AVCHD: Extreme settings

otcx

Kae and Butt could you put the your latest settings with Ptool 621, that I could test them.
Incam with 65 mbs dose not work in my camera, maby because Trancend class 10 SD.

stoney

88Mbps GOP3 AQ3 didn't pass my bush of death test. As Vitaliy suggested, I'm going to double the GOP size. Hopefully the higher B-frame bitsize does the trick. That would be GOP6 for 24p and GOP12 for 60p.

Dusty42

@OSGondar

Based on the settings you posted on the 16th, can you play the clips in camera? What type of cards are you using?

OSGondar

@Dusty42 yes they played in camera just fine however in general anytime a clip doesn't play just restart camera =) and it will, clears buffers or something i'm not sure but it works.. Right now i'm testing pushing bitrates without touching the buffers since i keep hearing they may not be critical. I am working to confirm this.

Edit* My card is a sandisk 32gig extreme class 10

cbrandin

True, the buffer settings are not critical, at least not from a stability standpoint. From an encoding standpoint they do have a significant effect. The ratio of I frame sizes vs. P and B frame sizes changes quite a bit, and that could have a big effect on image characteristics - especially with motion appearance.

Chris

cbrandin

By the way, you don't have to set the buffers to 3x if you are setting the bitrate at 66M, for example. You could set them at 2x, or leave them alone. One thing, though - you do have to set them all at once - bad combinations will cause the camera to crash - requiring battery removal.

You can see the effect the buffer patches have by looking at files with StreamParser. If you shoot something at 24H, for example, without the buffer patches, and then shoot the exact same thing with the buffer patches, in many cases you will see a difference in the streams. It doesn't show up with all types of shots - highly detailed ones mostly.

Chris

OSGondar

@cbrandin to clarify, thats: FB1, FB2, With the1080 24 Frame limit? = all of them.

Thank you for the input i would actually prefer to have the ideal 66mbs settings for now since my card is only a 32gig and i'm already at hardly an hour run time from the 3hrs+ stock gave me. Its not bad for film making but i also almost always carry my camera on me casually as well. Between raw photos and high bit rate videos the 32gig is looking on the small side these days so ill hold off on the 88 + for now but i want the best 66mbs can offer at the moment.

cbrandin

@OSGondar

That's right.

Chris

cbrandin

A comment about testing extreme settings... My wife has shown great kindness, having an interest in plants with extremely small and fine details. I have a rather busy back yard that works better for stability testing than any chart I have. One thing, though; If I test hand-holding the camera, it will happily grind away with no failures for as long as I am willing to hold it. If I lock it down on a tripod, and set focusing mode to AFS (that avoids any focus hunting), the camera will fail almost instantly if anything isn't right.

Chris

cbrandin

@OSGondar

One more thing about the buffer settings. If you end up with fractional settings for the buffer settings you should round up the FB1 and FB2 settings, and round down the 24 frame limit setting. This is because the FBx settings set up buffers that the limit setting uses. If they are even a little bit too small, the camera will crash.

Chris

LukeV

@cbrandin
Hello I've been using 66mbs -GOP3 without touching buffer size for the same reasons OSGondar wrote. I'm on a Mac and not able to use StreamParser.
Are you saying that increasing buffer size gives a better out come? Under the hood.
Thanks for all your work & Happy belated B'day.
Luke

cbrandin

@LukeV

I tested it only with 24H and the buffer settings produced streams that looked more like high-bitrate versions of the factory firmware - as far as the size of I frames vs P and B frames is concerned. A haven't done any low GOP tests. I understand that the buffer patches do make the B frames bigger. It will take more testing to determine whether that leads to a better image or not. With 24H-GOP 12, I think it does.

The Guys on the low GOP thread know a lot more about the low GOP stuff than I do.

Chris

LukeV

@cbrandin
Will do Thank you

Luke

ivan858 · Screen shot 2011-08-17 at 8.13.08 PM.png

@stonebat Hi first post here, I applied your 88bps settings with 6GOP. I'm getting some macroblocking flickering in and out when the camera is in motion.

LPowell

@ivan858 I've seen alternating 6-frame GOPs like those displayed in your second Stream Parser screenshot above. I'm very interested to find out what causes and/or cures this aberration.

plasmasmp

115mb gop3 is amazing. It only works on my best class 10 cards, and to get it stable I need to leave on the lens cap for 5-10 seconds after pressing record. 80% mode always crashes even when giving ten seconds lead time with the lens cap.

ivan858

@plasmasmp I just saw your eye video, impressive detail! Have you tried any other GOP settings with the 115Mbs?

plasmasmp

I haven't tested other GOP settings yet, still testing the limits of 3

ivan858

@plasmasmp i'm not having any issues with 100mbs at 6gop at all, mounted..no lenscap stabilization needed :P, movement is a different story. i have yet to try 3 though.

@LPowell do you think the 6gop changes you mentioned may be due to buffer limits?..i tried forcing the b and p frames to change by messing with the detail/lighting but not success in a pattern change.

blues

am doing 110 and 88, with GOP 3 and 4 more detail.....am getting average 50 mbps rate

so far so good, will try to go higher like 120

Ausboy

Here the settings I am using - mixture of a few contributors including LPowells Mjpeg - use manual lenses and class 10 cards. The odd write error but the quality is off the charts!!!!!!!

stoney

@ivan858 I'm trying out GOP6... later today. The 88Mbps setting failed my bush of death test. So I'm moving back to 66Mbps. Bye Extreme~~~

stoney

Hmmmm.... just one more thing.

@ivan858 can you try the same test with AQ1?

ivan858

@stonebat 88mbs 3200iso 6gop 1AQ. Plays in camera. no macroblocking.

I was playing with 100mbs last night, with just the 6gop and 1AQ checked, and it was behaving just as well. is anyone testing anything higher at the moment besides plasmasmp?

sSTEV

Hey everybody, I wanted to see how the FB1 and FB2 for 24p and 60i values are calculated

@OSGondar Thanks

Howdy, Stranger!

Categories

Tags in Topic

Top Posters