What lens were you using on the GH3?

@maddog15 I'm experiencing same thing in terms of noise. Sometimes I get extremely noisy and crushed/harsh/dithered shadows, even at low ISO. Look at the detail attached. This was shot at 800 ISO (not so low I know, but not even so high) on a Tokina 11-16, Natural, with settings down to -5, -5, 0, -5. And that noise is really "fizzy". It was 50p anyway, but I experienced same thing even at 25p 50Mbps

Avchd codec is much cleaner at high iso. I don't use MOV in low light for this reason

@flablo Yes that looks firmiliar. My tests have produced the same results. And for the most part with AVCHD and MOV. I'm trying to aquire a Nikon 20mm f2 AI manual lens. Or perhaps the Lumix 20mm Pancake. Either way I've got to get more light into the GH3 for low light if I'm going to use it for client jobs.

If any one has the 20mm lens and a GH3 I'd love to see screen grabs from your tests. In theory more light should help the issue.

This really makes me appreciate the work of VK, Drifwood and others even more. Makes me look at that "little plastic GH2" with new respect. Don't get me wrong I really live my GH3 - but the irony of this lesson has been enlightening to my brain as we'll as my bank account!

@maddog15 I figured that may be your lens. First, my GH2's are better in low light than my 3. That said, the 14-140 is a nice lens in good light and sucks in low light. Settings aren't going to help that much. Try a faster lens, that will help a good bit and hope VK hacks the GH3.

I also don't set my GH3 as flat as my GH2's but again, that isn't going to help with the noise.

Yep, the 14-140 is a dog in anything but daylight. That's your biggest problem.

Shouldn't there be concensus regarding low light performance between gh2 and gh3?

did other tests and AVCHD is incredibly better than MOV. Furthermore, shooting at 50p also eliminates the jitter/stutter problem associated with pans which has always bothered me! And bitrate is the same. And it is 50p so has "incorporated slow-mo". @Sph1nxster I should buy you some cookies

@manstok Nice test. Huge difference in my eyes also.

Granted neither is "usable" but I'm assuming you pushed it to reveal the noise for the purposes of the test? And just to confirm you did no adjustments to the images per Photoshop, Lightroom etc? These are straight screen grabs from your movie files?

Thanks for posting the images

@Sph1nxster Yes I've really become partial to All Intra recording. Edits wonderfully also. After hacking my GH2 with Moon T5 from Driftwood I doubt I'll go back to Long GOP's like DREWnet V8. In fact I'll probably leave Moon as my GH2's resident firmware. Absolutely beautiful footage and grades well.

That's to other thing about my GH3's footage. (very limited experience at this point as I just bought the camera) But grading in post can quickly reveal artifacts, crushed blacks and noise. Very tricky. But like I said I'm assuming a lot of "user error" at this point until I've lived and worked with the camera for a few weeks.

@vicharris of course with more light you have less underexposed area, but you can't presume you will never deal with an image with an underexposed area, even with faster glass. It is like saying "don't shoot dark scenes, just shoot daylight" (I'm exaggerating of course), it doesn't solve the codec issue we are discussing in this thread, in my opinion.

I'm posting 2 shots, mov and AVCHD. The latter seems objectively better (less harsh) to me

@maddog15 I used after effects and graded both exactly the same to push the noise.

@flablo the exposure and WB in your images look different, agreed the second image looks better

Here is one I just did, see names on windows to know who's who f 5.6 s 600 ISO1600, Panasonic Leica 25mm 1.4