Basic Picture Profile Questions

M

Mark-P

Well-known member
I'm new to the picture profile aspect of shooting and editing with the A7S2.

1. Do all the camera manufacturers use universal gamma curves and just rename them S-Log, C-Log, etc. or are they all completely different and proprietary? Does S-Log2 have anything in common with C-Log2?

2. I have some LUTs that I plan to apply to some footage. What is the best profile to use if I only plan to do simple adjustments to the WB and exposure? The footage will be shot in daylight 4K 24fps.

3. From what I understand, the different profiles offer different amounts of dynamic range and S-Log2 and S-Log3 have the flatest and highest amount of dynamic range. Is the primary reason why you wouldn’t always use them being that you need to use a higher ISO with them and they require much more work grading in post?

Thank you in advance.
 
Log spreads the sensor dynamic range into the recordable data.

Each maker does it a bit different but similar.

The relevant manufacturers luts untangle that spread of dr

--

If you want to make only minor grades you might be better to record in a baked profile and not use a lut.

--

The primary reason to use slog is to capture the most sensor data and manipulate to taste in post.

The secondary reason is that you may not like Sonys baked looks or they may not fit your scene dynamic range.

--

It is worth understanding that with an 8bit camera that recording a wide dynamic range compromises how much tonal detail you record per stop of light and that recording lots of data per stop gives a rich image. Recording a wide dynamic range is not always the best thing
 
1. I do not know enough yet about color grading so I plan to just adjust exposure and white balance in Premiere, and I was hoping that the LUTs that I intend to use would get me where I want to be with regard to my desired look-and-feel. Perhaps with some minor adjustments to the settings that the LUTs apply. With my clarification, which profile should I record in on the A7S2?

2. Although I am not planning to do the grading myself, I thought I would want to use slog in order to capture the highest amount of dynamic range and allow the LUT to essentially color grade my footage. Am I misunderstanding how this works?

3. "It is worth understanding that with an 8bit camera that recording a wide dynamic range compromises how much tonal detail you record per stop of light and that recording lots of data per stop gives a rich image." I thought the whole point of recording slog was to get a flat profile with a wide dynamic range with lots of tonal detail?

Thanks in advance.
 
Consider a scene with two stops of brightness change

Your 8bit camera records 255 tones per colour channel -

With a two stop profile you could allocate 127 tones to each stop.

With a 10 stop profile you could only have 25 tones per stop.

The more dr you record the less tonality you get - it is a simple mathematical trade off.

---
Yes shooting slog and applying a lut should "grade" your image in one click.

--
Which profile? Well that really depends on your scene - but slog covers all scenes - othe profiles may cover specific scenes better than slog.

Personally I'd stick with slog until you find it a problem and seek other routes

----
 
morgan_moore said:
With a 10 stop profile you could only have 25 tones per stop.

The more dr you record the less tonality you get - it is a simple mathematical trade off.
Click to expand...
That is not quite an accurate description.

Sure, if a camera's dynamic range in log recording is 10 stops, you have nearly 25 code values per stop in the luminance channel - on average across the stops. (Or 22 or 24, or 4 times any of those, depending on what digital encoding the camera uses.) The distribution of code values is not uniform. Towards black, each stop will have fewer code values. There are infinitely many stops going down to black, and most of them have no code values.

Now say you take that same camera with 10 stops of dynamic range and set it to a standard BT.709-like profile, instead of log. Does the camera's dynamic range suddenly decrease from 10 stops? No, it's still 10 stops. The camera has the same ability to distinguish some darkest grey value near black, and the noise in the shadows has not changed. The gamma curve does not determine the camera's dynamic range. The maximum dynamic range that can be captured is a characteristic of the camera, not the gamma curve. All of these gamma curves, log or not, are defined down to 0% reflectance, which means the curves all have infinite dynamic range. Just having a flatter curve doesn't give you any more dynamic range. This is a common misconception and it really needs to be put to rest.

Changing the gamma curve changes how code values are distributed across the dynamic range. Log curves tend to have fewer code values per stop above middle grey, and more stops above middle grey, compared to BT.709-like curves. That means more code values in the stops furthest below white clipping. 8-bit BT.709, for example, counting up from black has one code value in the 11th stop below white clipping and then one code value in the 10th stop below white clipping. 8-bit S-Log2, in comparison, has a code value in the 15th stop below white and then two in the 13th stop below white. So the log spaces get more precision in the deep shadows by sacrificing precision in the midtones and highlights. A log gamma can be good if your camera has a large dynamic range and you might want to lift deep shadows later. But sometimes a log gamma's distribution of code values is a bad thing. In 8-bit recordings, I recommend against using S-Log2, S-Log3, Canon Log 2, and V-Log L. These curves are too flat in midtones and highlights, not allocating enough code values in those ranges and giving them poor color precision and causing visible banding. Banding becomes worse when these flatter curves are combined with a wide recording gamut. Canon Log 1, Canon Wide DR, and Sony HyperGammas, for example, are curves with extended highlight ranges that have better distributions of code values in 8 bits.

On some cameras, selecting a log curve or a curve with an extended range of highlights like a HyperGamma will change the camera's operating mode such that the camera's highest exposure that it can record before clipping to white is increased compared to recording in a standard gamma curve. This actually does increase the dynamic range. But it's not the curve that's doing it. It's just that the camera wasn't recording all of the dynamic range that it had to offer in its standard gamma curves.

