AVCHD high bitrate tests- MUD FREE footage

I just tried to torture test the D settings using a sandisk extreme class 10 card. I shot steadicam of a field of tall grass at 125 shutter speed.

I got about a minute in each time before I would get the memory card speed lockup. I have a couple of different brands of class 10 cards coming tuesday I will test them out this week.
 
Oedipax said:
Maybe it's because the framerate is coming out at 23.971 right now rather than 23.976 and that's confusing FCP? Try conforming one of your L&T'd clips in Cinema Tools and see what happens. I've been conforming to 23.98 in After Effects using the original MTS files and everything seems to be fine.
Click to expand...

Oedipax,

Tried the Cinema Tools Conform route and unfortunately it didn't change anything. Right now the only way I have been able to get the footage to be exactly 23.976 (which it appears FCP wants) and to get rid of the Upper (Odd) field dominance, is to take each clip and open them in Quicktime and export them as ProRes 422 (HQ) at 23.976 fps clips. That works, but I am always concerned about multiple compressions.

I blew up a before and after sample of one clip and I can't detect a quality difference, but again, I am very leery about too many compressions.

Wish there was some way to get it exactly 23.976 out of the camera, because otherwise, the quality is really smoking.

Best,
-Tim
 
nathankw said:
I've just been trying out the higher bitrate AVCHD and I've got an odd result....

But looking back at my footage I've found one odd moment where the codec seems to hiccough for no obvious reason.
...

As you can see there's a sudden loss of detail and artefacting all across the frame.
Although its wide and sharp so there's lots of details there's very little movement and
nothing obvious to make the codec lose it. The image recovers over a couple of frames
and then exactly 13 frames later the same things happen....

so is this some kind of overheating/overloading problem?
Click to expand...

I don't think so. I think what you're seeing there is the "I" frame in the GoP, the first frame of a group of 15. The way a long-GoP codec works, the first frame is an intraframe, like a JPEG picture. Then every other frame in the group is made off of the changes it can detect from the first frame. So if there isn't enough bandwidth allocated to making the "I" frame, it'll be muddy like that, and then the codec will use the next frame or two to bring it up to full quality. And that sounds like exactly what you're experiencing, yes?

If there were a way for tester13 to increase the amount of bandwidth that can be allocated to an i-frame, then you should get better overall quality if you had a better-quality i-frame and a longer GoP.
 
Barry I have a question. I read that AVCHD is supposed to go up to 24 mbits per second. Wouldn't trying to make it go much higher than 24 mbits be detrimental to it in any way? I think the B option at 32 mbits be more within what it should take. Your expert opinion please. Thanks.
 
AVCHD, as a defined format, is limited to 24mb/sec. What tester13 is doing here is exceeding the specifications of the format. Is that "detrimental"? I don't think so, but it is nonstandard.

With a codec, the question is usually: what's the right balance between file size and image quality? The smaller the files, the longer the recording times and the easier/faster to offload and archive. The larger the files, the better the quality, but the longer it takes to offload and the less recording time you get per card.

At 32mb/sec, you're choosing a moderate increase in file size, a big boost in image quality, and probably avoiding any card compatibility issues with any Class 6 or above card. For my tastes I'd love to have 100 megabits, maximum quality, etc., but the problem would be card compatibility (Class 10 only "officially" supports up to 80 megabits).
 

AVCHD B setting. a test shot. i wonder why when i convert MTS files to voltaicHD, and when i dropped it into FCP, it just turns out to be 25fps. where am i wrong?
 
I dont know if this has been addressed yet but Im having a problem with the length of the footage. I record 10 min in the camera and convert it with TMPGnc and it becomes 37min. Windows media player plays it back at normal speed but says its 37min in length. And this footage crashes my After effects CS3. It does it with the B and C setting and with 24 Native and without.

UPDATE: turns out tmpgnc was the problem
 
Last edited:
Most newer NLEs will edit the native files. Vegas, Premiere CS4.2/CS5, Edius... All edit native mts files without conversion.


Dang Car3o beat me by... same exact time!
 
Now delete your post. /Sarcasm.

On a more serious note, I'll be testing 3 settings today that people have had good results and bad results and will report back here with the footage. Everything will be recorded on Ultra II. Each clip will be 1min 30 seconds long with pans and lots of detail with GOP settings changed.
 
Hi Everyone!

I would like 720p/60fps with reasonable bitrate that does not crash even with intensive movement, which settings should I use?
 
You must log in or register to reply here.
Back
Top