Personal View site logo
GH2 in 24p: noticeable "strobe effect" with any motion in frame?
  • 259 Replies sorted by
  • Well, pantyhose/filters will reduce overall resolution, while slower shutter speed only affects moving objects. IMHO, the latter is clearly preferable, because moving objects already are blurred by motion and thus lack resolution.

    BTW, stock lenses like 14-42mm are also very sharp.

    In any case, my current experience is to shoot at 1/30 or 1/25 shutter, and the strobing goes away.

  • Yea, possibly driftwood, or flow motion or sanity v4, one of these it must be.... (it wasn't my intent to hack my new gh2, i wanted to get the best out of normal settings first, but panicked with that strobing issue and thought that "flow motion" or "aquamotion" meant that the movements would be more fluid :D

    well, i don't regret the patches anyway, and i've been learning a lot lately.... still need to find a way to be sure that i can get no excessive strobing after editing (like when putting it on a dvd or so)

  • Just got back from a car drive with my GH2, had some trouble with the light (sunny day, lots of trees, and random clouds) and a very dirty windshield (oops), other than that, movement is great... really need to focus on details to see some strobing, but this could be just normal "cinema" style. (i'm talking about clips i viewed MTS, on windows movie player, 24p mostly, also tried 25p, but doesn't change anything). Hack pushed writing speed to 68 Mb/s , everything works fine on my sandisk class 10.

    At the end, i wanted to take a static shot of many cars passing by, cars in shadow against a sunny background, but i had inadvertently touched the shutter speed (while rapidly trying to change aperture), put it at 1/20, and strobing was awful :s I hope i can try it again with 1/25 soon.

  • At 1/20 shutter, you are forcing it to be open longer than the duration of each frame which is still set at 1/24 of a second. So you get a stop-motion effect. This is very different from strobing, though.

  • Yea something like that, it was like mega-strobing :D I was too frustrated at my mistake to take time and analyse the default :s (i'm trying to master all the technical stuff in order for me to use this cam soon on fiction shooting, where it's absolutely important to avoid such mistakes or else it can ruin the work of a whole team). I tried an average-speed pan earlier in my room, with the cam on an office chair, i'm still not very very pleased with the fluidity, but strobing seemed to be less (it was also a low-contrast image, with very little detail, bit rate remained around 15 Mb/s)

  • @mo7ies

    Hi.. I have done a lot more tests on this issue, incl using hacks like Sedna (I frames) and Cluster (Mixture IPB Frames) etc... bottom line is there is not much difference when you use different hacks, which is to be expected. Tho I suspect that the more I frames there are, the worse the strobe will be, because of the efficiency of the codec. it seems that a mixture (IP and B) lessens the effect somewhat.

    Its true that using a shutter speed of 30 on 24P does lessen the effect of strobbing when panning...not completely, but it is an improvement. 29.97 is still the best when it comes to panning the camera without strobe affects, although if you pan quite fast any setting will do it. However 29.97 or 25P (HBR) has its problems (as pointed out earlier) plus it renders less details than 24P when the camera is still on a complex scene. I did extensive tests on this...all types nature, guitar pans, close ups etc..

    Also testing with the old Kit lens 14-42 vs the 20mm F1.7...surprisingly the 14-42 is MUCH BETTER for moving or panning the camera (I tested the 14-42 with the cameras internal stabilizer/shake on and also off) With the stabilizer enabled or disabled its simply much better than the 20mm lens (although I love that lens)

    The 20mm is much harder to control and somehow moves very stiffly in 24P especially (compared to the 14-42) Now I know the 14-42 gets a bad rap, but personally if I shoot in 24P and want to pan then I will be using the 14-42, of course the 20mm is probably better for stills and obviously low light as well (DOF)

    I also tested footage in Adobes CS6 warp stabilizer in Post...nice results (like a glide cam) but it does soften the detail a little, so I am not sure about that.

    Well thats it for now.

    Cheers

  • Thanks for the tests and for your detailed post. Did you test any manual lenses, like Nikon/Canon with adapters?

  • @Astro : and to complete mo7ies' question, were you able to compare panning quality of 14-42 lens with that of 14-140 lens ? (or 25mm panny, or olympus 12mm, 45mm)... i know, asking a bit too much, but seems i'm thinking about ordering these lenses soon, i'd like to know if i should also have a 14-42 for panning scenes only

  • I also would be very interested to hear the difference in strobing amount between the Wide and Tele ends on the standard 14-42mm lens, in your opinion.

