SIGMA FP with ProRes RAW and BRAW !

[Llaasseerr]

7 hours ago, OleB said:

Have investigated this further.

Look what I have found in regards to the fp behavior in RAW on the Atomos.

Monitoring of RAW input in:

- Native – Linear to VLog for use with 3D LUTs – can use any existing VLog LOOK LUTs
- Rec709 – to view RAW input in SDR up to 6.7 stops of dynamic range
- HLG – to view RAW in HLG with up 10.5 stops of dynamic range
- PQ – to view RAW in PQ/HDR10 with up to 10.5 stops of dynamic range, for scenes with extended spectral highlights.

Am about 99% sure that the camera is showing Vlog when switched to native.

 

OK there you go. So yes it's confirmed that Atomos decided "Native" = Vlog. After you said it, I googled it and found the AtomOS release notes update PDF that you're quoting from. This certainly reduces a lot of uncertainty. It's a shame the Atomos rep was not able to clarify this when I emailed them a while ago.

The Vlog curve spec defines 8 stops above middle grey, so the max linear value is 46.0855. (0.18*2*2*2*2*2*2*2*2). The fp sensor clips way before the max value that Vlog was designed to hold so it should not be an issue. Based on the test clip you recently uploaded with the sun clipping out of the window, it's clipping at about 0.833 in Vlog so that would be like 83 IRE. In scene reflectance terms once converted from Vlog to linear in Nuke, that's 9.36. So it's clipping more than 2 stops below the Vlog max value.

This screengrab illustrates where that is in relation to the entire scene reflectance range defined by the Vlog curve. Specifically this shows a Vlog to linear transform, so it's the inverse of a Vlog curve.

At the bottom, the input numbers represent the equivalent of an "IRE value" (0.83) and the output numbers represent the scene linear value (about 9.4, matches Nuke). So as you can see, there's a ton of headroom remaining that will never get used.

 

7 hours ago, OleB said:

As far as I understand from the Vlog white paper 90% reflection is located at about 60 IRE, on ISO 100 the fp clips at about 62 IRE. The higher you go with the ISO (up to ISO 800). The more headroom on top is available.

 

That means on the same Vlog to linear transform the input is about 0.59 and the output is 0.9. Here's a screengrab showing you how much extra headroom you get for capturing highlights above that. You basically get all of the curve to the right of the green arrow I drew.

 

So while it's not ideal that Sigma don't have their own log curve, Vlog is okay as a monitoring option on the Ninja V.

8 hours ago, OleB said:

However I am not too sure if this is at least for me a good way of working since I do not know if the next step would be now to figure out the native ISO of the camera for maximum DR in Vlog?

It doesn't matter what ISO you use, you will never use up all of Vlog. This is assuming there's no funny business going on where the Atomos is somehow clipping the highlights as they come in from the camera, but I doubt that.

8 hours ago, OleB said:

For now I prefer the manual colorspace matching in FCPX directly from the untouched linear RAW, which gives me great results. And with the PQ view in the monitor I can get a really good idea on how the picture will look like.

Just to clarify, it doesn't matter which monitoring option you choose on the Ninja V, the underlying file is exactly the same. The difference between monitoring in PQ or V-log is that Vlog requires you add your own LUT, while PQ out of box creates a "display" image that takes advantage of the extra nits of the Ninja V but it's just a consumer HDR display standard for TVs.

So if you want to view the same image while shooting as in your timeline, you would match the display settings in both the Ninja V and FCPX, which is all you do in any software/hardware combo anyway. All that's been figured out is that the "Native" input is a known log format (Vlog), so that means it's easy to figure out how to load the look from your editing/grading software into the Ninja V. 

 

 

 

 

[OleB]

Thank you for the extensive write-up Llaasseerr.

Agree, that with the V-log LUT loaded on the Ninja V you have something to work with. If you select the same in FCPX the result is very close to what has been shown on the monitor.

Now the only thing I do not understand ist that in native (V-log) mode the clipping points of the sensor are different with the various ISO settings. I would have expected that the DR is the same for every ISO setting...Let me give you some values to show you what I mean.

ISO 100 - clips at about 65 IRE 

ISO 200 - clips at about 75 IRE

ISO 400 - clips at about 82 IRE

ISO 800 - clips at about 90 IRE

ISO 1600 - clips at about 90 IRE

ISO 3200 - clips at about 85 IRE

ISO 6400 - clips at about 90 IRE 

rest stays at 90 IRE

 

The noise floor / black level stays the same for every ISO setting. About 8 IRE.

What might be behind this? When reading these figures I would assume that for max DR I should use ISO 800...

