What I understood about compression algorithm based on DCT, is opposite: where is an image with more contrast and color vibrance, more easy to compress. The CODEC destroys any detail or data that is no visually "important" to our eyes, for example in the dark areas, where color and the zone is easy resumed by less tones/colors, less detail.

Aron - same Nikon vintage glass I've been using for years

@SuperSet yup

@heradicattor you'll find out that I buck conventional wisdom at every turn with pretty astounding results. I never base any conclusion I reach off of a pre-set theory. I tend to be more scientific and base it off observable and repeatable results. When doing camera and patch tests 2 years ago for a short I was getting ready to shoot with the hacked GH2, I noticed a little something peculiar http://www.personal-view.com/talks/discussion/comment/53940#Comment_53940

Whenever fog or smoke were introduced to an image thereby bringing up the shadow detail and lowering contrast in a very similar way to what the camera now does internally, the data rate dropped... significantly and consistently. On some VBR IPB patches this was a terrible thing, but on the more robust VBR ALL-I patches the footage was astounding and yet still compressed at a lower data rate. Now this tells me either Digital Compression Theory doesn't work the way you think it does, or the GH2 and GH4's internal computers have the algorithm all backwards... I found over time that the flatter the image : the lower the data rate, and the cleaner the image....every time.

So either I and the users I instruct to adopt my methods happened to buy the only GH2 and GH4 cameras on the market that actually defy DCT, or your approach is flawed. I teach all my students - Shoot, analyze, revise. Base you methods on the observable results of practical research.

@shian, fog and smoke don't increase shadow detail in any way. It might refract light into the shadows bringing up the levels in that area, but total scene detail to be captured is radically lower. Similar effect to using super shallow depth of field.

Shian Yours uneducated expletive language is unnecessary and unacceptable. I think your post should be deleted.

Matt, I don't think of the curves adjustment as increasing or decreasing sensitivity but more as a response curve. I'm probably mistaken in thinking that. I have no idea what the camera is actually doing to the image...I'll admit that right now. I don't have the time to run it down. BUT - This I know. I used to have to use Lo-Con filters and fog to get lower contrast. Now, with whatever the camera has going internally. I can do the same thing with Cine D, the curves, and lowering the contrast to get a flat image. I'm aiming for what to my eyes is as similar to 35mm film scans and RED raw footage as I can get. These settings get me in the ball park (to the naked eye) Yes I know it's not a raw cam. Everything is an illusion in filmmaking. It doesn't matter what is, it only matters what it looks like.

I stopped at (+2-2) cuz in my tests, if I made the image any flatter it starts to fall apart when I try to stretch it back out along the luminance spectrum. I don't know why I'm not having the same noise issues everyone else has. OR maybe it's just taste. I like grain. The 4k images of the cam have a very nice, fine film-esque grain to them. As a film guy I'm used to that... I like it. I'd be adding grain if it weren't there. That hacked GH2 had very nice grain, and this is even better. So maybe that's the sticking point.

@aaronchicago Master Pedestal does not increase the shadows like everyone seems to think it does (the shadow curve is actually what reveals more of what's already hiding down there). Instead, Master Pedestal takes the entire image and bumps it up (or down) on the IRE scale. Pushing up +15 is simply raising your entire image and pushing the top end of your highlights closer to their ceiling (clipping point). Master Pedestal was originally used to try and match one or more cameras together, not to reveal more information in shadows. By the same logic that people are using, you could lower the Master Pedestal -15 and push your highlights away from their ceiling even further, but then your blacks would be mush. It's best to leave pedestal at 0 unless you are matching your GH4 to another cam. If you want more information in the shadows (or highlights) lower your contrast or use the curves feature (but the more you open up those shadows the more noise is going to shine through). However, @shian is right, it's a fairly filmic grain (especially when 4K footage is downscaled). From my experience, however, -5 contrast with Cine-D is as far as I like to push it before the image becomes too noisy. I am really intrigued by @shian being able to get away with even more on the curves - probably due to his perfect 20%-80% metering.

good luck, we're all counting on you.

@shian stop being so childish in a conversation with different opinions.

@Lpowell thanks for the clarification. I will file that away for future testing. It explains a lot about my findings.

To those with their dander up: Yes, I'm an asshole who loves provoking those I find to be idiots... it's fun, it's like a sport to me.

But there are those who wish to emulate my results. The things I share are the methods and findings I use to get those results. Some will notice that I haven't been posting here much for nearly a year... I have my reasons. The only reason I post here anymore is to help those who seek to learn my methods, settings, etc... I mastered the GH2 and so did those who listened to me, The GH4 is a fun new toy... it's gonna be a fun new adventure finding it's sweet spots... oh, and cuz there's a fresh new crop of morons to throw bananas at.... So... (shrug) 2 reasons.

If you've already got it all figured out, you can either ignore me... or poke the lion with a stick. Just know that I like being poked...

Just completed an experiment with an X-Rite color checker and DaVinci Resolve 11. Shooting the same outdoor scene (late afternoon) with a chart and female model in shadow, I switched camera scene profiles with NO exposure adjustment & also with exposure adjustment to ensure a white histogram on the camera. There was full sun on the fence in the background to set the exposure; zebras at 95%, recording bits 16-235, WB set to cloudy.

Using the Resolve Color Match feature on all shots revealed three optimum camera scene profiles; Standard, Natural & Cine-D. The others all lifted the skin tones so they appeared too bright. Using curves pulled the skin back into an acceptable look, but the three profiles above still looked best. Using the shots with camera exposure adjustment also revealed the same three camera profiles as the best looking on skin tones.

Next test will be using zebras to set skin exposure.

Sounds like you have using the Ninja Star already figured out.

Just be sure you get a "no quivel" return guarentee incase it does not work out.

And, please report your results here.

I have not used the ninja star but have used a Nanoflash which is equally high bit rate but 8 bit. The improvement is quite dramatic to me as far as depth is concerned. The sharpness of the GH4 can be too much already and it is not helped by some noticeable aliasing using the downconvert from 4k to 1080p when recording external. And it is there on the horizontal and vertical - the scaling algo seems lacking somewhere - but don't get me wrong - it does grade better and it is really sharp - if you get some nice vintage glass on it looks great. So yes it is a step up from the internal codec but depends on what you are comparing. 4k internal ( 1080p timeline ) vs 1080p external - sharpness I would give to the 4k but colour to the 1080p external. Of course the Shogun will fix all this.......but a ninja will give you nice results.

@theconformist I bought the Ninja Star for the exact same reason.

I can confirm that the GH4 does in fact do 4K 422 10Bit and downscales internally to 1080p. I shot some test to compare side by side 4K modes and 1080p modes, and I wouldn't say there is TOO much of a difference on face value. Unless you really need the difference than it isn't a big deal, I would say Sharpness is about the same (Might give the edge to the Ninja Star) but the color and just the life in the image seems to be alot better with the Ninja Star. Not to mention the better flexibility in post. I would be comfortable shooting mostly with the Ninja Star, and if the occasion would arise that I need 4K to crop in, or for multiple angles, or something I would shoot 4K internally. They both look great in the end. I never saw any problems in the way of Aliasing and Moire that wasn't in the same 4K internally.

Here are some tests I did with different settings.

Pros: Already NLE friendly format and renders fast. More flexibility in post. Better looking colors. Already set to 1080p.

Cons: Double the file size. Recording externally with no monitor to reference. Cord can potentially unplug accidentally. Not 4K, no cropping in post. 10Bit can't record internally simultaneously. Need Ninja Star +CFast 1.0 cards + CFast Card Reader + Mini HDMI to Mini HMDI