A lot of cameras don't give you any more highlight exposure range when selecting a log curve, due to the fact that the log curve is not available at the same low ISO settings. Take for example a camera that has BT.709 available at ISO 100, and an extended-highlight gamma curve with 800% the highlight range of BT.709. If that 800% extended-highlight curve is only available at ISO 800 and up, both curves have exactly the same white clipping exposure at their respective lowest ISO settings, and thus the same dynamic range. The extended-highlight gamma curve places its middle grey point at a lower exposure, which is how it can have more stops above middle grey. If you expose the scene for each curve to place 18% reflectance at the gamma curve's middle grey value, the 800% extended-highlight curve will need the exposure set three stops lower. That's where the extra highlight range comes from.

To answer the OP's questions:

The camera log curves like S-Log2 and Canon Log 2 are all proprietary curves defined by their manufacturers. They have some things in common, but they're different and need to be treated differently in post. I have picture profile recommendations for shooting with Sony cameras. Since the A7SII only records in 8 bits, Cine1 or 2 will be a better choice than an S-Log.
 
Last edited:
balazer said:
That is not quite an accurate description.

The distribution of code values is not uniform.
Click to expand...

Save yourself some typing by noting the word in my post 'could' not 'is'.

---

As for log increasing the DR - no - log only records what the sensor can record, some other settings however clip the digital file at a narrower DR than what the sensor can record, meaning log can deliver a wider DR.

As for not giving more highlight range you are nailing yourself in by suggesting that mid grey should be recorded at a fixed level. Mid grey should be exposed depending on many things. Scene, gamma, final desired look etc.

--

My core point is that with an 8bit camera and a narrow DR scene the op may get better results by choosing a profile that is not log.
 
morgan_moore said:
As for not giving more highlight range you are nailing yourself in by suggesting that mid grey should be recorded at a fixed level. Mid grey should be exposed depending on many things. Scene, gamma, final desired look etc.
Click to expand...
I said 'if', not that you need to set exposure that way. The placement of middle grey in the gamma curve is related to how the manufacturer rates the highlight range of the profile.

My core point is that with an 8bit camera and a narrow DR scene the op may get better results by choosing a profile that is not log.
Click to expand...
No disagreement there, but you said "The more dr you record the less tonality you get - it is a simple mathematical trade off.", seemingly saying that the total dynamic range delivered by the log profile should be the consideration of the tonality you get. It's not as simple as looking at the total dynamic range recorded by a profile. You need to look at how the tonality is distributed over the dynamic range.

BT.709 in 8 bits, for example, has 10 stops of dynamic range if you ignore black - the same as a log profile on a 10-stop camera. But that doesn't mean BT.709 is equivalent to log. HyperGamma 7 in 8 bits, to give you another example, has 15 stops of dynamic range if you ignore black, the same as S-Log2 in 8 bits. But that doesn't make HyperGamma 7 equivalent to S-Log2.

The profile that distributes its code values most similarly to your final output is the one that minimizes banding. So if your final output is BT.709-like with no highlight roll-off, shooting in BT.709 will minimize banding. If your final output is BT.709-like but with a large highlight roll-off, a HyperGamma will minimize banding.
 
Want to jump in this discussion - for perhaps some clarification and to make sure I've understood Balazer correctly. "The profile that distributes its code values most similarly to your final output is the one that minimizes banding."
So - if the final deliverable for the video you are shooting is the web - this mean sRGB, correct? Which, I believe is very close to BT / Rec709 - with a slightly different gamma - but colors that are very similar (in practical purposes). So, if the camera's sensor you'll be using is capable of recording 14 stops of dynamic range, and you want to be sure your video won't have any banding when posted online, are you proposing the operator choose a profile in camera that only will capture 10 stops (assuming 8 bit instead of 10 files)?
 
Last edited:
This is hard to answer.

So many topics!

709/srgb in theory can only handle about 7 stops.

Flattering the roll to clip can add a couple of stops.

So tht is nine stops.

You should control your scene to nine stops by using fill light and Nd.

So yes the best is a profile that has has nine stops with the top two being "roll off"

A profile with more stops will have that log flat wooden feel unless you grade.

Hyper-gamma is therefore logical.

And will deliver the best quality.

But this "perfection" does not account to the real world where your exposure might not be perfect and you may not be able to control the scene dynamic range.

You might be able to get this perfection in a studio.

Out in the world you make trade offs - a camera profile that gives wiggle room but risks banding when you wiggle in post.

Ultimately location experience and practice will lead you to slecting the right camera setup for the right shoot/scene
 
One problem with Sony PP is that they can go very horrible if the scene dr is beyond the profile dr..
Nuclear highlights
Wooden skin
 
morgan_moore said:
This is hard to answer.

So many topics!
Click to expand...
Agreed - it is hard to answer and I'm not sure a one-size fits all approach is the best one. 10 bit Slog adds much flexibility and can "hide" being off on your exposure. But to the OP - while I do not own nor have I shot with the A7s2 - just know - if you are shooting internally - then you are not only in 8 bit but also in 4:2:0 color space - and pushing the image around more than just subtle tweaks in post will likely not go well. In certain situations - keeping exposure consistent - slog and a conversion lut in post might work just fine. In a very low light scene - or one like Sam talks about presenting narrow (say 4 stops) of dynamic range - then another profile would likely be better. But I would suggest testing these yourself in various lighting conditions rather than believing what someone else tells you.
 
I'm also confused by Picture Profiles and Creative Styles.

Are Creative Styles intended for people who do not intend to color correct their photos or color grade their video? Should it even be turned on if you intend to color grade video?

I want to capture as high of quality video as possible, and edit/ color grade in Premiere. Since I have limited experience all around, I'm concerned with what I have read with the ISO requirements, not being able to use the light meter, and some other things that I have read about S-Log2 and 3. I have also read that using Cine4 is a good alternative since it does not require as high of a ISO and it maintains a good level of dynamic range. Does this sound correct?

Thanks.
 
You must log in or register to reply here.
Back
Top