Possible breakthrough on Native 24p!

Vitaliy,

I tested a bunch of different modes to try and determine what the 9000000 vs 18000000 constants mean. Nothing changes when I switch between FHD, SH, H, and L modes. This leads me to the conclusion that it might be related progressive vs. interlaced (wrapped) frames, using one value for 60p frames and the other for wrapped 60i frames. Any thoughts?

Chris
 
We need to move lower in AVCHD analizer to understand this empty or almost empty frames.
I can understand why frame can be almost empty after key frame (in static situation).
But why last frame in GOP have some problems is not easy to get.
 
cbrandin said:
Vitaliy,
I tested a bunch of different modes to try and determine what the 9000000 vs 18000000 constants mean. Nothing changes when I switch between FHD, SH, H, and L modes. This leads me to the conclusion that it might be related progressive vs. interlaced (wrapped) frames, using one value for 60p frames and the other for wrapped 60i frames. Any thoughts?
Click to expand...

May be, it is hard to tell.
I am working of some things to allow speed up and improve testing.
It includes IDE modification, special IDC script, plus ability for PTool to use external patches (and use large arrays for them also).
Some file operations will be also opened in next database release.
Plus one safe location (routine that is executed upon Bracketing menu item selection).
 
In the 720p clips the 192 byte frames are devoid of image data. My tests are mostly, but not entirely static. When I look at the clips frame by frame there is absolutely no change - and there should be some changes. I test with a very slow pan on a static scene. I can't explain why they occur after an I frame instead of before, but I assure you they shouldn't be emply in either case. They happen with some regularity in both cases, so I don't think the Native 24p option is causing them - it's more like the wrapped 24p is hiding them.

Chris
 
cbrandin said:
In the 720p clips the 192 byte frames are devoid of image data. My tests are mostly, but not entirely static. When I look at the clips frame by frame there is absolutely no change - and there should be some changes. I test with a very slow pan on a static scene. I can't explain why they occur after an I frame instead of before, but I assure you they shouldn't be emply in either case. They happen with some regularity in both cases, so I don't think the Native 24p option is causing then - it's more like the wrapped 24p is hiding then.
Chris
Click to expand...

In this case we must approach this professionally.
Look only on unmodified 720p footage. And strictly document testing procedures and provide short clips as well as analizing results.
After this we must publish open letter to Panasonic on one of the large techno sites. :-)
 
I wonder if you shoot 24p in 60i wrapper and then remove the pulldown, whether you also see empty frames? As you say, the wrapper can be hiding empty frames, but they should reappear if unwrapped. Anybody tried this?
 
I don't know how you could do this. The best I can come up with is to process it with NeoScene and look frame by frame for non-changes. NeoScene will transcode, so looking at the output with an analyzer won't tell. But if you look at the transcoded clip with an editor you should be able to see if nothing changes from one frame to the next.

Chris
 
cbrandin said:
Did you get the failure with Navive 24p checked? Stream Parser shows a maximum bitrate of 110,297,900 bps in my tests. Does that match what you got?
Click to expand...
Yes, my tests were with Native 24p at 86M Overall Bitrate. Stream Parser showed a max bitrate of 110M.
 
Yes, sorry, I forgot that none of the pulldown removal programs put things back in AVCHD, so the analyzer doesn't work. But, I think your suggestion would work. We would need to put it through NeoScene, then look for repeated frames. Unfortunately, I don't have NeoScene, otherwise I would look at this. Thanks for putting up with a suggestion from a noob.
 
Lpowell,

Can you tell me a little bit about your test setup? I would like to try to replicate it as best I can.

Chris
 
I tried this with
Video Bitrate FHD/SH: 50000000
Overall 86000000

Ran before an after with my old ptools settings which were 62,66,68 and I seem to get the same bitrate shooting the same scene.

Could not get new setting to crash with my patriot lx
 
cbrandin said:
Can you tell me a little bit about your test setup?
Click to expand...
I put the GH1 with a Canon FD 28mm lens on a tripod and point it at dense shrubbery. Then I record 8-sec FHD video clips on the Class 10 Sandisk at 1/60 shutter, f8, and 100-400 ISO. Two out of six attempts failed with a write-speed limitation error, producing a zero-length file.
 
cbrandin & lpowell;

Just a FYI;

I Tried the following setting :


Get lpowell's latest Tool and 50M setting
http://www.dvxuser.com/V6/showthread.php?t=222791

load setting I then play with the following;
Check 24 fpp native
Overall = 120,000,000 (the Max we can set, it goes with class 10 card)
FHD= 51 then 53 (both looking good, over 53 will turn the rate back to lower rate)
GOP=6 then 4 (default 12 is not as stable as 6, 4 seems more stable than 6)

shot PAPPASART's motion chart more than 60 sec a few times with GOP=4 without problem (produced rate at 46-47)

just let you know put you guys's talent together , it has produced a test result that has never been so good before.

You guys and Vitaliy deserve another round of applaud

Cheers.
don
 
cbrandin said:
My tests are mostly, but not entirely static. When I look at the clips frame by frame there is absolutely no change - and there should be some changes. I test with a very slow pan on a static scene.
Click to expand...

lpowell said:
I put the GH1 with a Canon FD 28mm lens on a tripod and point it at dense shrubbery. Then I record 8-sec FHD video clips on the Class 10 Sandisk at 1/60 shutter, f8, and 100-400 ISO. Two out of six attempts failed with a write-speed limitation error, producing a zero-length file.
Click to expand...
Chris, while I haven't tested with new OBR setting yet, it has been my observation from prior testing that "native freeze" happens on very static scenes. I.e. camera locked down on a tripod and pointing at a motionless or slightly moving subject, with majority of scene in focus motionless. Hand-holding or panning the camera, however slowly, let's one dodge the bullet and avoid "native freeze" (in my experience).
 
Hey guys,

one quick question about testing procedure with Pappas chart of Doom. I have a 24inch monitor and the way i test is I open the chart at 100% and have the monitor fill up the entire frame on the GH1. I have just gone through almost all the suggested settings here and not one gave me a stable 1080p (24p or 60i). Most just crashed after 1-2 sec. So, how is everyone testing? Why is there such a giant difference between cams? I mean not being able to make it crash on Pappas chart? Really? I'm with a Class 10 Sandisk Extreme ... very weird.
 
Just tested donwang's settings and it crashes immediately no matter what scene I have. 720 just crashes while 1080p freezes the whole cam. Something is fishy here. I suggest if we do test we should give all the information necessary to replicate the experiment - setting file for ptools, camera settings, testing procedure. All of this is pointless otherwise.
 
Guys, I want to warn you that this is thread in development section.
So, if you want to start testing and reporting here, better make new thread in general hack section.
 
You must log in or register to reply here.
Back
Top