Stuttering image while Panning

[hoylecd]

Stuttering noticeable without panning.

Stuttering noticeable without panning.

I am currently working on a feature and stuttering has become very apparent. Particularly, a scene in which the main character is lecturing in front of a dry erase white board. The scene involves him moving around quite a bit, and any quick movement displays this stuttering effect. While I have not tried watching it on a CRT monitor, we've tested a Pro Res clip on several different monitors with the same outcome, and importing into FCP or using VLC does not help solve this issue at all. While I understand that the quick panning issue is relevant to both film and digital shooting at 24p, I'm curious if this stuttering would be noticed had this scene been shot on film. Also, we're shooting on Build 20, which is still beta I know, and the DP was wondering if this problem is more or less apparent on previous builds. I've worked on several RED shoots, and never once noticed this stuttering to be so apparent. Any thoughts?

 

[Uli Plank]

But Sanjin, you can't just tell the director 'sorry, I can only pan at x speed' because the camera won't let me go any faster! :rolleyes5:

You have to! Every manual for serious cinematography tells you the correct speeds.

Plus, the situation is very different if you follow an actor/subject and keep the BG out of focus. Since the viewer's attention is drawn to the moving target, which is kept relatively stable in the frame, you may pan faster.

Did you lately watch an action-movie in cinema? They don't follow the rules any more and it stutters like hell!

 

[Tony Covell]

The problem is the speed of the video acceptance of picture to memory. Its always happened on Video cameras right from the start. Tube Tape or Chip it always happens.
Check out ASC Manual of Cinematography that has 35mm pan speeds and figure its a lot lower. I dont know of any tables as it will change according to picture set up and content. Thought: You dont see whip pans on motor racing just cuts or slight movement while tracking a particular car and its always in relief in picture. Thats the reason.
Hope this assists.
Tony Covell.

 

[Daniel García]

I think there are people here with a slight misconception of the "too fast" issue.
My understanding of the panavision/ASC panning speed charts goes like this:
Slow pans under the stated threshold are good.
Faster pans show the stuttering effect.
But there's another threshold, approaching whip pans, where motion blur takes over and there's no visible stutter again (at least when using standard 180º shutter).

So the problem is not the whip pans, it's those in between, neither slow nor fast enough.

To be clear this is general 24p knowledge, I'm not discussing whether the RED threshold is lower than usual or worse than other cameras. I suspect this is more of a display issue, but it's something that should be throughly investigated.

BTW thanks for the BBC whitepaper.

 

[Sanjin Jukic]

I don't my RED with me right now, it's rented but I had a quick look through my RED test archive and found one example that is pretty close to quick whip pan:

Here we go with the footage: The Joker>>>

Shot on RED1 with Nikon 17-35mm f/2.8 zoom, ISO 320, 25 FPS, shutter speed 50, exported straight away from 1K QT_M proxy file,

Rode NTG 3 mic and Avid-M-Audio Micro Track II recorder line out to RED1 mini xlr line in.

 

[D'Arcy Foley-Dawson]

Here is a white paper repport of a test BBC did last year to find the framerates that eliminate film/motion judder and the ability to capture fast moving objects without judder and smearing in a possible future TV standard.

The white paper both describes the reason for why motion judder happens and concludes that 300fps seems to be the necessary framerate to obtain this.
http://downloads.bbc.co.uk/rd/pubs/whp/whp-pdf-files/WHP169.pdf

i love it that the bbc publishes white papers on research into frame rates! it's so wonderfully nerdy and obscure. thanks for the link!

 

[Stephen Williams]

What can i say? I saw KNOWING at the theater and didn't notice any stuttering problems at all! And Steven says it was unwatchable because of all kinds of movement?! Wow! That's strange!

Something else strange about KNOWING was that the print i watched was slightly soft. Soft in a very un-RED kind of way...

Hi Miltos,

I saw Knowing on a big screen projected from film, same screen as Ben Button a couple of weeks before.

Actor movement on screen form Knowing looked awful, wides were hopelessly soft & the focus often seemed in the wrong place or just soft.

Stephen

BTW none of those problems were seen at the 4k showing @ IBC. Who does the post seems to make an awful lot of difference.

