I realise that this is not directly relevant to filmmaking but just to add more info to assist in the overall analysis of what's good or bad about the the GH1, here's my first impressions comparing a Panasonic SD5 with the GH1 - the SD5 is an AVCHD device with a similar bit rate to the GH1. (Oops, no, it's rated at only 13Mbps CBR).
Sorry, words only at this point rather than samples.
I did a simple test, recording the passing of a couple of cars past my house, then panning a bit, slowly. I used the 1080 and then 720 settings of the GH1 then the 1080 setting of the SD5, using the same card.
It's a suburban street - opposite is a handy brick wall, tiled roofs, trees and shrubs, houses, and nearby are shrubs with fine foliage in my front yard.
I inserted the card into my PS3 (using an adaptor) and displayed the result on a 42" Bravia TV.
My comments -
1080 from the GH1 is too jerky for my taste if there's fast motion (passing cars) or other than snail-like panning. Alternate frames show the car at a certain position, next frame is a re-interpretation of that frame (some reconstruction of the 'muddied' image of the wall that the car is passing across, for instance), then the next frame has the car at a new position, then the next frame another reinterpretation, etc. I guess this is the 25fps (PAL) output from the sensor, with the codec trying to make a better job of the reinterpretation of the previous frame in between the 50fps that's recorded. There are dropped frames leading to even more jerky motion.
The SD5 version is much smoother with the passing cars and the panning. No dropped frames (on the same card).
The 720 setting of the GH1 gives the kind of smoothness that the SD5 provides, but (as with the 1080 setting) almost-black becomes black (eg in the darker parts of the trees) while the SD5 continues to provide some detail there. Where the very fine foreground foliage is in the shot, the SD5 (1080) does a better job of reproducing it than the GH1 in either setting - but, the GH1 was set to a wider angle than the SD5 so I need to retest that aspect with a similar field of view for fair comparison.
The PS3 does not play the GH1 720 footage so smoothly if I play the actual file. If I play the AVCHD folder, it seems like the PS3 gets additional info from elsewhere in the folder structure and interprets it correctly. Consequently, I'd question whether the actual video files will always work correctly if downloaded from the net without the whole AVCHD structure.
Last night I wandered round the house (lit by a mixture of fluorescent and halogen lights) with the GH1 and then the SD5. Until I replayed the result on the big TV via the PS3, just judging by the SD5 monitor there wasn't much difference. But replayed properly, it was chalk and cheese - the GH1 footage was perfectly usable whereas noise seriously degraded the SD5 to the point of unpleasantness.
I guess overall I'd endorse what others have said here - for low/lower light, and for shallow depth of field, the GH1 is pretty impressive. In full light, with full HD definition, I wouldn't dream of using the GH1 in preference to the SD5 unless there was very little movement in the shot and no panning or unless I really had to have shallow DOF. Different tools for different jobs.
My guess at the problem with GH1 full HD is this. Maybe AVCHD needs a lot of incoming temporal data from the sensor to be effective. Maybe the 25fps from the GH1 sensor in full HD is not enough for it to work with. I would also guess that the sensor output is built right into its design and will not be fixable via a firmware update. The GH2 may boast yet another iteration of the sensor and perhaps the engine too - then it might work properly for full HD.
Sorry, words only at this point rather than samples.
I did a simple test, recording the passing of a couple of cars past my house, then panning a bit, slowly. I used the 1080 and then 720 settings of the GH1 then the 1080 setting of the SD5, using the same card.
It's a suburban street - opposite is a handy brick wall, tiled roofs, trees and shrubs, houses, and nearby are shrubs with fine foliage in my front yard.
I inserted the card into my PS3 (using an adaptor) and displayed the result on a 42" Bravia TV.
My comments -
1080 from the GH1 is too jerky for my taste if there's fast motion (passing cars) or other than snail-like panning. Alternate frames show the car at a certain position, next frame is a re-interpretation of that frame (some reconstruction of the 'muddied' image of the wall that the car is passing across, for instance), then the next frame has the car at a new position, then the next frame another reinterpretation, etc. I guess this is the 25fps (PAL) output from the sensor, with the codec trying to make a better job of the reinterpretation of the previous frame in between the 50fps that's recorded. There are dropped frames leading to even more jerky motion.
The SD5 version is much smoother with the passing cars and the panning. No dropped frames (on the same card).
The 720 setting of the GH1 gives the kind of smoothness that the SD5 provides, but (as with the 1080 setting) almost-black becomes black (eg in the darker parts of the trees) while the SD5 continues to provide some detail there. Where the very fine foreground foliage is in the shot, the SD5 (1080) does a better job of reproducing it than the GH1 in either setting - but, the GH1 was set to a wider angle than the SD5 so I need to retest that aspect with a similar field of view for fair comparison.
The PS3 does not play the GH1 720 footage so smoothly if I play the actual file. If I play the AVCHD folder, it seems like the PS3 gets additional info from elsewhere in the folder structure and interprets it correctly. Consequently, I'd question whether the actual video files will always work correctly if downloaded from the net without the whole AVCHD structure.
Last night I wandered round the house (lit by a mixture of fluorescent and halogen lights) with the GH1 and then the SD5. Until I replayed the result on the big TV via the PS3, just judging by the SD5 monitor there wasn't much difference. But replayed properly, it was chalk and cheese - the GH1 footage was perfectly usable whereas noise seriously degraded the SD5 to the point of unpleasantness.
I guess overall I'd endorse what others have said here - for low/lower light, and for shallow depth of field, the GH1 is pretty impressive. In full light, with full HD definition, I wouldn't dream of using the GH1 in preference to the SD5 unless there was very little movement in the shot and no panning or unless I really had to have shallow DOF. Different tools for different jobs.
My guess at the problem with GH1 full HD is this. Maybe AVCHD needs a lot of incoming temporal data from the sensor to be effective. Maybe the 25fps from the GH1 sensor in full HD is not enough for it to work with. I would also guess that the sensor output is built right into its design and will not be fixable via a firmware update. The GH2 may boast yet another iteration of the sensor and perhaps the engine too - then it might work properly for full HD.
and thats normal. 
