GH4 James Miller's Camera Settings for Best Dynamic Range

Benjobe

Well-known member
James Miller just posted a short video describing his camera settings for best dynamic range. He's contributing to the evolving GH4 camera review on Philip Bloom's blog. His settings use Cine-V and a Master Pedestal of +15 with pretty impressive results.

http://vimeo.com/95589923

Here's the link to Philip Bloom's evolving GH4 camera review: http://philipbloom.net/2014/05/12/gh4/

I'm intrigued by his use of the Master Pedestal and Cine-V. So far I haven't seen or heard of anyone else using those settings. Most people here seem to prefer Cine-D. He also uses i.Resolution which again most people here seem to avoid. Anyone else interested in exploring these new settings and comparing them to the Cine-D settings?
 
Last edited:
The thing with non raw or LOG capable cameras are that you have to fiddle with settings to get the most out of the camera. Why not just use a camera that does it right from the start.
 
The thing with non raw or LOG capable cameras are that you have to fiddle with settings to get the most out of the camera. Why not just use a camera that does it right from the start.

Abelcine's 'alexa classic starter kit'

AB-ALEXA-B862
Options View Rental Package
MSRP $80,110.00

SAMMY'S
GH-4 PRICE: $1,697.99

ADORAMA
GH-3 PRICE: $997.99

Now what was your question?
 
The thing with non raw or LOG capable cameras are that you have to fiddle with settings to get the most out of the camera. Why not just use a camera that does it right from the start.
Even the best of the best need some fiddling. If a camera came out with no user adjustable settings it wouldn't sell.
 
Tooled around with his settings a bit and didn't really like how contrasty it turned out. Going back to CineD. Also I'm not sure why folks are choosing Saturation -5 when it seems easier to desaturate in post vs adding it back.
 
I have to agree with the growing suspicion here that he's going down the wrong track, starting from Cine-V which was designed as a compromise for people who don't fully grade.
 
I started doing my own testing, using a BlackMagic SmartView Duo. My goal is to try to match my GH4 with a RED EPIC Dragon. I found that I can get the CineV waveforms and vectorscope to look more like the DRAGON (DragonColor/RG4) than CineD will do. I do have to dial back contrast (-4) and saturation (-5) for sure. I haven't yet delved into noise reduction and sharpening settings, but will likely follow the wisdom shared by the OP. Ditto i.Resolution and i.Dynamic.

I do realize that RedGamma 4 is a LUT, not a LOG format, so perhaps that's why CineV can begin to match it, whereas CineD does not. I will look at trying to match RedLogFilm with CineD later this afternoon.
 
Everyone's entitled to their own approach with the GH4, but I have to say I disagree with most of Miller's settings, based on my use with the camera so far.

Cinelike-V is too contrasty, even at-5, if the point is to have room for grading. It's gamma curve leaves less room for highlight retention than Cinelike-D.

Setting the Master Pedestal to +15 is just weird to me. It lifts the blacks by about 10 IRE, which doesn't increase shadow detail, it just lifts the darkest pixels in the image up in brightness, making everything milky. This is 10 IRE of potential tonal range now gone.
Compound this with the use of 16-235 (which I understand he's using for ease of workflow), and you now have about 20 percent less luminance range to grade. Given the 8-bit codec, this is not a good idea, even if you are color correcting inside a 32-bit float environment.

iDynamic "increases" dynamic range by lifting the shadows in a way that creates noise. My experiments with it have shown that lifting the shadows later in post looks better than using iDynamic. Plus, iDynamic changes as you're recording if the brightness values in the scene change. This is unacceptable to my style of shooting.

And iResolution adds an electronic, fake sharpness to parts of the image. Again, not something that I'm a fan of.

Of course, what matters is the end result. I'll be interested to see what his final, graded images look like.
 
Also I'm not sure why folks are choosing Saturation -5 when it seems easier to desaturate in post vs adding it back.

Plus, you may want to saturate an 8bit compressed codec as far as you can without clipping, since you'll likely maximize the amount of tones in the image, which will reduce banding.
 
