Summary of my findings (and video) of GH13 and GH2 ("strobing", detail, ...)

[PJF]

I did some other tests where I filmed the same (static) scene and quickly swapped lenses (to have the same lens on GH13 and GH2). Very similar results.

In this context of an "internet look see" (i.e. not a scientific test) - that's good enough for me.

My comments were not intended to dispute your results, they were suggestions to make an interesting comparison more complete (sorry if I came across as churlish). The tandem camera mount is an excellent method of eliminating scene discrepancies. With other discrepancies removed, diminished or compensated, tests like these can be very useful additions to the record.

Thanks again for your efforts. I understand work pressures!

 

[DrDave]

I tested the two side by side for a week indoors, outdoors, sharp primes, Panny 20mm, and so on, and under certain conditions the GH13 was better, and in others, the GH2 was better. In most cases anything from the GH2 that "looked" sharper had sharpening halos, but the sharpness was compelling, and from a few feet away the artifacts disappear. I would not bet that I could tell them apart in a double blind test, which is, of course, what no one has done. The colors on the GH2 often looked a bit off, but one has to use a white card anyway. And so it goes.

 

[mimirsan]

And I am talking about the hacked GH1 too! You don't think the folks at Panasonic looked at the hacked version? Then gave the GH2 Native 24p, 24 mbits, better sensor, etc.? Please!

I dont really think panasonic gave two $$$$ about the hack. Give them credit though they listened to the 24p demand.

 

[crunchy]

Having the B frames seems to improve this like in the Gh2 and HMC 150 version of AVCHD

In my opinion, B frames, by themselves, cannot solve problems with mud. If this would work, then lower GOP sizes would already solve the problems. Moreover, each I frame should be entirely mud-free (since it contains the original picture), but it isn't.

I suppose that there exists some kind of internal threshold or parameter (within AVCHD coding routine) which controls a level of filtering (mud) in "not so bright" areas before doing actual coding into AVCHD. Probably such kind of filtering is not so "strong" in GH2 or some other cameras than in GH1?!?

 

[Blackout]

That is very true, no amount of bitrate will improve the areas of the video that are being processed by the Gh13 as "not requiring many bits". Green leafy details in shadows or near the edges of the frame seem to be muddy. Having the B frames seems to improve this like in the Gh2 and HMC 150 version of AVCHD.

R

I am not talking about very dark areas. I am saying "wide shot: busy city street" in the shadow areas (not dark, just less bright than other areas) on the gh2 I had and the footage I dled I see things like peoples faces turning into little blocks and wall textures of bricks turning into mush or macroblocking on the gh2 with busy scenes.

I do not see any of that on the gh13 with my current hack settings.

Times square - the golden magic hour.... wide lens.... people bustling all over.. please.. gh2 people show me this or any bustling city or crowd scene like it and the wider and more detailed movement the better. Do a tripod lockdown wide.... do a pan... do a handheld slow move.... let us all look at it... what I found with the GH2 and these type of scenes was a lot of issues... none of these pieces of footage push the codec at all. If you are just shooting locked down close ups of flowers sure... no problem... GH2 all the way.

The GH2 does too much artificial sharpening which to me looks like it makes the aliasing worse but on certain shots it may appear to make the image look better.. this is a matter of taste though... I don't think it resolves more resolution.

 

[Ian-T]

Halos and sharpening artifacts are one thing (I think it's just a matter of user settings) but from the samples I've seen it does resolves detail more than the GH-1. The GH-1(3) got a boost from increased data rate...but still.

 

[rambooc1]

In my opinion, B frames, by themselves, cannot solve problems with mud. If this would work, then lower GOP sizes would already solve the problems. Moreover, each I frame should be entirely mud-free (since it contains the original picture), but it isn't.

I suppose that there exists some kind of internal threshold or parameter (within AVCHD coding routine) which controls a level of filtering (mud) in "not so bright" areas before doing actual coding into AVCHD. Probably such kind of filtering is not so "strong" in GH2 or some other cameras than in GH1?!?

Agreed the mud is still there in the I frame in underexposed areas probably from the way the variable codec allocates pixel blocks like 4x4 to 8x8 16x16 etc. This explains the macroblocking in the "not so bright " areas that the codec allocates less data to. I read somewhere the B frames have more to do with motion rather than "not so bright"areas.

Blackout, those macroblocking in faces from the GH2 "busy" scene will probably benefit from a higher data rate when it gets hacked. I suspect the same blocks are there in bushy green scenes, just that it's hard to see them in that situation.

R

 

[Ian-T]

Blackout, those macroblocking in faces from the GH2 "busy" scene will probably benefit from a higher data rate when it gets hacked.R

I also think it will. When comparing a split screen HDMI capture with the AVCHD (from another post) it was like night and day (no pun intended). The dark areas kept its detail while the codec smeared it out. A higher bit rate would definitely help.

 

[emory]

I dont really think panasonic gave two $$$$ about the hack. Give them credit though they listened to the 24p demand.

Panasonic didn't give two $$$$ about the hack? Then why did they make the later GH-1s unhackable? And as far as the GH-2 is concerned: as yet they haven't listened to the 25p demand from PAL customers. Sorry, but I just couldn't let your posting go unanswered.