  • @mo7ies and @Zeko OK...I did a test the other day with a Legacy Konica 57mm F1.4 and adapter ...using the Sendna Hack (all I frames)

    I panned across three guitars on a standard tripod indoors, one is a sunburst Les Paul, the other a light blue Strat, and the other dark blue Strat. I am in PAL land and the lights were on so I had to set the shutter to 50, but I did test on the 20mm as well. Both of those lenses strobed quite badly ...especially on the light blue guitar, whereas the 14-42 was better. So high contrast definitely makes the strobing a lot more apparent. Also I processed them in the Adobe Warp Stabilizer and although the pan became buttery smooth, the strobing was still there (especially on the light blue guitar)

    Sorry I dont have the 14-140 Lens (I imagine it would be good tho) or any older Canons (although I plan to get some of those sometime) I have heard that the Olympus Prime 12mm F2.0 is very steady when handheld in the GH2...but I have never seen any examples of this in footage. I have also converted quite smooth TM700 1080p/50 frame footage to both 25P and 24P...and it becomes quite jerky and stutters when panning, although quite nice at 50 frames.

    The whole point of me testing this is...I am trying to figure out what the best plan of attack is for future clips, so I know what the hell I am doing LOL!!.

    Basically the conclusion I am coming to is

    1. That I would use lenses like the Panny 20mm and the Konica or any other legacy style lenses as portait style lenses, and keep them more or less steady on a tripod. And use the 14-42 if the Camera needs to move quite a bit.

    2.I still may check some more into slightly longer GOP hacks if I pan the camera a lot, or there is a lot of movement on 24P (Maybe Cluster which has I,P and B frames...or Flowmotion2 which has just I and B frames, its hard to get to the bottom of wether this affects the final result that much in regards to moving the camera...so I will check around more.

    The 30Shutter on 24P (as I mentioned earlier) does definitely help quite a bit too, but as I am in Pal land I can only do this in the daytime. Other than that if there is lots of panning and movement I can shoot at 29.97 from beginning to end.

    Decisions...decisions

  • Interesting results, but they go contrary to my own experience. IMHO, the kit lens 14-42 is bad for panning, especially at the long (telephoto) end. I thought the consensus was, because this lens is so sharp, it exaggerates the edges and thus increases the strobing effect.

    Other lenses may not be as sharp, and therefore do not produce that much strobing.

    My own experience was that the strobe/stutter was worse on tele end, but I'd love to hear what the others say about it. @Astro info did not clarify this, AFAIK.

    @Astro, to control exposure at daytime, may I recommend optical ND filters in front of the lens, instead of changing the shutter speed. Variable ND is also great for fine adjustments (at the expense of some image quality loss.)

  • @mo7ies Hi... Yep...you could be right, I only tested the fully panned out version of the 14-42 (14mm). So I compared that with the 20mm lens and its almost impossible to pan smoothly with the 20mm F1.7 Lens or the Konica in 24P, but the 14-42 (set at 14mm) was pretty good actually.

    You are right about sharpness too, I have come to the same conclusion, I do now avoid hi contrast areas...leaves against the bright sky and such, and thanks for the tip about the ND filter, I do have a variable one and use it a lot and it helps a lot (especially outdoors).

    I have yet to confirm wether an all Intra hack...such as Sedna or a mixed GOP such as Cluster may be better in 24P. I did another test and that was in 24P with 50shutter speed, camera fixed on a tripod with human motion...arms waving and stuff and the motion looked fine, (to me anyway) it appears that the strobing issue is a problem mainly when you move the camera. Mind you the human motion was indoors and not hi contrast, so it may be a different story outdoors in more contrast, but indoors it was good and quite natural looking.

    Another test I did was with Cluster (Version 3 I think it was) ...a close up of a highly detailed dobro in 24P and 29.97 with a very slow pan...actually both looked really good, I was surprised how much detail the 29.97 version showed (directly off the .mts files) I think I read a while back Driftwood saying the Cluster was good at HBR (30P or 25P) and it is. Well thats all for now...I think all this helps to make informed decisions about what to do before you decide on a project I am hoping to arrive at that point soon. The GH2 definitely is a great camera with some weaknesses that can be sidestepped if you know exactly what they are. Cheers

  • @Astro : talking about human motion, i tried to film myself in dark blue (black) PJ's walking around the room, and when i pass before a white closet (yellowish light from the lamp), i can see some strobing behind me. Nothing huge, but still noticeable :s (using the 14-140 mm, it's the only one i have for now)

  • Here's something new I learned today. @LPowell produced Flow Motion hack, which currently is in version 2.02, which looks like one of the best to balance high bitrate (100-140Mbs), GOP length (3), and image quality.

