GH2 in 24p: noticeable "strobe effect" with any motion in frame?

Roberto

@MisterJ Yes, that pic shows the right amount of blur to produce smooth motion. I'm expecting your others are about right, too.

I'm really more interested in something quantifiable - if possible to reliably produce stuttering video. On my own I can only test a few targeted samples. I keep wondering if there's any benchmark we can all test...

how about..

Freight trains often carry universal-standard shipping containers. Where I live we get long trains, (4 km!), moving at a limited speed through town.

Instead of trying to pan, maybe I'll see if the contrasting vertical flutes on the sides of moving containers can produce stutter.

Best of all, it's quantifiable:

The speed that matters is the movement across the field of view, not the Km/hour. The passage of a given point on a container - say, from left to right of the picture - can be accurately counted by the number of frames;
A train is unlikely to make any sudden irregular movement, accelerate or slow down.
A train takes long enough to pass by for a number of takes at different shutter speeds.

At present, I'm inclined to use firmware 1.1 (no hack) and the lenses I have. (14mm pancake might work with such a big subject).

MisterJ

I did really quick test with @LPowell latest hack. I'm not sure is this the right place to put material for You guys, but here is the one picture from FCP X. I don't have best monitors and videocard at home, but my 1st feel is that I really like, how @LPowell's latest hack captures motion. Specially in 1/25 shutter speed.

So this shot was f 6.3, shutter 25, ISO 200. It's 24H and kit lens 14-42. Jpg is exported from timeline. I did variation of same kind of movement with shutter speeds of 25, 30, 40 and 50. So @Roberto or anyone else, tell me, if You want to check those.

MisterJ

@bwhitz

Please come on. I've been working with Alexa and RED many times. Have done fast dollies where in front of a frame there is lots of vertical objects and we did not use 60fps. Here is picture attached. That's from a dolly track shot, with RED. No strobing, no problem. Theres no irony here. You are terribly wrong if you think that all Alexa and Red material in 24p should have strobing effect. It's about shutter speed. And @Roberto is right about motion blur. You are right that even RED and Alexa has it limitations. Strobing will come out using those cameras if you have, lets say, fast pan and lots of thin vertical lines. But I've done shooting with GH2 hundreds of minutes. I've shot in all shutter speeds there is. I've used 1/25 shutter speed and took a handheld shot with a little movement and there it is. Strobing. At this moment, GH2 generates strobing too easily.

All I wish, is to look this problem through. Without arguing. I'm not expecting to have same quality in motion as Alexa, but if there is something to do to get better and less strobing shots - great.

@LPowell I'll download your latest and take a look.

Best, misterj

Roberto

claims of judder don't have any accompanying videos for us to look at

Yep, talk is indeed cheap. post some samples, to accompany claims, please!

Until such time, going by @Astro's sample of 1, I hereby find all stuttering video to show measurably less motion-blur than the stated shutter speed should be producing. ;-)

mo7ies

Well, @bwhitz got me thinking. Maybe in fact GH2 does record a very sharp video with enough bandwidth to also produce rather low-blur motion. Question is, what is considered visually pleasant? To me, "strobe effect" in place of the expected smooth motion is Not pleasant. I have not worked with Alexa or RED personally, so can't compare. But I did work with 7D extensively, and its motion was never objectionable.

Now, if @bwhitz is right, maybe 7D's 24p motion is better because it is simply not sharp enough to begin with, plus its codec is rather inferior.

Then like @Roberto pointed out, most likely there's a combination of factors with GH2. For example, stock 14-42mm lens may exaggerate the motion artifacts by forcing its own over-sharpening algorithm.

Well, at this point, whatever is the cause, if I shoot 24H at 1/30 shutter, the annoying Orange... err.. motion :) is fixed. If I shoot fast sports, 1/30 feels a tiny bit long on blur, BUT nowhere near as objectionable IMHO as a stuttering look at 1/50.

Roberto

@bwhitz

@mo7ies said

not enough bandwidth to record enough data when some motion is in frame

Which I am prepared to accept as part of the general "failure to compress" problem which may or may not apply to strobing. Lots of busy scenes (like analogue TV noise) fail to compress, whether it's issues of codec, processor speed, card writing, de-compression and bandwidth etc. There could be just one bottleneck anywhere - or even a combination of several.

mo7ies

@LPowell no need to be condescending here, pal. So you had a little back-and-forth with Driftwood, and now you are fuming on him and everyone else, I see it from other threads I'm following. Tone it down, my friend - you are being both unfair, factually incorrect, and too personal.

My thread is based on my own observations, and I have researched personal-view and other resources before posting here. I also did post a link to the video that illustrates one of the cases, earlier in this thread, together with the "cure" - longer motion blur.

