GH4 Lots of tests with the new GH4

I totally agree mcbob also is the 1080p really that bad ? I thought it looked excellent.
 
Good morning. I updated the 1st post with a couple more clips I had.


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:
 
Last edited:
macgregor said:
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. Video in 4K from Sony Vegas.
Click to expand...

I didn't care for the 'redish' cast on the girl's arm, interior side, that seemed to be in several of the profiles. "Cinelike-D" and "Natural" seemed to have less of such.

Does the GH-4 have the ability to create custom profiles?

I have been waiting for 1) a good christmas bonus, 2) something to upgrade from my GH-1. (when the GH-2/3 came out... no christmas bonus...).

And if it is really, really good, go into hock... well, once the new computers are paid off...
 
j1clark@ucsd.edu said:
Does the GH-4 have the ability to create custom profiles?
Click to expand...

Well, once you select one of the profiles, you can adjust Contrast, Saturation, Sharpness, NR, (just like the GH3) and Hue (newly added). You can also add a custom curve for shadows and highlights and go with one of the existing ones or add your own.
 
j1clark@ucsd.edu said:
I didn't care for the 'redish' cast on the girl's arm, interior side, that seemed to be in several of the profiles. "Cinelike-D" and "Natural" seemed to have less of such.

Does the GH-4 have the ability to create custom profiles?

I have been waiting for 1) a good christmas bonus, 2) something to upgrade from my GH-1. (when the GH-2/3 came out... no christmas bonus...).

And if it is really, really good, go into hock... well, once the new computers are paid off...
Click to expand...

The reddish cast on her arm??? Has nothing to do with the cameras color profiles. That is mixed lighting. Look at the water jug on the table. What do you see reflecting off the top of it? Yep, that would be a incandescent bulb right over the top of the girl spilling warm light right down her left shoulder and arm. You can even see it's hot spot on her shoulder.
 
I'm of the "the more information and data, the better" persuasion, so I definitely appreciate the gentlemen posting their tests for us to peruse.

However, is the method of measuring DR by under- and over-exposing using exposure compensation valid? My first inclination is that it might put us in the ballpark, but not very accurate.

In fact, the comparison between the F35 and the GH4 isn't that far off (I don't know what the DR of the F35 is; guessing around 12-13?). Certainly, the highlights on the GH4 get blown out quicker, but is the difference that big?
 
One thing I just noticed is that at 1600ISO the image on the LCD looks very noisy, however, in the computer the footage looks fine. Sure it's got some noise but totally acceptable for 1600 from a tiny camera.
 
A quick Google search seems to indicate the F35 has 14 stops of DR. So, is the difference between the F35 and the GH4 truly 6 stops? (Honest question!)
 
hoju said:
I'm of the "the more information and data, the better" persuasion, so I definitely appreciate the gentlemen posting their tests for us to peruse.

However, is the method of measuring DR by under- and over-exposing using exposure compensation valid? My first inclination is that it might put us in the ballpark, but not very accurate.

In fact, the comparison between the F35 and the GH4 isn't that far off (I don't know what the DR of the F35 is; guessing around 12-13?). Certainly, the highlights on the GH4 get blown out quicker, but is the difference that big?
Click to expand...

Hi Hoju, not sure what you mean by exposure compensation. The DR test was performed with constant lighting, cine lenses with T stops, and a lightmeter to set up the correct exposure from which to start.
 
Good morning! Thank you for the nice comments. It was a ton of fun working with Matt Uhry and Macgregor and I learned a lot from them as usual. Thanks to Mac for making his space available too for what turned out to be a full-day (and evening) test.

First off: in some tests, I mis-labeled some shots as H+5 S+5 when they should have been H-5 S+5.

Second: Don't worry, we are trying hard to find the best way to get max DR with good color out of the camera. More tests were done, exploring the issues we found and looking for workarounds, shooting with different WB, with 1/2 CTB, etc. I also brought some contrast filters but didn't get to that because I needed to dig in and debug Cine-D.

I find DR subjective. I'm sure other people will find other numbers (including 11), depending on tolerance for noise, color content, etc. But you can compare DR between two cameras, which makes the F35 side-by-side useful. Anyway, Mac wanted to see how this camera compares to the F35 which he usually shoots on... and the answer is "awesome resolution but you need to be REALLY CAREFUL vs the F35 due to more limited DR."

I suspect that the red channel clips quicker than the others - and this causes the desaturated or green artifacts on very bright saturated objects that we're seeing. Unfortunately Cine-D accentuates these a little! The other modes have their own issues, though. And they also still have that green issue to a lesser degree. Pictures will follow.

Actually, the color processing seems to be really hurting what this camera can do. Hopefully Panasonic will update the firmware. Or (please please please) just send the data straight from the sensor out through the HDMI with no processing at all. They don't need to de-Bayer it - they could just send the raw values from each pixel. 4K via HDMI at 4:2:2 10bit is actually a higher data rate than the raw output from the sensor would be - you're sending out a 10bit luma and a 10bit chroma value per pixel, and I don't know of any raw sensor that needs more than 20bits to describe each sensor pixel's value (14 is usually plenty). I guess this is why I am keen on that Apertus stuff...

