JELLO question???

J

JerryB

Well-known member
I have been following the hack since day one and am overly impressed with what has been done with the improvement of the image and added functionality (hopefully AGC and ext. monitoring are on way!!)

What are the effects of the hack (raising the bitrate) in regards to JELLO? I realize the dreaded "jello" effect is due to the way the sensor captures the image rather than settings, but I was curious if raising the bitrate would require the camera to process more information thus possibly creating more jello.

May have no effect (or possibly a positive one)??? Just curious if anyone has had first hand experience or any insight or ideas on how to improve.
 
The jello is timing based. It's a fixed number that cannot be changed unless you increase the timing of the processes that control it. This *could* be done but in all likelyhood the camera was designed around a specific timing and budget constraints usually dictate using the cheapest parts that allow you to get your result thus it is probably near it's maximum.

This wouldn't get rid of jello, only reduce it. The only way to get rid of it would be post processing of the image file that changes the justification of the sampled rows to match them all equally. This would have the effect of picture lag but the output would be jello-less. Don't get your hopes up though, this would be DSP intensive and most certainly can't be dealt with in the GH1. I'm betting it's fixed in the pro camcorder model this way though.
 
Jello effect (aka Rolling Shutter) is much less pronounced with 720/50p shots. The thing is that the CMOS sensor can be only read sequentially, from top to the bottom. The camera has the time of exactly one frame to read all the data, then it must capture the next frame.

In 50p mode this time is less, therefore the jello effect is much less obvious. In 25p mode the camera has more than twice the time and it does seem to take its sweet time reading that CMOS sensor, resulting in visible jello.

While experimenting other issues Vitaliy did some hack to change 720/50p to 720/25p. I guess he changed the sensor's output from 50p to 25p. I am curious if the opposite is possible: change the 1080 mode sensor output from 25p to 50p, then drop every second frame to keep the real video file output at 25p. With the faster sensor speed the CMOS read would have to be faster and jello may be reduced.

Obviously the NTSC version would be even more tricky, if at all possible, as you cannot just drop frames (sensor output 24p -> 60p, drop every 2nd frame: 60p->30p, not 24p).

I am just guessing here, probably the whole thing is only theoretical :( At least until Vitaliy says otherwise :) After all, a few weeks ago we didn't really believed that the 17mbps bitrate can be uped to 40mbps :)
 
By the way, VirtualDub with DeShaker plugin does a fairly good job at reducing jello. At 720p it produced perfect results for my shots (no jello at all). At 1080p it was not perfect, but much better than the original MTS files. I used a shutter correction of 60%.

I also tried The Foundry Nuke with RollingShutter plugin and Vegas MovieStudio with NewBlueFx RollingShutter, but those seemed to be inferior to VirtualDub+DeShaker.

To be fair, Nuke+RollingShutter has the edge when there is also movement in the picture in addition to the shake of your hand, but it is incredibly(!!) slow (0.2fps), not user-friendly, and does not natively handle MTS files.
 
you may be right im forcing it..
"slow" pan would be changing person to person, so need to put the video ehre, i dont know how..
i ll try more natural scenes..
 
You must log in or register to reply here.
Back
Top