LPowell

@mo7ies So this thread is really nothing more than rehashed rumor-mongering? Your evidence is a second-hand account of what "one of the guys from Universal" said about the subjective results of a completely undocumented workflow that produced a Blu-ray rendering that no one here has ever seen? Here's what @Ian_T had to say about that in the same thread you linked, and I agree with him completely:

"...um...is it me or does it seem strange that all of these claims of judder don't have any accompanying videos for us to look at? I sincerily have no idea what Dalefpf is talking about. I've heard of these judder claims before with the GH-1. But I've yet to see someone show an example of this "judder" (that is supposedly different from any every other 24p camera). "

mo7ies

@bwhitz I see your point. What about this, however?

http://www.personal-view.com/talks/discussion/comment/8687#Comment_8687

"I asked one of the guys from Universal to come take a look at it. He agreed it looked terrible." The way I understand it, he says GH2 motion looked bad comparing to film, if we take that as a reference point (I see that whatever else I refer to here, is getting rather strong reactions, so let's try old media, eh!)

mo7ies

@Rambo GoPro reference was to illustrate that GH2's motion looks similar to the much lower-class video acquisition device which is HERO2 camera. I do understand your point that Hero has very high "saving private ryan" shutter speed outdoors in bright light, which I'm sure you're correct about. Then why would GH2's motion look very similar at only 1/50 shutter, which is 10x slower than 1/500 shutter of Hero cam?

RRRR

Another thing to consider, when comparing and talking about the temporal look of 24 frame footage, in film you basically have a black frame between each image. As far as I know, no digital camera yet reproduces this. Instead you have image, image, image..

bwhitz

@Roberto

not enough bandwidth to record enough data when some motion is in frame

That's one idea which deserves its place in any synopsis as well as any other.

Not really. Like I said above...

Where's the logic in a GH2 intra patch with 100mb/s+ not having enough bandwidth for motion... but the 7D you referred to, reproducing motion more faithfully ,with only 50mb/s? Or the F3 with only 28mb/s?

But as filmmaker I think at the moment, that if I use GH2 in my film, I must compromise too much if I want to have a cinematic feel and movement in shots. Meaning how fast I can pan or tilt? What kind of objects I can have in a front of a frame during a dolly track?

Haha, oh sweet irony. This is EXACTLY what directors and cinematographers run into when shooting film and digital-cinema. Most good directors, DP's, and operators that shoot the high-end films, have to be constantly aware of the pan/dolly speeds they can use and plan around them. This is just what you have to deal with when shooting 24fps! Use 60p for sh*ts sake!!!

LPowell

@mo78es - "I'd say we are dealing with not enough bandwidth to record enough data when some motion is in frame."

Any downloadable footage that demonstrates this speculation?

Rambo

mo7ies, the GOPro is a variable shutter speed camera that adjusts exposure using mostly shutter speed, usually at a very high rate 1/500+ in average outdoor conditions for what it was designed for. Of course it will display unpleasant ( for some) stutter motion, it's designed this way because it is an action cam that is subject to high camera motion. If it were 1/60 the results would be mostly blurred video and very unhappy users. Why even mention this camera in this thread?

Roberto

@mo7ies said

not enough bandwidth to record enough data when some motion is in frame

That's one idea which deserves its place in any synopsis as well as any other.

bwhitz

Well, for the 7D, as much as a fan of it as I was... and I still use it, has a muddiness to the motion that is again, caused by long-GOP temporal compression. This is why I really don't like to use my 7D any more for personal projects.

But... there's no reason to NOT like this, it could just be your personal taste, but the motion from Intra-GH2 patches is nearly identical to that produced by an Alexa, Epic, Scarlet, probably F65 (haven't worked with footage yet though), and yes, even a real film DI. This is just how real 24fps footage looks.

I think it's just personal preference here. You must just prefer the smoothed out long-GOP look. It's kind of like how some people prefer the heavy compression of lower bit-rates because it takes away the images texture and looks "less noisy" to them.

But I can assure you... besides some weird possible glitch...the GH2 isn't stuttering more than other 24fps cameras, it's studdering completely accurately. Other cameras like the 7D, F3, AF-100, are studdering less.

If I had to guess, I'd say we are dealing with not enough bandwidth to record enough data when some motion is in frame.

Kind of backwards here. Not having enough bandwidth to render all the motion is what causes the GOP to "blend" frames, for lack of another word, and LOSE some of the studdering that the footage should have. This is actually what you seem to prefer.

Plus, where's the logic in a GH2 intra patch with 100mb/s+ not having enough bandwidth for motion... but the 7D you referred to, reproducing motion more faithfully ,with only 50mb/s? Or the F3 with only 28mb/s?

mo7ies

Would it be better if I compared GH2 (unfavorably) to Canon 7D's 24p?..