    But what makes it stand out, he says this hack is specifically designed to handle MOTION SMOOTHLY.

    I have not tried it yet, but people are raving about it (although mostly about the quality of the picture, and not many posters mentioned strobing improvements - but then again, some people don't even realize it it a problem to begin with.)

    Will be very interested to see in LPowell's Flow MOtion makes a sizable reduction in strobing on GH2...

    Here's the links with downloads, latest one at the bottom:

    v1 http://www.personal-view.com/talks/discussion/2099/100mbps-flow-motion-v1.11-failsafe-patch-with-hbr-25p-50p-modes/p1

    v2 http://www.personal-view.com/talks/discussion/3337/gh2-flow-motion-v2.0-patch-release/p1

  • @mo7ies : nah, i had precisely tried that patch first because i came here hoping to solve the strobing issue...and i didn't see any improvement on that :s

  • @mo7ies Hi... I actually tested Flowmotion2 quite a bit,I have folder full off stuff I have filmed in it. No real difference as Zeko said, its an I frame and 2 B frame style codec, strobes just the same as all the others really. I guess maybe for running water and stuff its cool on the details, but generally its no real improvement in the areas we are talking about. I also have tested Cluster 2 (king of the long Gops Driftwood calls it) and thats I frame 4 B frames, and 2 P frames. I filmed a lot of cars a buses on a road with that one and ...plenty of strobing in that one too. I am pretty much convinced no hack will fix this.

    I have come to this conclusion really (see if you agree or have differing thoughts on this), I am keen to know your and Zeko's opinion. Anyway...Heres what I have found after extensive tests

    1. Shutter Speed of 30 or less definitely does help with 24P on the Gh2, 180% shutter (to my eyes is not like film) so a shutter speed of 40 or 50 is where the strobing is most apparent...obviously it will get worse in higher shutter speeds. You both have probably read this article (I posted it earlier I think)

    http://www.hdvideopro.com/technique/miscellaneous-technique/help-desk-did-i-judder.html?start=1

    1. ND filters help too, as does shooting in low contrast situations, but none of these remedies completely eliminate the problem (and it is a problem) despite people saying its how film looks. Yes I have seen strobing in films, but not as bad as the Gh2 does in 24P.

    2. Vitaliy mentioned earlier about time to start shooting in 50 or 60P (higher framerates) ...and he is dead right. The older TM700 at 50P or 60P does not have this problem at all, 24P does tho. Some people said that frame rates wont fix the problem, well that has not been my experience ...over and over. But most places...YouTube, Vimeo, Movies etc...dont accept these higher frame rates, and so you have to convert back ...and the problem returns. (tested this extensively)

    3. It must be some kind of shutter problem in CMOS based cameras, and made worse because the GH2 is so clean and sharp. Panasonic obviously knows this, and thats maybe why there is a lot of talk about a Global Shutter being developed...but not on the GH3 (unfortunately)

    A 29.97 frame rate pretty much eliminates the problem, but once you decide on that, you have to stick with it all the way. This is something of a solution to me, that and using 24P at 30shtr speed along with ND filters and super slow panning. Its kind of annoying because it limits the 24 and 25P option quite a bit, but thats life I guess for the time being. Hopefully there may be better solutions in the near future. Sorry for the long rant, tried to cover everything.

    Cheers

  • My solution when shooting 24p with GH2 is to simply have shutter as slow as 1/30s or even 1/25s. Then the motion strobing is no longer as crazy.

  • @Zeko @Astro

    Could you please post stills of:

    1. Shots of moving background which show strobing
    2. Similar shots which do not show strobing

    I never saw strobing which did not have something to do with inadequate motion-blur. Other kinds of blur can mask the too-static frames (say, out-of-focus blur, softer-lens blur, pantyhose filter), or you can watch your video on a fancy player (Splash PRO?) but this is like lipstick on a pig. You might as well drink a stiff whisky. As long as you've got insufficient motion-blur, we haven't got to the bottom of this long-term problem, and I'd like to.

