GH4 Essential Settings & Essential Post for GH4 Film Look

[Thomas Smet]

Thomas, this is really helpful. Regarding bypassing any in camera NR, does this NR in post become necessary on most all shots or just more taxing low light or stressed dynamic range shots?

Do you have any concerns about the better noise reduction that the GH5 employs to be so much better in lowlight, or is that different? Should we expect to see lowlight performance similar to GH4 if we set the GH5 NR to -5?

Why does leeming luts recommend NR only at -3 (for non native M43 lenses, that is)?

And lastly, does turning down contrast and saturation get you closer to the GH4 raw image, or does it start to artificially reduce contrast and color? Wondering why if anything higher than -5 is "Fake" for sharpening and NR, why this wouldnt be the case with contrast and saturation?

Contrast and saturation are a bit different because you are recording to a limited format. Typically any time a person tried to force a flat look with the contrast and color it artificially hurt the image. Color is also subjective and also easy for the camera to do so there isn't a processing speed concern or a concern over a fake look by letting the camera handle the color.

I always use -5 NR in the camera. Typically I try not to shoot in a low light situation where I need to bump up the ISO a lot. Really if I have to start raising the ISO I prep to raise the lights instead. Of course that can't always happen. For NR in post I only do it if it really needs it. NR of -5 doesn't really need it as lower ISO levels. Noise isn't always an evil thing either.

 

[visceralpsyche]

Contrast and saturation are a bit different because you are recording to a limited format. Typically any time a person tried to force a flat look with the contrast and color it artificially hurt the image. Color is also subjective and also easy for the camera to do so there isn't a processing speed concern or a concern over a fake look by letting the camera handle the color.

I always use -5 NR in the camera. Typically I try not to shoot in a low light situation where I need to bump up the ISO a lot. Really if I have to start raising the ISO I prep to raise the lights instead. Of course that can't always happen. For NR in post I only do it if it really needs it. NR of -5 doesn't really need it as lower ISO levels. Noise isn't always an evil thing either.

Exactly All these people who recommend flattening everything never think about the fact that you then have to restore the colour from less information, than if you had left it alone to begin with! And then they complain and say the camera is noisy....

 

[filmguy123]

Paul, would you say -5 on non native lenses (like Voigtlander) would come out TOO soft then? I do notice about film, or higher end cameras, they have a ton of detail but are very smooth (don't want to say soft as that can be confusing), but wouldn't want the images to turn into mush either.

Re the other settings, let me see if I can phrase this back correctly:

Don't adjust contrast down, because it basically artificially lifts the image and introduces noise
Don't adjust the saturation down, because there is no reason to not do that in post for the same effect, and even with saturation at 0 you can still shift the color tones well in color grading. Reducing just gives less data to work with.
Don't adjust hue because that bakes in an unbalanced look and you can do that in post

Am I tracking?

RE a couple other settings, I have googled and found stuff all over but I find both of your explanations far more enlightening - and TRUSTWORTHY - as there's a lot of non sense and old posts all over the web. Could you talk to me about 4 other settings?

iResolution
iDynamic

^^ What are they, why WOULD someone want them on, and why do we NOT want them on?

Master Pedastal

^^ Is this similar to contrast? Early on there were lots of people swearing to boost Master Pedastal to 15, but now I am hearing to NOT do that. Why the change, what discovery? What's the rational and what exactly does M.Ped control anyhow? Is this just a GH4 thing to leave it alone?

Lastly, the doozy:

Luma 0-255 vs 16-235

I understand this maps pixels and luma shades to different values for broadcast, but I don't *really* understand this - as in, what potential downsides mapping to 0-255 could have, how it interacts with my software and output and any needed setting changes in workflow, why and for who this is an option at all, and what I miss out on IQ wise by leaving @ default 16-235.


Thank you both!!

Paul - if I buy Leeming LUT for GH4 today, will that support GH5 or would it be a separate purchase?

 

[Thomas Smet]

Paul, would you say -5 on non native lenses (like Voigtlander) would come out TOO soft then? I do notice about film, or higher end cameras, they have a ton of detail but are very smooth (don't want to say soft as that can be confusing), but wouldn't want the images to turn into mush either.

Re the other settings, let me see if I can phrase this back correctly:

Don't adjust contrast down, because it basically artificially lifts the image and introduces noise
Don't adjust the saturation down, because there is no reason to not do that in post for the same effect, and even with saturation at 0 you can still shift the color tones well in color grading. Reducing just gives less data to work with.
Don't adjust hue because that bakes in an unbalanced look and you can do that in post

Am I tracking?