I start thinking that maybe this is a psychovisual (invented word?) phenomenon.

Clearly, some people see it (and bothered by it), and some people just don't. Kind of like rainbow effect on the old-time DLP projectors...

I thought we established this earlier in the thread, but if not, here goes: GH2's 24p motion is MORE stuttering than what is usual for the 24fps time sampling. To me, it looks like a night-club scene with strobe light on - for the lack of better analogy.

I recently had a chance to use my GoPro HERO2 camera, and it has very similar unpleasant motion stutter - in 1080p30 mode!

If I had to guess, I'd say we are dealing with not enough bandwidth to record enough data when some motion is in frame. Once again, just a speculation. Whatever it is, the effect is pronounced. I see that people pounce on F3, which probably was a bad example. So then, let's compare to Canon 7D - which had way more pleasant 24p motion rendering. Is it too much to ask for GH2 to have the same 24p motion quality as Canon 7D? Now that both devices are clearly of the same class?

bwhitz

Ok seriously this is crazy... The strobebing on the GH2 is what gives it such an authentic real 24p look. Have any of you guys complaining ever shot Red or Alexa and watched the uncompressed non-vimeo footage? This is how they look at 24p. It's the real-deal digital cinema look. All frames rendered individually (Intra patches) means more stuttering. This is a good thing. The motion on something like the F3 screams digital to me for this reason alone... the temporal compression smooths out the stuttering to a degree and ruins the true-cinema motion. If you like "smooth" realistic footage... use 60i or 720p60fps mode. If you want cinematic motion... you really can't get any better than the Intra-GH2 for under 20k...

@itimjim

There's a chap on here who can tell the difference between a 1/40 and 1/50 shutter.

This is actually pretty noticeable. And using something like a 1/30 shutter is getting into cruddy lookin' footage territory...

@mo7ies

What we are saying is that GH2's 24p is visually more strobing than, say, Sony F3's 24p under exact same conditions.

Well yea, and this is why the F3 (internal codec) has a bit more of a video look to it... even with the extra latitude. Temporal compression.

driftwood

@nomad With the F3 can't you turn the shutter off to sort out strobing like the EX3? It works like an auto shutter.

Roberto

@itimjim

To be fair, there's only a select bunch of people that have found this a problem.

That's true - but these threads are linear and when we repeatedly raise the perception aspect it ends up sending us around in circles. I really would like to do a synopsis of this entire phenomenon, including the valid aspects of viewer perception and the context in which we view the footage.

Roberto

@Misterj says

Most of the shots I see posted here are static ones.

Yes, because one of the reasons 24p stutters is under motion-blurring. I have repeatedly asked for users to post stills from stuttering video to see if it's blurred as much as it should be, given the supposed shutter setting.

SO far, i've only seen 2 examples on this thread - both suffering from momentary handheld shake. That's not good enough. I'm quite sure there are other 24p stutter culprits. But we have to rule out the obvious one first. As I keep saying (sigh):

I've never seen stuttering video which didn't suffer from under-blurring.

@LPowell wrote:

no one can alter the properties of human vision to make 24fps video look less stuttery than higher frame rates.

Well, no - but since we can correct it using motion-blur in post, we can adapt to suit the human eye. @Mo7ies uses 1/30 shutter to do it in-camera.

nomad

I can see strobing with both the F3 and the Red Epic at 24p with 1/48th shutter too. It all depends on camera and object movement, contrast and DOF. To me it's always a bit more than with analog film.

LPowell

@mo7ies Sorry, I don't know of any way to make a GH2 look as good as an F3!

mo7ies

@LPowell wrote:

no one can alter the properties of human vision to make 24fps video look less stuttery than higher frame rates.

But that's not what we are trying to achieve here.

What we are saying is that GH2's 24p is visually more strobing than, say, Sony F3's 24p under exact same conditions.

It is very prominent, and annoying to the viewer.

LPowell

@mo7ies All AVCHD hacks have inherent limitations - they cannot alter the properties of the camera's image sensor, nor the electronic shutter that controls it. Likewise, no one can alter the properties of human vision to make 24fps video look less stuttery than higher frame rates. As @P4INKiller pointed out, high-contrast highlights can emphasize this inherent stutter, as can fast shutter speeds.

@MisterJ "Fluid cinematic motion" is precisely what I aimed to produce with Flow Motion v2:

http://www.personal-view.com./talks/discussion/3337/gh2-flow-motion-v2.02-spans-in-hbr-25p-24h-720p/p1

Howdy, Stranger!

Categories

Tags in Topic

Top Posters