Anyway... if you are testing DR with a grey chart, you will get one number. But if you test DR with colors, you might get a different number. Definitely I would like to find my Stouffer wedge and shoot it not just with greyscale but also with different lighting gels on top of it. But that can come later (and doesn't need an F35 side-by-side).

I must admit that I am not yet 100% sold on Cine-D. I am also playing with Portrait with contrast, sharpness and saturation way down, i.Dynamic on Low (I am expecting to discover some pitfall somehow but so far have not), and for highlights / shadows... well, I like shadows plus 5... but I don't really like those highlights so am playing with keeping the highlights section flat or even just pushing them up +5 (this looks like it will be awful but I don't think the graph in the menu reflects what the cam is actually doing) so that you can spend that 8-bit tonal budget on what the camera does well. This might be a terrible mistake so please don't follow this blindly.

If this were recording raw or Log 10-bit we would just expose more to the left (and plan to blow out the dodgy highlights with an aggressive S-curve) but since this is an 8-bit file we're making I am looking to find a way to not waste a lot of my bit budget on highlights whose colors I don't like.

I definitely think that it is worth it to consider the scene you are shooting and not pick a mode that is too flat for the scene because then you might accentuate 8-bit issues.

Bruce Allen
www.boacinema.com
 
macgregor said:
Hi Hoju, not sure what you mean by exposure compensation. The DR test was performed with constant lighting, cine lenses with T stops, and a lightmeter to set up the correct exposure from which to start.
Click to expand...

Hi Macgregor,

Thanks for the reply and clarification. It sets my incorrect assumption straight!

:)

I get my GH4 on Tuesday, and looking forward to it! Poor planning on my part, though, means I won't get the actual memory cards needed to make the damn thing work until Friday ;)
 
hoju said:
Hi Macgregor,

Thanks for the reply and clarification. It sets my incorrect assumption straight!

:)

I get my GH4 on Tuesday, and looking forward to it! Poor planning on my part, though, means I won't get the actual memory cards needed to make the damn thing work until Friday ;)
Click to expand...


We were just using class10 30MB/sec SD cards. 30MB/sec sould be plenty of speed for 4K 100Mbits if I'm not doing the math wrong.
 
Glad Liam got the joke :)

OK after further looking at what we shot yesterday... if you want max dynamic range (on subjects other than pure grey):

Photo Modes: Portrait and Cine-D both work. Cine-D is by no means a magic bullet (you're not going to get another stop of highlights or whatever).

I don't have much luck trying to recover highlights with Highlights -5 and Cine-D. You just get a little bit more desaturated grey or green in the highlights. The color science mangles the highlights before Cine-D can help much. Fun. Raw (or LogC-like) mode please, Panasonic!

If you want to hold highlights, set the ISO lower. ISO 200 and expose more to the left, maybe. Bring up the shadows as much as possible. Set your viewfinder / LCD to unnaturally bright :)

i.Dynamic is definitely worth a look. It brings up the shadows in a different way to Shadows +5. I am currently using a combination of both when I shoot.

I'm going to romp around and shoot a little more with this in mind.

Bruce Allen
www.boacinema.com
 
Bruceallen,
So, changes to the S curve not making much difference? What about the Picture profile adjustments (i.e. Contrast, Saturation), etc?
REALLY appreciate your testing and posting about your findings. Thank you!
 
SuperSet said:
Bruceallen,
So, changes to the S curve not making much difference? What about the Picture profile adjustments (i.e. Contrast, Saturation), etc?
REALLY appreciate your testing and posting about your findings. Thank you!
Click to expand...

The S-curve works in the shadows but the highlights kinda suck (they seem to be mangled by the color processing earlier in the chain) so there's little point in trying to get more stops there.

Contrast, Saturation, etc have a big difference because a lot of the color profiles crush the blacks by default (WHY?). Also, the default sharpening is HORRIBLE and makes everything look like video. You'll definitely want to turn it down. Why do they make the camera look so bad out of the box?!

I will stop talking and start posting images soon, promise.

Bruce Allen
www.boacinema.com
 
Bruce or anyone involved with the tests, have you experimented with adjusting the Master Pedestal setting? By default I assume it is set to 0. I've heard that by lifting it to+1 or +2, you can avoid crushing the blacks so easily.
 
macgregor said:
WB set at 5600. Lighting source: kinoflo diva lite 400.
Click to expand...

First off thanks for sharing your extensive tests!

I'm out-of-country & have a dodgy internet connection, so haven't looked thru all the footage, but am shooting with the GH4 & Diva-Lites.

Like previous Panasonic's cams CineLike-D & Divas (with daylight bulbs) work best with a custom white balance. The skin tones we shot without a custom white balance looked magenta & plain icky.

Best,
Bern
 
You must log in or register to reply here.
Back
Top