RE a couple other settings, I have googled and found stuff all over but I find both of your explanations far more enlightening - and TRUSTWORTHY - as there's a lot of non sense and old posts all over the web. Could you talk to me about 4 other settings?

iResolution
iDynamic

^^ What are they, why WOULD someone want them on, and why do we NOT want them on?

Master Pedastal

^^ Is this similar to contrast? Early on there were lots of people swearing to boost Master Pedastal to 15, but now I am hearing to NOT do that. Why the change, what discovery? What's the rational and what exactly does M.Ped control anyhow? Is this just a GH4 thing to leave it alone?

Lastly, the doozy:

Luma 0-255 vs 16-235

I understand this maps pixels and luma shades to different values for broadcast, but I don't *really* understand this - as in, what potential downsides mapping to 0-255 could have, how it interacts with my software and output and any needed setting changes in workflow, why and for who this is an option at all, and what I miss out on IQ wise by leaving @ default 16-235.


Thank you both!!

Paul - if I buy Leeming LUT for GH4 today, will that support GH5 or would it be a separate purchase?

Luma range is an interesting subject, especially if you use an external monitor/recorder that only likes a 16-235 range which in fact is a limitation of most external monitors and no manufacturer seems to be interested in fixing it. BMD did recently add a full luma HDMI range to their Video Assist monitors so yay for them for actually addressing this issue.

For the most part you shouldn't have to worry about it. A NLE knows how to deal with 0-255 or 16-235 if the proper flags are set in the video file. The internal GH4 material does this correctly so you have nothing to worry about. It is only when a 0-255 signal is recorded externally on a device that expects a 16-235 signal that things get funky. I created a lut and a native FCPX effect to correct this issue on external recorders. I have begged Atomos to work on a solution but they ignore it.

The -5 sharpness issue on non native lens is really up to you. Paul feels it is too soft and that -3 helps a bit without looking too digital. I tend to agree with him when using my 12mm Rokinon lens. It is clearly softer than my other m43 lenses. I personally hate digital sharpening of any kind but if you feel the non native lenses are too soft it is safe to go to -3. Again it is really up to you however and it isn't a rule written in stone. These are just our own opinions.

The detail issue you are referring to is called acutance. http://www.cambridgeincolour.com/tutorials/sharpness.htm. This is one of my favorite links that describes the difference between acutance and resolution. Digital sharpening in the camera can give the illusion of higher detail but it is fake detail. It creates a lot of edge artifacts that do not look as natural because our eyes do not see the world like that. The GH4 while 4k doesn't resolve as much detail as other 4k sensors can and tends to use a bit more acutance than normal to compensate for that. This is partially why Panasonic is using more pixels on the GH5 and a more advanced form of sampling to increase the amount of high frequency detail the camera can create in the image. In theory (haven't seen any -5 sharpness samples yet from the GH5) the detail could look higher with less sharpening done in camera and bring the GH5 up to the raw natural detail level of other 4k cameras.

 

[visceralpsyche]

Paul, would you say -5 on non native lenses (like Voigtlander) would come out TOO soft then? I do notice about film, or higher end cameras, they have a ton of detail but are very smooth (don't want to say soft as that can be confusing), but wouldn't want the images to turn into mush either.

Most likely, yes.

Re the other settings, let me see if I can phrase this back correctly:

Don't adjust contrast down, because it basically artificially lifts the image and introduces noise
Don't adjust the saturation down, because there is no reason to not do that in post for the same effect, and even with saturation at 0 you can still shift the color tones well in color grading. Reducing just gives less data to work with.
Don't adjust hue because that bakes in an unbalanced look and you can do that in post

Am I tracking?

Yes, although I and others have found that occasionally, depending on the saturation of the source, -1 Saturation is safer (think fluorescent colours, flowers etc) to avoid colour channel clipping.

RE a couple other settings, I have googled and found stuff all over but I find both of your explanations far more enlightening - and TRUSTWORTHY - as there's a lot of non sense and old posts all over the web. Could you talk to me about 4 other settings?

iResolution
iDynamic

^^ What are they, why WOULD someone want them on, and why do we NOT want them on?

iResolution tries to create additional sharpening at a localised level, as far as I can make out. Like primary Sharpening, it's best left off as you can do this better in post.
iDynamic is trying to fit extra dynamic range into a shot, but does so in an uncontrolled and not altogether good way. Again, best left off.

Master Pedestal

^^ Is this similar to contrast? Early on there were lots of people swearing to boost Master Pedestal to 15, but now I am hearing to NOT do that. Why the change, what discovery? What's the rational and what exactly does M.Ped control anyhow? Is this just a GH4 thing to leave it alone?/

Master Pedestal is there for getting black levels in sync with other cameras, in a live shooting environment (eg. broadcast). It reduces dynamic range and compresses all the other levels upwards if set to positive. If set to negative you could clip your blacks unnecessarily. Basically, leave it at 0.