Tooled around with his settings a bit and didn't really like how contrasty it turned out. Going back to CineD. Also I'm not sure why folks are choosing Saturation -5 when it seems easier to desaturate in post vs adding it back.
My feelings on saturation too. Sure, I probably don't want as much contrast in the image in final, but as long as you're not overdoing saturation to the point of maxing out, you can dial it back down and you have more dynamic range per color. Dialing up saturation is what can give you that jagged digital look or banding. If you were shooting in a 4:2:2 codec it wouldn't matter as much as you can push the saturation a lot more.

*edit* Just saw that if I would have read further down, Joshua made the same point. :P
 
Last edited:
I looked into this briefly and here is what I sorta find.

This is the J. Miller profile set up per his video:
J Miller.jpg

Here is CinelikeD set up as follows:
COntrast -2
sharpness -2
NR -5
Sat 0
Hue 0
Curve zeroed
Master Ped +10 (this is where you want to adjust to taste and subject).

CineD.jpg

Between the two you first see the blue shift from the skylight coming through the window on the CineD where the JM profile is neutral due to the -5 Sat. Other than that the hues of the colors are about the same. Now sure if this will show on these compressed uploads but here is what I see. If you look inside the colored cups what I'm seeing is that the shadow gradation is far smoother in the CineD versions where the JM profile shows a much less smooth grad. And this is in all three cups. Personally I would prefer the CineD profile for those smoother grads because they will hold better to color correction.

I haven't really tested the effects of the I-dynamic setting too much. Reason being is I'm not too sure of how the camera processes I-dynamic by means of questioning whether the camera reads a scene and adjusts the image or if its a set constant setting in camera. It is called "Intelligent" dynamic which would lead me to believe the camera is making adjustments as to the scene. I guess we need to figure this out some how but as of now I avoid it. I don't want the camera making adjustments for me.

I will stick with my settings on the CineD for now but will set a custom dial setting with this JM and shoot a few same scene comparisons. But my first impressions were that the color gradation is smoother using the CineD settings.

Keep in mind that I dynamic could maybe go the opposite way we want too. Is the setting seeing a flat scene and adding possible micro contrast? This is why I would rather set hard settings in the camera. With contrast setting, master per, and the curves adjustment you should be able to dial in the luma you want. Then use Sat for your chroma values.

I too don't automatically set the Saturation to -5. If I'm filming something that I want some vivid poppy color why would I want -5? If I know I'm going muted maybe I would set -5. But man, the NLE can desaturate the color to taste. Keep in mind, every time you make a color or contrast adjustment to ANY file other than RAW, you are degrading the image. ANY adjustment. So if looking for color that pops, -5 sat might not be the best way to go. I think this method jut comes from the 5Dmk2 days where a handful of guys posted about using minimal saturation and it just stuck with people over time to automatically set cameras that way.

Think of those settings as film stock. Im my film days I would have never taken a flat low saturation negative film to photograph an image destined for high valued color and vice-versa. Pre-visualize what you are doing and set your camera as close as you think it should be for the shot.
 
Michael, I've very interested in your tests and what you come up with here. Granted, the GH4 probably doesn't have the DR the Epic has, but getting things dialed in close on the GH4 would be very beneficial when mixing cameras.
Please keep up posted with your tests and results. I lack the experience and tools to do the test you are doing, but would love to know the real world effect of various settings like sharp and Noise Reduction etc. Thanks!
 
Everyone's entitled to their own approach with the GH4, but I have to say I disagree with most of Miller's settings, based on my use with the camera so far.

Cinelike-V is too contrasty, even at-5, if the point is to have room for grading. It's gamma curve leaves less room for highlight retention than Cinelike-D.

