GH4 Lots of tests with the new GH4

macgregor

Veteran
Hi everyone.
Bruce Allen came today with two brand new GH4s so we had no other option but to test these new little camera.

I'll be quick and just post the results with a few comments. All clips imported/exported in Davinci Resolve 10 Lite. Camera set to full range (0-255).




Resolution test on all the different modes resampled to 1080p. 4K is perfect. 1080p is bad (as most DSLRs), and 1080p 96fps is really bad.
Shot on Sony Cinealta 4K 35mm PL mount lens.
 
Last edited:


GH4 custom photo style settings explored. Definitely bring down sharpness all the way to -5. Even at 0 the halos are way too strong. No matter if you're shooting for a 1080p or 4K finish, -5 looks the best.
 

Same lens (Nikon 24mm f1.4), just different nikon to MFT adapters. With and without speedbooster.
I think the footage with the speedbooster looks really good.
Notice the moire on the building on the left side, with both adapters. This is visible in 4K.
 
Last edited:

Super quick dynamic range test.
Well... this is the worst part on this camera. It's not terrible but definitely not awesome.
I can only see there +3, N, -4 stops, for a total of 8 stops of DR at 400ISO.
 

Just so you can compare the DR against a high end digital camera here's the same set up shot vs the Sony F35. You're not going to like the results...
 
Last edited:

Real life effects of different picture profiles in a high contrast scene with skin tones. DR seems to be about the same in most of them. So any settings/adjustments/tweaks in the profiles only seem to be moving the curve up/down, not really getting any more info. Video exported at 4K from Sony Vegas.




And knowing the limitations in DR, I decided to go a bit flat and underexpose by one stop to see what I could get in post:



And finally some real life footage with lots of detail and tricky highlights. shot with nikon lenses (14-24 and 70-200) and a speedbooster on Cinelike-D gamma. No duck shots.
 
Last edited:
Conclusion:
- fantastic resolution and image quality (specially for a 1080p finish)
- very limited dynamic range for what the the competition is offering
- for the price this thing sells and being able to record 4K internally, this little thing is quite impressive
 
Very nice tests, thanks.

The green in the skin tone test...would you put that down to the camera or lighting?
 
Hi everyone.
Bruce Allen came today with two brand new GH4s so we had no other option but to test these new little camera.

I'll be quick and just post the results with a few comments:




Resolution test on all the different modes resampled to 1080p. 4K is perfect. 1080p is bad (as most DSLRs), and 1080p 96fps is really bad.

While the 17th Century Still life may be good for some people, I'd prefer to see basically a step wedge or even a grey card and waveform displays of such.

I'd also like to know the amount of light used in footcandles/lux.

And an external meter reading of the scene and what that meter indicates for the exposure for the set ISO value.

I've been testing my cameras to have the 18% grey card level at the 50% IRE, and then from there open up N stops or close down M stops to find the effective Dynamic Range.

At this point I compensate for my GH-1 ISO setting about 1 1/3 stop 'faster', that is if I set the GH-1 to 100 ISO, I meter at ISO 250. For my Nikon D600, I set the camera to 200 and meter to 250.

Otherwise, just putting a 'step wedge' in the scene, perhaps with a color chart in the back ground, and a few 'fruits' in the front, really doesn't give a setup that people can realistically setup and test them selves, either if they get a new camera or want to compare with their current camera.

I know that grey cards are boring... but that's somewhat the name of the game with technical things.

For example:

13926600054_b6ca2d2b9e_o.jpg


13903013251_427182b299_o.jpg


Boring... but the first image, gives the 'composite' image with a graph of the small rectangle on the grey card. The graph shows the approximate level of the area, for each f-stop setting.

The wave forms show where these stand on the IRE scale, and give a pretty good idea about what to expect on 'dynamic range'.

Some people do say that the grey care should be at some other value on the IRE scale... which may be 'fine' in some sense, but it would be nice if they were clear on the point, and then others could use that as their point of reference, or chose another that yields for their purposes a 'better image'.
 
Last edited:
j1clark, thanks for the lecture but my 1-bedroom apartment flooded last week in a tragic water heater leak last week, and my Stouffer wedge chart and grey card went missing when I carried my stuff frantically to the storage room. So that was not an option.

In the meantime, if you can't get any value from our tests... well, I dunno. Sorry dude. You are entitled to a refund on these tests posted on the Internet for free. I personally found them useful. Especially since for us it was about things like, oh, trying to figure out the camera's look, taking a preliminary look at color profiles, etc. Personally I find grey cards not vastly useful in figuring out a camera's colorimetry.

The most important thing we learned is to turn sharpening way down, which the "17th century still life" was very useful for. It was also very useful in helping us determine the resolution loss when we went to 96fps. Actually, screw it... it was just plain useful.

Anyway, sorry you didn't like our tests. We did them for ourselves, to be useful for ourselves. Then we realized that others might find them interesting, so we put them up. We did it for free. Your post encourages us to keep them private in future.

Bruce Allen
www.boacinema.com
 
Yep, it's odd to see someone take this tone when someone makes such an effort.

Bruce, care to share your impressions? No other cam does what this cam does for the price, but I think I've been spoiled by the cams that do colour well (and dr for that matter).
 
So the 1080p is bad unless resampled from 4K ? Also Is the GH4 1080p as good as the GH3 1080p ?
 
Back
Top