Official Panasonic GH3 topic, series 2

RRRR

I was just comparing some gh2 footage with the 50mbps mov. Different lenses, subjects and all but: I´d say there´s a 10/15% drop in resolution... (actually the gh3 footage in this instance reminds me of a little better 5d mk II)

And while "a bit soft" isn´t necessarily bad, then there has to be enough pros in there to make it worthwhile. And the above discussed issues definately needs to be manageable!

amateur

Hands on : http://www.expertreviews.co.uk/digital-cameras/1295134/panasonic-lumix-dmc-gh3-review-hands-on

http://www.trustedreviews.com/panasonic-lumix-gh3_Digital-Camera_review

bg2b

After looking very carefully, I did find a little bit of moire in Genesis. The most noticeable is when the car pulls out around 0:41 on the water tower(?) in the background, and I think there was also a bit on a window screen (maybe around 1:30; I can't view the downloaded version here to check). Overall, it didn't seem that much.

From the Panasonic interview on Imaging Resource, it sounds like the GH3 does 2x2 binning on-sensor, then reads out, and then downscales further. Possibly the latter stage could be somewhat flexible in firmware. I don't recall any official statement about what the GH2 does on-chip, but I can't imagine it's reading out the full sensor and then downscaling. Certainly the GH2's correlated stuck-pixel patterns suggests that it also does something with 2x2 groups on the sensor.

christianhubbard

every single time I watch genesis I see the moire in the numerical keypad outside of the apartment. even on the 1080p source file.

really hoped that was minimal and they would fix it before release, but judging from all this footage i'm not sure about that... :/

amateur

1440x1080 sample from http://www.photographyblog.com

if a firmware or a hack could make a video frame look like this it would be beautiful. see no moire in the hats girl! and it is actualy a downscaled like you guys know

(don't even know if All-I 72Mbps is able to produce frames like this) NO CONSIDERABLE MOIRE IN THE UMBRELLA'S MISTER !

DrDave

Moireee. Blech. A filter might help, but I have to say if you look at cams that have it, they still have it. On the bright side, my GH2 just increased in value.

Aria

It seems that there's a good chance that the Moire can be avoided with a better understanding of just what causes the GH3 to produce it. It doesn't seem to be ever present but rather on some parts of some shots. Far from the worst i've seen on video footage. It just makes me think that there is possibly a way to shoot with settings that will minimize the chances of it popping up. Can't wait to see what the production models actually do.

pundit

Yeeech! The fucking moire is simply killing the GH3. I've seen enough. 2nd GH2 coming next week before the prices go up as the stock runs out.

sunboy

If you look at the samples you can clearly see oversharpened edges everywhere. That makes the video look very videoish and despite that there is still a relatively bad resolution of details. Seems there is no way to get a natural looking image as decreasing the sharpening in camera will most probably simply make the image unsharp, not soft in a positive way.

RRRR

@sunboy

Resolution definately is lower than gh2. However, I think that images I´ve seen recorded with in-cam sharpness dialled down to a maximum has looked pretty good, and it´s not hard to create an appearance of sharp footage in post. It won´t be as handy as the gh2 but it might have it´s uses. I´m thinking of low light situations, closeups.. But there might be other situations as well.

brunoalmeida

Some new samples:

I chose the GH2 because of the resolution, if the resolution becomes equivalent to the canon or sony, panasonic loses the brand differentiator.

Vitaliy_Kiselev

@pundit @sunboy

Guys, stop this attack of hysteria.

PhilMyself

From the original files, I think it looks ok for me. The only problem I see is that the test are always done with the 12-35 lens. I have one and I can tell you that you should not use the AF while filming because it creates a lot of mini "backfocus" that make picture look bad. The filming set provide by panasonic are also very poor in lightning !

John_Farragut

Apart from the aliasing and moire, these samples show that the video is very clean. So the codec is holding up. Also the uniform walls seem to do fine. A more static camera will do better.

PhilMyself

I have downloaded the original and convert in proresHQ to do some grading. It holds very well. Don't know if I can post a video sample in response...

Original file is poorly shot (white are "overblown", video signal looks very bad !)

When graded in correct norms and a warmer look it's ok...

So let's wait for a better shooter with a better lens and a good eye !

Tron

LMFAO... Another hilarious GH3 freak-out moment. I hope you all cancel your preorders so I move up higher in the queue :)

So far I have only seen moire (without false color) with the 12-35 lens, which seems to indicate that sharpening or CA algorithms could have something to do with it. Hopefully someone can moire test with a Nikon 50 or Samyang 35 to help determine it is not related to the downsampling or AA filter. The codec holds up very nice and seems to buck banding very well which is my main issue with the GH2.

svart

You guys doing the bitching do realize that you are only causing others to doubt their purchases, not "force" panny into changing anything, right?

And for sensor output, it's doubtful that the downsampling is done in hardware. It's usually done in DSP after the sensor outputs raw A/D samples.. So yes, this could possibly be fixed, but you either get sharpness for pictures or you get no moire/aliasing for video. You can't have both. Yet you asked for it.. So you got the one panny thinks will sell.. Great going guys. Your constant bitching about features ruined this for us!

onionbrain

The mosquito noise you're seeing on the “All-I” footage is exactly what you get with ALL-I on the 5D Mark 3. Clearly this relates to lower bit rate intraframe – but I'll leave it to you encoding experts to figure it out.

The lenses used in these sample clips aren't necessarily the greatest – they're f/2.8 (I think) and then you factor in the additional light loss for MFT, etc.

I've seen plenty of GH2 sample footage that looked worse than these clips.

So – what these clips tell me – is that if this camera can be hacked, if, then increased bit rates and whatever else may push it to very impressive levels. That the stock camera is produced to create intraframe footage along with the multiple processors – means it may have been designed to handle much greater loads – like higher bit rate hacks. If so – the GH3 will be a winner – in six or eight or twelve months when hacks (if) approach maturity.

onionbrain

Also, regarding these sample images -- we have no idea how they were encoded for web delivery. Single pass, standard or low profile -- and a dozen other things could have easily screwed up the image files we're all attempting to pixel peep. Did any of these Vimeo uploaders specify that they uploaded the files directly from the camera?

The only one that appears to be camera-direct is that water fountain shot -- that exhibits the mosquito noise you'd see on ALL-I from a 5D Mark 3. Did anyone analyze that clip to see if it was native 25p, or was it laid (flagged or unflagged) on a 50i stream?