GH4 Different GH4 settings, beware of Highlight and Shadow

soulofrev

Member
Hey I have been using the GH4 for a while and I made a video on some of the different settings. Watch out for the highlight and shadow options.
http://youtu.be/YE7Gd3r4Lek
Thanks Chris :)


 
Why beware? Share your experience and details!

I might be wrong but I think he is meaning 'pay attention to the effects on shadows & highlights as you watch the different picture profiles'

I'm not trying to speak for someone but that is how I read into it.
 
If you watch the video, he verbally shares his opinions about different settings. In short, he finds monkeying with highlights/shadows to make the image look too processed and creates unwanted artifacts.
 
Hey I have been using the GH4 for a while and I made a video on some of the different settings. Watch out for the highlight and shadow options.
http://youtu.be/YE7Gd3r4Lek
Thanks Chris :)

Thank you so much Chris, for sharing that extremely helpful video!! May I ask what lens you used for the portrait shots of the man?
 
I totally agree about getting a 'plastic' look with -/+5 so I sort of settled on a -/+3 but after a few days I've now dropped it down to -/+1
iDynamic LOW - but I'm thinking of turning that off as well.
Keeping Master Ped between 0-5.

Thanks
 
Yes, I am still orbiting around the perfect settings. I'm doing some experiments using Natural instead of Cine-D because I've found Natural to handle highlight roll-off better.
I also shot a little test tonight toggling between a Master Ped of 0 and +15. There's a case to be made for lifting the pedestal to that level, as the worst aspects of h.264 compression tend to live down in the shadows. My test is to see if lifting the pedestal to +15 will reduce macro-blocking in the shadows without creating any other problems. If I don't see any evidence of improvement, I'll return to keeping the Master Pedestal around 0 to +1 depending on the lens I'm using, the presence of flare from backlight, etc.
 
Raising the Master Pedestal to +15 will only simulate the "flat" low-contrast look, it won't actually give the shadows any more tonal detail than leaving Master Pedestal at 0. If you watch closely at 0:52 in the video (where Master Pedestal is raised to +15), you'll notice that not only the shadows, but the mids and highlights are brightened as well. The brightness of the entire scene is raised, making shadow details more visible, but not in a way that counts. The reason lies in the math.

When you raise the pedestal, you are adding a constant offset factor to the entire range of luminance values. The near-black values that would have been recorded at levels 0-15 will now be recorded at levels 15-30, and everything else will be raised by 15 8-bit levels as well. This will make the video look brighter and the shadows look grayer, but so long as you don't clip the highlights, the encoded results will be no different than before.

To get the grading benefits of a genuinely flat encoding curve, it's not enough to raise the pedestal. You need to lower the overall contrast by boosting the shadows with a multiplicative factor (i.e. more exposure) while reducing the exposure of the highlights. That's the reason for using Cine-D and adjusting the shadow and highlight settings.

It's likewise a rare situation where you'd want to raise the pedestal in post, since that will typically produce a visible gap between the darkest gray value and solid black. It's usually better to lift shadow details with the Shadow Gamma control, preserving a smooth gradient from black to dark gray.
 
Last edited:
Raising the Master Pedestal to +15 will only simulate the "flat" low-contrast look, it won't actually give the shadows any more tonal detail than leaving Master Pedestal at 0.

Yes, I understand that, and I understand the math.
The reason that I'm doing the experiment (I can't vouch for the OP), is to see if lifting the blacks would combat problems with macroblocking and other h.264 artifacts that manifest in the shadows. My hypothesis (and it's just that...the footage I shot tonight as a test is inconclusive) is that lifting the image into a range where the GH4's encoding does less damage might help. In post, I plan to return the black level to 0 and thereby suppress noise and artifacts.

But I am aware that this doesn't give me anymore dynamic range...in fact, it may even lower it. It isn't Log, etc.

Here is my temporary conclusion based on monkeying with the footage in Premiere tonight: it's not worth bothering with. After color correction, I can't tell any difference between the footage I shot at Master Ped 0 vs. Master Ped +15. So if I can't tell the difference, and I know that Master Ped 0 is going to give me more range in the image, I think I'll stick with that.

If this had been a GH1, however, Master Ped +15 might be valuable. I loved my Gh1, but it had a horrible case of noise and macroblocking at about 20-30 IRE. Lifting the Master Ped and thus the whole exposure a bit higher might have helped back in those days. But I feel the GH4 does not have this problem. Maybe at really high ISOs it may be of some help. I may do a test and compare some low light footage shot at 3200 ISO and see if there's a difference in the noise level in the shadows.
 
Shooting this weekend and playing around with Curves, picture profiles and Highlight / Shadows.

Definitely think that excessive use of Highlight reduction (in order to avoid blown highlights) is going to mess with the looks of features/ skin tones etc. Its great for wide shots where people aren't a problem, to keep a bit of sky from blowing out, but on faces it creates a painterly look for faces.

For shooting people - I've been leaving Highlight reduction on -1 and shadows on 0. Cine Like D, seems to work best, 16-235 and i'm not touching master pedestal. Contrast is -3, Sharp - 3, NR-5, 0, 0. The most important thing for colour seems to be nailing the white balance. I generally find the AUTO WB to be very magenta orange, so dialling it in via the kelvin adjustments is crucial to me. I'm just dialling it in by eye matching the LVF with the real world as close as possible. I've got I Resolution on 'Low' and IDynamic off.

Next thing i've noticed is that exposure via Zebras is a bit off (even set to 95% they're a bit 'conservative' - i.e. not overexposed at all), so i'm doing all my exposing via the histogram and it seems by far the most accurate.

Get the exposure 'in the zone', white balance right via the kelvins and using the above picture profile settings and its a very grade-able and very lovely looking image. Even when using Vari-ND filters (I'm using a Tiffin).
 
Last edited:
Back
Top