Personal View site logo
Can the hack affect the jello... or the perseption of it.
  • Look at the video below shot by @bwhitz to test the low gop avchd hack . If you have not seen it before it sure is very very good and the best advert for the GH23, GhVK etc... I find it to be very jello/wobble free, more so from how it was shot. If you read the comment section, you will see that he could not get any stabilizer and had to shoot handheld and many shot are... in a moving train. I can't think of a lot of situation that can be worst than that. Now if you look at the shoots you can see that it has a lot of motion and in the train ones vibration. But strangely it is quite jello/wobble proof.




    I know that as some said it before that the speed of the sensor does not change with the hack. In this case (low gop), I am asking myself if on longer gop the jello is more amplified because the codec is calculating much more from a single I frame every 12 frame. Like a cascading effect, where the next frame has to calculate from the last one that was compromised by the jello, and adds it next little motion which will affect the next one which will calculate from it, which result in a jello fest. I might be completely wrong in this but I am a bit intrigued by the above video, where with my experience you should see much more wobble. I did not see any post about it, if it has been discussed thoroughly, I don't mind if it is put back in the thread or deleted.
  • 17 Replies sorted by
  • Just to add that it is better to see the motion in the 720p version on vimeo, because the downloaded 1080p version does not play very well in many players I have tested.
  • No, it can't affect jello.

    Next topic without tags and in wrong category will be deleted without warning.
  • " the downloaded 1080p version does not play very well in many players I have tested. "

    It plays perfectly fine on mine. You must not have a powerful enough computer to run it.
  • You will see that many people have problem playing back with some players.
  • It plays choppily on my 8-core MacPro with ATI 5870 graphics card running 10.7 as well ;)

    Just transcode to ProRes and it will play back great.
  • So for those that have been able to look at it what are your thoughts. I know it is very subjective and not scientific at all.
  • @Oedipax

    I've got pretty much the same computer, but with the stock G Force GT. It plays back fine for me. Weird. What's your RAM like? I've got 12gb in my system... don't know if that really makes a playback difference though.

    +1 for transcoding to ProRes. That's probably the best option.

    As for the jello... I don't think it effects the native speed of the sensor, there's no way it could. But, like @danyyyel said, it could have an effect on the perception or amplification. The cascading theory is interesting and is kind of along the same line of what I usually see with long-GOP footage. Or maybe the GH2's sensor is just fast enough that hand held footage just doesn't show jello either way...?
  • I too have noticed this, my only answer is that the jello effect is compounded by compression, (makes sense- objects moving in funny way - with high levels of compression would effect quality).

    So I believe that I the Jello looks 'nicer' with the patch- (but it's still there- and even RED has jello so no-one really cares anymore anyway...) (Except the global shutter crew- who secretly want to squash a Phantom camera into a Lumix, it will happen - but maybe in 10years! WoooHooo Home movies will ROCK in 10 years!)
  • I also don't think that the refresh rate would be higher that is why I am thinking it might come more from the higher bitrate or lower gop, if ever it is not only perception. I have tried to see the different subway video but they were all wide angle, so it a bit difficult to judge.
  • There is a very interesting hardware solution for rolling shutter and temporal aliasing (wheels turning backwards) they do say that they need clean video out of the camera in order to synch. http://www.tessive.com/home/time-filter
  • @bwhitz, I'm running 14GB of ram in my machine. What app are you using for playback? Maybe it's a Quicktime vs VLC vs whatever thing.
  • The now very usable MJPEG does seem to have much less jello compared to the AVCHD.
  • I took a risk and used Kae's settings on some professional work this week. (Not just charts and leaves!) I've spent the last few days editing it and unfortunately, I do not see any reduction of jello over what I would expect with the standard firmware. However, I've color graded several shots that I'm certain would have shown posterization banding without the firmware hack (because I've seen it on several previous shots filmed under similar conditions.) The low light/high ISO noise is also greatly improved. However, I'll probably scale back my settings because I'm losing some shots and not able to play back in camera--both of which are crucial to my work.
  • @DouglasHorn

    I haven't noticed any jello reduction with any of the settings, my variations included. Were you losing shots in 24P or the other modes? Also, sometimes with the higher bitrate settings I get the clips not playable in camera message. I turn the cam off and on and the clips will play back fine. Annoying, but a workaround in a pinch. I too have noticed the banding decreases a lot at the higher bitrates and at 2500-3200 iso the shots are usable (noise almost filmic) where it wasn't at the lower bitrate. It would be interesting to test where the cutoff is, i.e. at 50M is the noise/banding benefit as good as at 60-65m?
  • Hi Kae,

    Thanks for the shot GOP settings (and of course, Big thanks to VK for the firmware hack!). I'm not able to get playback in camera of the 24P clips. Not a real problem on what I just shot, but it will be on a couple upcoming projects, so I must dial this in.

    The posterization reduction is very real. I was grading a shot of a china ball light right up against black duvetyne and the flare and falloff both were very soft no matter how I lifted or crushed the shots. That made me a believer. That shot would definitely have banded on the standard settings.

    I have 2 GH2s so I've rigged them side by side, hacked and unhacked with matching FD lenses for tests. I'll probably re-shoot this weekend. Hopefully I'll be forgiven shooting some charts and foliage if I can do apples-to-apples comparisons like this.
  • @cosimo_bullo is MJPEG at 30 fps? If so it will have less jello than 24 fps AVCHD...