortion, artifacts, etc.) combined with subjective judgements (taste, context, etc.). Carry on!

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...