Lastly, the doozy:

Luma 0-255 vs 16-235

I understand this maps pixels and luma shades to different values for broadcast, but I don't *really* understand this - as in, what potential downsides mapping to 0-255 could have, how it interacts with my software and output and any needed setting changes in workflow, why and for who this is an option at all, and what I miss out on IQ wise by leaving @ default 16-235.

I think Thomas Smet answered this one quite well, so I shall defer to his answer! In short - leave it at 0-255 unless you need broadcast output to an external recorder, then set it to 16-235. Whatever you do, DON'T use 16-255 as it's non-standard in every way!

Thank you both!!

Paul - if I buy Leeming LUT for GH4 today, will that support GH5 or would it be a separate purchase?

I won't know until I have a GH5 in hand. I tried my existing GH4 V-Log LUT on the GH5 footage from Neumann Films and it looks pretty good at first glance, but for the Cine-D side I won't know as it's got a new sensor and new colorimetry according to reports.

Most likely I'll have a new LUT package for the GH5 to get it really optimal, since it's capable of 10bit 4:2:2 internal, which will allow for better LUT precision, but I have already announced a discounted rate for existing GH4 LUT owners on my Facebook group page, so you will get any GH5 LUT cheaper than first time buyers at least!

HTH!

Cheers,

Paul

 

[Joshua_G]

…
Regarding -3 Sharpness for non-native lenses, that's the setting which I found gave the most sharpness before artifacts started creeping in. Below -3 the image got softer …
…

When the image is soft, isn't it better to do some sharpening in the NLE, rather than in the camera?

 

[Joshua_G]

Contrast and saturation are a bit different because you are recording to a limited format. Typically any time a person tried to force a flat look with the contrast and color it artificially hurt the image. Color is also subjective and also easy for the camera to do so there isn't a processing speed concern or a concern over a fake look by letting the camera handle the color.
…

Exactly All these people who recommend flattening everything never think about the fact that you then have to restore the colour from less information, than if you had left it alone to begin with! And then they complain and say the camera is noisy....

Concerning contrast, which is Luma information, it seems to me that the amount of details on the Luma channel is being determined by the bit-depth (8bit vs, 10 or 12bit). Lowering the contrast in the camera affects only the contrast curve that the in-camera processing does, but, unless one is under-exposing, it doesn't affect neither the amount of details on the Luma channel nor the amount of noise.

Concerning color saturation, it seems to me that the amount of details on the color channels (RGB) is being determined by the color matrix (4:2:0, 4:2:2 or 4:4:4). Increasing the color saturation doesn't add any information, or details to the colors.

When shooting with low contrast and low saturation in the camera and increasing them in post, one have better and more refined control on the outcome, without losing anything (like increased noise, or loss of colors details).

Am I wrong here?
If yes, what have I overlooked?

 

[sigamy]

Thank you all for passing this information on. Would everything here apply to the G7 as well?

 

[filmguy123]

This is really helpful thread. I know there are other GH4 setting threads but as I mentioned, this stuff is often spread out over the net and over years of trial, error, and argument. It's super helpful to hear from people really experienced with that and not random YouTubers from a year ago, or trying to dig through ancient threads with no clue if who posted what has since changed their mind.


A couple questions I have in this is, does shooting in CineD necessitate a certain grading workflow past the obvious bump contrast, saturation, etc.? I'm not thrilled with some of the images I am getting so far, but I am still experimenting.

Another question is, what is the difference between shooting CineD with -3, -5, etc. than doing CineV with the same and perhaps just bumping down saturation and contrast a notch in CineV?

 

[Thomas Smet]

This is really helpful thread. I know there are other GH4 setting threads but as I mentioned, this stuff is often spread out over the net and over years of trial, error, and argument. It's super helpful to hear from people really experienced with that and not random YouTubers from a year ago, or trying to dig through ancient threads with no clue if who posted what has since changed their mind.


A couple questions I have in this is, does shooting in CineD necessitate a certain grading workflow past the obvious bump contrast, saturation, etc.? I'm not thrilled with some of the images I am getting so far, but I am still experimenting.

Another question is, what is the difference between shooting CineD with -3, -5, etc. than doing CineV with the same and perhaps just bumping down saturation and contrast a notch in CineV?

Cine-V is more of a baked in look while Cine-D gives you more room to adjust that look in post. Cine-V is also a higher contrast profile and has a bit less dynamic range. Cine-V is a good look but its really for those who don't want to grade a lot and want to use the camera as is. You can grade all the profiles to a certain degree but the point of shooting flat is the not have an 8bit format that already has too much of a look baked in. This is why I prefer V-log since you don't really get much flatter than that outside of shooting raw.