 

[Bill Anderson]

Well, physics isn't going to bend its laws, so what is the difference between film and Digital acquisition? Why should one discreet frame at 1/50 sec be more conducive to showing what happens when you pan a camera at various speed etc.? Apart from the huge variables from one camera operator's expectations/claims and another's, we might expect a slightly different strobing behavior between film and digital; apart from other film post Vs digital post considerations, the film image is made up of arbitrarily positioned grain, each frame unique in its physical makeup. Does the grain of each frame of film tend to have a slight (masking?) effect on strobing because it doesn't (physically) exactly mirror the frame before it? Digital images on the other hand are made up of an identical grid pattern, locked exactly into position - no breathing, no weave or transport through a gate, no color layers and developers waiting to respond uniquely to every exposure - very few variables between frames, just a perfectly matched grid suited to show what goes on when you pan a camera... Whip pans can often be free of strobing because the information (image) in each frame is mostly unique, it no longer mirrors the info from the previous frame. Subject matter aside- there is a panning speed(s) focal length combo conducive to strobing (considered too fast a pan) because it sets up a sort of harmony between each successive frame's imagery texture: things have changed within the frame, but only slightly, not enough to break the visually perceptible connection of the textural or frequency pattern between the previous frames. It's physics, and the slight variables from here to there in film might help a little- but not a huge amount. My guess is that pan induced strobing is a pan speed X shutter X textural frequency (for want of a better word) and that it would be possible to induce strobing even at slower pan speeds if the subject matter was carefully chosen for this.
It's been pointed out: one option is to decrease strobing with slower exposure rates- induce blur, something that is less uniform in the makeup of the frame to frame image- not to mention decreasing the lag between each exposure.

 

[Bill Anderson]

My guess is that any strobing claim made about RED or any other camera is only valid under very unique circumstances and isn't an inherent problem - every camera has its own achilles heel when it comes to this effect: subject matter- light ratio- pan speed-shutter speed- focal length- post methods- combinations and so on. I doubt RED is any more susceptible than any other camera. Maybe the experts can chime in with their thoughts.

Will larger sensors have an impact on this phenomenon???

 

[DCC Erickson]

I am currently working on a feature and stuttering has become very apparent. Particularly, a scene in which the main character is lecturing in front of a dry erase white board. The scene involves him moving around quite a bit, and any quick movement displays this stuttering effect. While I have not tried watching it on a CRT monitor, we've tested a Pro Res clip on several different monitors with the same outcome, and importing into FCP or using VLC does not help solve this issue at all. While I understand that the quick panning issue is relevant to both film and digital shooting at 24p, I'm curious if this stuttering would be noticed had this scene been shot on film. Also, we're shooting on Build 20, which is still beta I know, and the DP was wondering if this problem is more or less apparent on previous builds. I've worked on several RED shoots, and never once noticed this stuttering to be so apparent. Any thoughts?

Just reading this, I wonder if the dry erase white board might be part of the problem - glare, very high contrast with the moving actor might exaggerate the effect, perhaps? Doesn't exactly fix your shot, I know.

 

[Jay Schweitzer]

well what if your shooting a high speed action sport like motocross and have no choice but to pan fast to follow the action, then what do you do??

 

[Steve Gibby]

well what if your shooting a high speed action sport like motocross and have no choice but to pan fast to follow the action, then what do you do??

I've shot a ton of high speed action sports with RED for television broadcast - fast pans, tilts, close action follow of high speed objects and people, etc. In 4k I usually shoot it a 29.97, with a shutter speed of around 1/96th second - and the results have been smooth. For higher frame rates in action sports I shoot 3k, project frame rate of 23.98, Varispeed up to 48 fps, and a shutter speed of around 120 fps, for 50% slow motion. The results have been glass-smooth.

If you're tracking action sports, and insisting on shooting 23.98 or 24 fps, yes you will have very stuttery footage in fast pans and tilts. In short, there are definitely times to use higher frame rates and shutter speeds - and action sports, nature (fast running animals), etc. are some of those times.

A "24 frames and 180 degree shutter or nothing" attitude works well with narrative genres, but there's an entire alternate universe of production genres beyond narrative that regularly call for different frame rates and shutter speeds - and that's precisely why RED has included a wide variety of frame rates and shutter speeds in RED one - and have announced an even wider range of them in Epic and Scarlet.

Obviously for televised content you need to meet the Producer's Guidelines for the tech end of your delivered master - so you need to plan out your acquisition, pre-post, post-production, and mastering accordingly.

 

[Eki Halkka]

can you get video look with a film camera?

