GH1 720 60p & 30p to 24p vs. HPX 170 24p
Rate Conversion Tests
Mad Props to Luis Caffese for running out last minute in the dwindling sunlight and shooting this with me on his HPX-170.
And many thanks to Ninja HD Guru Barry Green for his help in figuring out how to set up the test.
5 min vid.
Click Image to go to Vimeo.
Orig file available for download there.
Tip: don't be afraid to turn HD off to eleminate possibility for streaming induced stutter vs. cadence issues.
I won't comment on the efficacy of each conversion method at this point (though I do have my opinions), but will point out some non frame rate stuff I found interesting:
GH1 handily wins for lattitude (HPX was a little over exposed in some shots, but you can see the latitude edge in all of the shots)
Even with the HPX 170 in 1080 mode and the GH1 in 720, it wins for resolution as well. Note the trees. But also note the murky gray pavement in the GH1 AVCHD shots. In 720 MJPEG shots everything is sharper than the 170's 1080 with no murkiness including the gravel. This is really interesting since in the AVCHD mods which are Long GOP, the still pavement should be where the codec excells and the blowing dancing tree leaves should challenge it, and yet the opposite seems to be true. I would suspect I had a big smear of sweat on my ND filter in the pavement if it weren't for the fact that the MJPEG (Intra frame shots) are smearless.
WHY NOT JUST SHOOT 108024p all the time? What's the point of these tests?
In 1080 24p because of the lack of B frames the camera is trying to predict frames as you shoot, and if you get a ahead of the prediction with your camera movement - you can get some macro blocking or "mud".
This problem is significantly reduced in 60p AVCHD making it suitable for most applications where 1080 24 might fail, and non existent in 720 30p MJPEG an intra frame codec. The good news is that 60p goes to 24p perfectly or near perfectly. 30p is a bit more of a challenge and should be avoided though it can still be used when mud is unavoidable. The good news here is that if you have a smooth moving subject or shot which would really reveal the stutteryness of 30p to 24p, you won't need to use 30p.
Rate Conversion Tests
Mad Props to Luis Caffese for running out last minute in the dwindling sunlight and shooting this with me on his HPX-170.
And many thanks to Ninja HD Guru Barry Green for his help in figuring out how to set up the test.
5 min vid.
- 4 conversion methods using Cinema Tools, Compressor, and FCP
- GH1 shoe mounted to the top of an HPX 170 for a similar frame
- Shot cars crossing the frame since they move at a fairly similar rate and any cadence jumps, stuttering should be apparent.
- All 20-faux p referenced against HPX 170's 1080 24p
Click Image to go to Vimeo.
Orig file available for download there.
Tip: don't be afraid to turn HD off to eleminate possibility for streaming induced stutter vs. cadence issues.
I won't comment on the efficacy of each conversion method at this point (though I do have my opinions), but will point out some non frame rate stuff I found interesting:
GH1 handily wins for lattitude (HPX was a little over exposed in some shots, but you can see the latitude edge in all of the shots)
Even with the HPX 170 in 1080 mode and the GH1 in 720, it wins for resolution as well. Note the trees. But also note the murky gray pavement in the GH1 AVCHD shots. In 720 MJPEG shots everything is sharper than the 170's 1080 with no murkiness including the gravel. This is really interesting since in the AVCHD mods which are Long GOP, the still pavement should be where the codec excells and the blowing dancing tree leaves should challenge it, and yet the opposite seems to be true. I would suspect I had a big smear of sweat on my ND filter in the pavement if it weren't for the fact that the MJPEG (Intra frame shots) are smearless.
WHY NOT JUST SHOOT 108024p all the time? What's the point of these tests?
In 1080 24p because of the lack of B frames the camera is trying to predict frames as you shoot, and if you get a ahead of the prediction with your camera movement - you can get some macro blocking or "mud".
This problem is significantly reduced in 60p AVCHD making it suitable for most applications where 1080 24 might fail, and non existent in 720 30p MJPEG an intra frame codec. The good news is that 60p goes to 24p perfectly or near perfectly. 30p is a bit more of a challenge and should be avoided though it can still be used when mud is unavoidable. The good news here is that if you have a smooth moving subject or shot which would really reveal the stutteryness of 30p to 24p, you won't need to use 30p.