 

[joe12south]

I've found with NR at -5 on the GH4, it has its own natural grain/noise and don't really need to add more, unless I'm going for an effect.

Indeed. I've always found the luminance-biased noise to look similar to grain. It is mostly only in the shadows, though.

 

[joe12south]

Oh, and between lighting and grading, you want editing.

Oh yeah. ;-)

Seriously, though, one thing that screams amateur is overly long takes, often forced because there is not enough coverage. A buddy of mine just shot a very nice looking feature on the GH4, but because he was so rushed, he simply did not have enough coverage of each scene and leans on overly long master shots. It's hard to watch.

 

[Vesku]

Indeed. I've always found the luminance-biased noise to look similar to grain. It is mostly only in the shadows, though.

Even NR -5 softens shadow noise. It starts to blur dark tones right after iso200. Iso400 has less detail in shadows than iso200.

If someone needs more accurate sensor noise than NR-5 you can use 4k photo for video shooting. It has less NR and a nice bonus of showing all exposure values with all modes.

 

[joe12south]

I would encourage you to evaluate saturation levels with a chart. If the image being recorded is too "hot", you can actually clip color channels. For example, the "Normal" profile is definitely unnaturally saturated (try -3 to bring it in range).

 

[filmguy123]

How does grading Vlog compare to working the Leeming LUT CineD settings? Similar, but just with more DR? Or...?

 

[Thomas Smet]

How does grading Vlog compare to working the Leeming LUT CineD settings? Similar, but just with more DR? Or...?

Depends who you ask. I love V-log because it is the closest we will ever get to raw recording on these cameras. No it isn't the same as raw but it is much closer than anything else we have. I like having a flat image to work with that has no baked in settings at all. I only use V-log external at 10bits however which in my view is really a must for a log recording. V-log is not perfect but with care it becomes a whole new way to work with the GH4.

If you are shooting on the GH4 and do not want to use an external recorder I would highly recommend Cine-D + Leeming LUT. It is so much easier to shoot with.

 

[filmguy123]

Question on exposure and ETTR:

Let's say you're filming a scene, and there's some rather unimportant highlights in the background that you can let blow out. Or not - because you have enough DR in your histogram to save them, too, and still keep all shadows above baseline.

Save them? Or just let them blow out if they aren't too important, and lift the shadow baseline even more into neutral to give even more light to shadows and midtones?

Just wondering if I only need to make choices to blow out some highlights when it means clipping the shadows, or if I should choose to do this even when everything fits well into the histogram in order to give an even cleaner image to shadows/mids for my post grading work?

 

[Thomas Smet]

Question on exposure and ETTR:

Let's say you're filming a scene, and there's some rather unimportant highlights in the background that you can let blow out. Or not - because you have enough DR in your histogram to save them, too, and still keep all shadows above baseline.

Save them? Or just let them blow out if they aren't too important, and lift the shadow baseline even more into neutral to give even more light to shadows and midtones?

Just wondering if I only need to make choices to blow out some highlights when it means clipping the shadows, or if I should choose to do this even when everything fits well into the histogram in order to give an even cleaner image to shadows/mids for my post grading work?

It is really up to you just like normal exposure is up to you and depends on what you want to save. Eventually you will have to sacrifice those highs so you don't grossly under expose the important part of the shot. It comes down to what are people going to see when they look at the shot.

 

[joe12south]

How does grading Vlog compare to working the Leeming LUT CineD settings? Similar, but just with more DR? Or...?

V-LOG L > REC709 (input LUT) > grade to taste
Cine-D > Leeming One (input LUT) > grade to taste

Essentially the same workflow. In fact, there is a Leeming One LUT for both camera profiles.

If you're recording internally, I would encourage you to use Cine-D + Leeming One. The bit-starved internal processing pipeline can't handle the low contrast V-LOG curve without introducing significant visual artifacts. If you're using an external recorder, V-LOG L will net you a bit more dynamic range. Color response is the same either way.

 

[joe12south]

It is really up to you just like normal exposure is up to you and depends on what you want to save. Eventually you will have to sacrifice those highs so you don't grossly under expose the important part of the shot. It comes down to what are people going to see when they look at the shot.

If your subject is a person, then you should expose for their skin. Getting that right is most important. Only after doing that, should you adjust the surrounding environment to increase or decrease light levels as desired (Adding fills to shadows, filtering harsh lights, etc.) to create the tone you are going for.

I know you'll most commonly see people advising your expose to the right, or expose a stop low, or something similar, but that's thinking like a camera tech, not a cinematographer. Light your subject. Expose for your subject. Then deal with everything else. This is a better path to "cinematic" images.