Made for settings that include lower than usual GOP values.
Reminder This is topic for testers and users who completely understand that they are doing. If you just happened to find it and don't know that the heck it is, go to PTool topic for settings for normal, casual suers.
driftwood, I see why your original 24L for GOP1 turned to mush. You must of accidently dropped a zero when entering the number. Like you said, you were not concentrating on 24L, just 24H.
Here's what's in the file: Video Bitrate 24H=220000000 Video Bitrate 24L=17600000
LOL, .. no wonder GOP1 at 17.6Mb/s looked like shit. hehehe
Continuation contribution I'm sorry. @driftwood Since the weather was extremely bad today, I improved the previous setting again. Can't I have it analyzed now?
Sorry Driftwood, but your patch isn't working for me. I may get one frame that's decent and then the rest is just mush. Is this the pulldown thing being discussed? Or is my GH2 somehow worse than the others?
Chaps I completely agree with you. Its a 24p N which is defo progressive, why the f*ck you have to use pulldown to get it in order is beyond me. Its the same for all GOP1 settings on 24p on the GH2 - GOP1 sticks one P frame in on the second frame at the beginning within 24p recordings. 720p does not. Can other people check to see if any dropped frames happen using fcp L&T with advance pulldown ? I need verification. Please do some tests.
@Isaac_B Thats not a bad idea. Ill run two gh2s and have a look tomoz.
@Stray & @Isaac_B It doesn't really make sense, you're right.
But since that function in FCP does not appear to be based on interlacing or a fixed pattern but rather on a motion detection method then I have a hunch it might be something connected to that. I don't really know, I'm just speculating here. For some reason it does work. However, while I haven't seen it remove any frames yet, that's not to say it can't happen so I'll be checking any clip I transcode for errors. I can always fall back on my re-order script if there's an issue with this.
@Isaac_B I have exactly the same concerns, pulldown doesn't make any sense at all as a solution to me either. I have no transcode options myself other than 5dtorgb which cannot deal with a GOP1 stream (and what premiere can give me). But I would be interested to here if anyone has found something else (other than neoscene) that can succesfully transcode a GOP1 stream. Anyway, I'm pretty sure during my brief time (er.. an hour or two) trying to play with GOP1 I managed to get Premiere to output a sequence of Tiff stills from a GOP1 stream. So.. you could try that and then reimport the sequence of stills and export to another Intra like format.
24P on the GH2 is already progressive, so not sure why an advanced pulldown would work. (Pulldown removal is unnecessary on native 24P footage) I'd be concerned that the advanced pulldown is removing incorrect frames of footage. Does a simple transcode not fix the issue?
If you want to test for what it's doing it would be helpful to take two GH2's, one unhacked and the other hacked with gop 1 settings. Both locked down and filming. You could overlay footage and check the gop 1 footage with advanced pulldown removal to see if it matches or if the pulldown is dropping your footage.
You have to use 2:3 pulldown. Don't know why, probably because of the rogue p frame is making it interlaced? I dont know why... 24pN should be progressive. Anyone? Anyway to transcode properly: For example, If you have fcp7, in Log & Transfer, check preferences in AVCHD, check import to prores HQ or your choosing but make sure you check the box 'Remove Advance Pulldown'. Stick it on a 24 timeline. Done. Works a treat.
driftwood I am so glad to use your new "GOP1 220M AQ1" hack today with TOSHIBA white Card (16G C10); But I see that it have obviously juming some like that when I playback with KM Player or TMT5 player; Could you pls give some advice for normal playback?
Just been shooting outside with an array of shutter speeds using my new GOP1 setting - hasnt crashed once! Seems very good on IQ still with the limiter in place.
@sam_stickland I am not sure but I think that look is happen with the specification of Streamparser? That will also be lost if the length of a time-axis is changed? Although it is visible also in other persons' result ...
@bkmcwd Looking at the first screengrab, what's according between frames ~270 - 340, ~438 - 505, and ~605 - 675? Is it representative of something in the stream?
@driftwood, latest 220 setting plan to test, just a question, the rest of the settings does not matter? thanks for all the work you do, I think from the whole community!
This is the 3GOP stable setting combined 88Mbps and 66Mbps which I made. (Of course, there is still spanning issue) As long as I tried, there is no problem in camera playing and the cadence problem has not occurred, either, inclueding 80% mode. Since I prefer medium size file, this is very convenient.
Incidentally, except 24H and 24L, it is the Chris's former setting. (Although I have not tried ... )
Since I thought stability as important, I am not sure for the quality of this to be how...
About attached snapshots, 24H,24L,24H80%,24L80%, the places which are valleys in the charts are the portion which moving camera purposely and quick focus out, in.
EDIT: If you test with Pappas death chart, in this setting, it may become an error. Although I think that it is satisfactory practically, if you change into QV=24 as the second ini file, it will be stabilized further.
EDIT: Sorry! Furthermore, when testing, it became clear that it was not stabilized unless it is 26 as the newest patch showing. Scatterbrain...
I tried the AVCCam Log and Transfer in FCP7 and it seems to work :) need to do some more testing though - but great news for us Mac folk. PC users can still use my workaround.
Interestingly, in VLC it plays back correctly, but if you go frame by frame (hit the E key) - it has the same frame order issues as everything else.