  • Post content deleted, as i already posted on page one this thread.

  • Once again, to test blurring of a stuttering video we need to know:

    1. How fast is the background passing by? e.g. if panning trees left to right, count how many frames it takes a single tree to enter the frame at right, move across and finally exit from the left;
    2. What was the shutter setting?
    3. We need to see a single frame from within that series.

    We can then calculate how much blurring there should be and compare that to how much there is in the frame-grab.

    In the "Farkle" http://personal-view.com/talks/discussion/comment/8788#Comment_8788 video example, I have to agree agree with @CobyD that the stuttering video No. 2 was shot at too high a shutter speed. Thanks for the attempted experiment, but we just need to reproduce this..

  • @Roberto Thanks for your interest, much appreciated. Just to make it clear tho...I have been testing this for ages, and basically I have arrived at my conclusions...and that is the 24P-180% shutter on the GH2 is troublesome unless there is

    1.Low contrast, 2.Very Slow Pan (painfully slow), 3.ND Filter outdoors. 4.Slower than 40 shutter (as mo7ies has mentioned)

    I also NEVER shoot at 24P with a shutter speed thats over 50 (unless I want a seizure)

    Anyway here are a few examples..I had to re size these 15 jpgs to 800 as the 1080 png's were a bit large.

    This is one of my faster pans on Cluster 2 ....but its still reasonably slowish (Long GOP)

    BTW Short GOP (All I frames) looks just as bad. My slow pains are very slooow..and I can still see the stutter. On these images the Shutter was at 50 and its 24P, Shutter 40 is not really much different, shutter 30 improves things, but does not totally fix the problem totally.

    I have read sooo many forums...about too much smear equals video look, stutter look is the film look etc......yada yada.

    However the GH2 stutters at 24P at 180% shutter or even at 40 shutter speed, many people accept this and some dont. All I can do is act according to my own tastes and that is the stutter is unacceptable in the GH2 at the "Film Standard"...for my work at least.

    Also...Yes I have set my monitor at different refresh rates below 60hz etc..

    Bottom line is I feel I have done enough testing on this, there comes a time when you just have to move on and create. For my 2 cents it is a weakness in the GH2 and there is no doubt about it in my mind, so I have to find workarounds...which I have. Hope this helps

    Thanks Again

    Caps-Jpg.zip
    3M
  • @Roberto BTW As I mentioned before on this forum. Nearly all these problems disappear as soon as soon as I use 29.97 frame rate at 1080P, pans are smoother, handheld is smoother, stuttering is 80 to 90% gone. Cluster V2 retains quite a bit of detail at this frame rate.

    As a last....Here are 2 caps from a slow pan at 1080P of a Dobro shot at 29.97 and 24P (both at 40 shutter) and with a 20mm F1.7 lens. Unfortunately the 24P cap is slightly out of focus and is closer, but the comparison will give you some idea of the level of detail with the Cluster V2.

    Dobro.zip
    1M
  • @Astro

    Thanks for the files [Nice maleleuca (?) trees and a bee-uutiful dobro!]

    Before unzipping them I'd thought I'd have to ask you to go back and count the frames in the pan - otherwise I have no way of knowing whether it's a 2-second pan or a 30-second one.

    However, on seeing these stills, I've sort of got an answer - I'd be forgiven for thinking there was no perceptible blurring at all. It's as if these were 20-second pans or something.

    To do this thoroughly, I'd still like you to go back and play a section of each, frame-by-frame, in VLC or something, so that either a tree or a pylon or something crosses the frame, one side to another, then count the frames. Without that, there's no point of reference, ergo no point.

    As I said before, A cyclist travelling at 20 kph will move 10 centimetres during the frame so we can expect a blur about half the size of his head... Even without speeds, the maths will tell us exactly how much it should be blurred.

    Bottom line is I feel I have done enough testing on this..

    I understand.

  • As I mentioned before: electronic shutters don't give you the same attack and decay structure as mechanical ones. It's a general problem of such cameras, not only the GH2, and one company (no affiliation on my side) seems to have a solution for serious dollars: Tessive. Lot's of examples on their site: http://www.tessive.com

  • @nomad

    Sorry, I have a completely open mind about this, but until we do real, simple, reproduce-able tests, talk about anything else just muddies the waters.