Setting the Master Pedestal to +15 is just weird to me. It lifts the blacks by about 10 IRE, which doesn't increase shadow detail, it just lifts the darkest pixels in the image up in brightness, making everything milky. This is 10 IRE of potential tonal range now gone.
Compound this with the use of 16-235 (which I understand he's using for ease of workflow), and you now have about 20 percent less luminance range to grade. Given the 8-bit codec, this is not a good idea, even if you are color correcting inside a 32-bit float environment.

iDynamic "increases" dynamic range by lifting the shadows in a way that creates noise. My experiments with it have shown that lifting the shadows later in post looks better than using iDynamic. Plus, iDynamic changes as you're recording if the brightness values in the scene change. This is unacceptable to my style of shooting.

And iResolution adds an electronic, fake sharpness to parts of the image. Again, not something that I'm a fan of.

Of course, what matters is the end result. I'll be interested to see what his final, graded images look like.
I also find conspicuously absent from this whole issue an attempt at consensus with respect to particular workflows that will inevitably include (for most filmmakers) plug-ins like FilmConvert, Lumetri Looks, Red Giant Looks, and maybe others. I'm worried about the FilmConvert folks in particular (since they're the best) being bold enough to "demand" a specific optimal starting point against which they will have built their referencing camera profile. It will be a disappointment if they just work off of Cine-D at defaults, for example. But the point is, we all eventually need to sort of get onto the same page. Organically (with no help from Sony, I should add), Frank Glencairn ended up creating the default LOG-wannabe profile for the NEX-FS100 that is still my go-to camera for overall performance, after adding Blackmagic and GH4 into the mix. But to take that example, the FilmConvert team never played game and simply built its camera profile upon Sony's contrasty, non-flat default.
 
I'm worried about the FilmConvert folks in particular (since they're the best) being bold enough to "demand" a specific optimal starting point against which they will have built their referencing camera profile....the FilmConvert team never played game and simply built its camera profile upon Sony's contrasty, non-flat default.
I'm a bit confused by this statement. From my understanding, Filmconvert profiled the GH3 using every picture style included in that camera. As near as I can tell, their contrast curve for the GH3 is created with the assumption that a picture profile is on -5.

That being said, you are correct in that we are all waiting for Filmconvert to have an actual GH4 profile. All this guesswork of trying other cameras as the source camera leads to weird results. And no matter what film stock I choose in Filmconvert, I can't find one that doesn't turn green foliage brown. Anybody had any luck preserving greens using Filmconvert?
 
I also find conspicuously absent from this whole issue an attempt at consensus with respect to particular workflows that will inevitably include (for most filmmakers) plug-ins like FilmConvert, Lumetri Looks, Red Giant Looks, and maybe others. I'm worried about the FilmConvert folks in particular (since they're the best) being bold enough to "demand" a specific optimal starting point against which they will have built their referencing camera profile. It will be a disappointment if they just work off of Cine-D at defaults, for example. But the point is, we all eventually need to sort of get onto the same page. Organically (with no help from Sony, I should add), Frank Glencairn ended up creating the default LOG-wannabe profile for the NEX-FS100 that is still my go-to camera for overall performance, after adding Blackmagic and GH4 into the mix. But to take that example, the FilmConvert team never played game and simply built its camera profile upon Sony's contrasty, non-flat default.
I am just not tony enough for this much name dropping. But yeah, Cine-V is for when you want your footage to look ENG.
 
I'm a bit confused by this statement. From my understanding, Filmconvert profiled the GH3 using every picture style included in that camera. As near as I can tell, their contrast curve for the GH3 is created with the assumption that a picture profile is on -5.

That being said, you are correct in that we are all waiting for Filmconvert to have an actual GH4 profile. All this guesswork of trying other cameras as the source camera leads to weird results. And no matter what film stock I choose in Filmconvert, I can't find one that doesn't turn green foliage brown. Anybody had any luck preserving greens using Filmconvert?

FC mentioned that when they released the GH3 profiles, they were profiled for all 0's. A few people have posted to their FB page that they should create GH4 profiles using -5 so they said they'd pass it on to the developers. No further word on it yet.
 
Don't worry, I am sure Barry Green is working on all these litlle details for "THE BOOK" on the the GH4. All these settings come from the good old days of the DVX, HVX, HPX, AF100 era and he knows them well. Waiting for your input Barry Green, Panasonic master.
 
Back
Top