Sure: Shoot at 50 fps (PAL) or 60 fps (NTSC), and play the movie back at that same framerate.

 

[Joe G.]

Steve, this is very interesting:

"If you're tracking action sports, and insisting on shooting 23.98 or 24 fps, yes you will have very stuttery footage in fast pans and tilts. In short, there are definitely times to use higher frame rates and shutter speeds - and action sports, nature (fast running animals), etc. are some of those times."

What if the task was to shoot an action scene, say at 48fps or 72fps, and convert it to 24fps final output? The goal being to just eliminate the stuttering by combining multiple frames in a fancy post program?

 

[Gavin Greenwalt]

It's not the framerate you shoot at. It's the framerate you deliver that's important.

Your "Fancy post program" would be to just increase the exposure time on the camera. The ultimate post program would just create a more motionblurred frame.

 

[Eki Halkka]

What if the task was to shoot an action scene, say at 48fps or 72fps, and convert it to 24fps final output? The goal being to just eliminate the stuttering by combining multiple frames in a fancy post program?

That wouldn't help at all. You need to both shoot AND playback at a higher frame rate for it to look smoother.

 

[Steve Gibby]

Joe G,

If you overcrank the camera, RED One in this case, to a higher frame rate than it will be played at in your delivery medium, then slow motion will result. For example, lets say I set up RED for this profile: 3k, project frame rate of 23.98, Varispeed frame rate to 48 fps, and throw in a shutter speed of maybe 120th second, then when I play those files back on the timeline in my editing system, they will play as slow motion - and yes it will be smooth. Even clicking on the Quick Time proxies of those files will trigger a slow motion playback of the proxies, because they are slaved from the R3D files, which were shot in the camera setup profile mentioned above.

If it is smooth slow motion of fast moving objects, people, etc. you're after you just need to overcrank the camera with a good combination of frame rate and shutter speed - then play it back on a slower project rate timeline.

If it is action sports or some other fast motion subject matter, and lets say the delivery medium is 59.94 fps, then you'd need to overcrank above 59.94 on acquisition at a frame rate/shutter speed combination that would get you the rate of slow motion you're after when it is played back on a 59.94 timeline. With the proposed (but not confirmed or finalized) overcranking capabilities of Epic and Scarlet extreme slow motion should be seriously enabled.

I know a lot of guys here are big fans of doing everything in 23.98 and 24 fps. I'm not - but to each their own. I prefer higher frame rates and shutter speeds, very low judder, and smooth motion. If you've watched the outstanding cinematography of NFL Films, that would be an example of what I like. They do shoot lower frame rates when interviewees and subject matter are stationary or relatively stationary - but for high action sequences they overcrank up to high frame rates and shutter speeds to get glass smooth motion upon playback on the slower frame rate timeline.

If you're shooting high-action sports with RED, and 4k is the resolution you want/need, and the footage is destined for a USA-televised delivery, then a camera profile of QuadHD, 29.97, 1/96th second shutter will give you smooth playback and low judder in your rapid pans and camera movements while tracking the sports action. USA TV channels that televise sports usually ask for mastering at 29.97 or 59.94.

There's more to a "film look" than 24fps/180 degree shutter. Just some other factors are lighting, camera movements, filtering, camera usage style, post production processing (in RED's case how you process the RC Raw files), and on and on and on. Beyond that, everyone doesn't have to seek a "film look" for their projects - they just need to seek it if their subject matter, distribution outlet, and/or delivery medium require it. Perfectly valid artistic expression with a camera can come from a wide variety of camera setups and their resulting "looks".

Again - to each their own - but IMO, and for a lot of the genres I work in, different subject matters call for different frame rates and shutter speeds. Certain genres and mediums call for dreamscapes, suspended reality, and blurry judder - but others genres and mediums call for bringing viewers into the present, smooth motion, and low judder. Both approaches are perfectly valid for what we're trying to impress upon each particular audience.

 

[Joe G.]

I was talking about "time remapping" in Adobe After Effects, for example. They compute new frames based upon the surrounding frames.

 

[Denis Buhot]

Hi Steve, I'm sending you a PM...

 

[Denis Buhot]

My question is ...

My question is ...

They compute new frames based upon the surrounding frames.

how do them people using red for the big screen (soderbergh and others) achieve a smooth rendering... Che is full of action scenes when no stuttering occurs... And no slow-mo...