When inspecting the scene with the LUT thrown on you are facing a situation where the absolute whites on ISO 100 are not reaching red, and cannot go higher than yellow.

This is confusing me, because I am not understanding the intention behind this, if any.

If you can shed some light on this, would love to hear 🙂

 

[Llaasseerr]

7 hours ago, OleB said:

Thank you for the extensive write-up Llaasseerr.

Agree, that with the V-log LUT loaded on the Ninja V you have something to work with. If you select the same in FCPX the result is very close to what has been shown on the monitor.

Now the only thing I do not understand ist that in native (V-log) mode the clipping points of the sensor are different with the various ISO settings. I would have expected that the DR is the same for every ISO setting...Let me give you some values to show you what I mean.

ISO 100 - clips at about 65 IRE 

ISO 200 - clips at about 75 IRE

ISO 400 - clips at about 82 IRE

ISO 800 - clips at about 90 IRE

ISO 1600 - clips at about 90 IRE

ISO 3200 - clips at about 85 IRE

ISO 6400 - clips at about 90 IRE 

rest stays at 90 IRE

 

The noise floor / black level stays the same for every ISO setting. About 8 IRE.

What might be behind this? When reading these figures I would assume that for max DR I should use ISO 800...

 

 

This clipping point is in line with the characteristics of the PRR test clips you uploaded before, and broadly speaking, also the behaviour of the DNGs, so I don't see anything strange here. We established that up to ISO 800 was an amplification of the base ISO 100 signal so the clipping point is raised accordingly. As the signal increases it should also be raising the noise floor obviously, but we noticed that the noise floor stays pretty low which is a nice win. 

For whatever reason, this taps out above ISO 800, so there are diminishing returns as far as overall dynamic range. When a camera is baking in the ISO change, the clipping point would ideally continue to increase. Maybe Sigma prioritized a clean noise floor instead of highlights for the higher ISO ranges (>800).

So using ISO 800 is a decent option. Regarding ISO 3200, the higher base ISO, it has a lower clipping point but my observation is that it has a proportionally lower noise floor. So my recommendation for absolute max DR was to shoot 3200 at least -1 stop underexposed since you could raise the noise floor and it would be the same as ISO 800/1600, but with additional highlight headroom. If you think about it, this is in line with the expected behavior with higher base ISOs: lower noise floor and slightly lower highlight clipping point. I understand this is a bit of a headache to track for a marginal gain, so shooting ISO 800 is a good rule of thumb to get max DR to keep things simple.

 

7 hours ago, OleB said:

When inspecting the scene with the LUT thrown on you are facing a situation where the absolute whites on ISO 100 are not reaching red, and cannot go higher than yellow.

 

Well again, the IRE value is LUT or view gamma-dependent, but your numbers make sense. The base ISO 100 has a lower clipping point than ISO 800 which is baking in an amplification of the signal, but somehow also keeping a similar noise floor.

For example with this ISO 800 test clip, when importing as V-log into ACES and viewed through the standard ACES output display transform the clipping point is in the high 90's so it stands to reason that ISO 100 would clip lower.

 

 

 

[OleB]

Thank you very much! I think I now got it. Weather is getting really nice here the next days, so finally I will try to capture some footage with this new knowledge. 🙂

When is your camera supposed to arrive? Guess you have a good starting point established now as well.

[Llaasseerr]

I'm actually picking it up today! I'll probably be super lazy about getting around to learning it and shooting with it, but hopefully I'll post a few tests here eventually. Thanks for all the help from people here willing to post test footage.

[Llaasseerr]

@OleB just an update on my early experience shooting with the Ninja V, monitoring as Vlog and with a correct LUT expecting Vlog and transforming to Rec709 works as expected, which is great and can replicate my timeline color at the start of a new Resolve project. In fact when I rewatched the Atomos setup video on Youtube, it even mentions on the screen that the "native" encoding is Vlog (in addition to the firmware release notes PDF you pointed out before), so it was staring me in the face before but I missed it.

So that's the good news. The bad news is like you said, it's also been awkward in my early testing to get exposure with a light meter. Despite all my ranting about how I don't really like false colors and IRE and whatnot, it is totally legitimate to expose middle grey at around 42 IRE on the waveform when displaying as Vlog as a sanity check for the light meter reading. But so far I haven't got middle grey to hit 42 IRE when exposing as per the light meter. This is easy to do with other cameras I've used. I did try with the underlying ISO that you suggested (either 100 or 640) and will do a few more tests to get to the bottom of whatever weirdness is going on there. Possibly I just made a stupid mistake. 

What would be really great is if Sigma did a firmware update that allowed you to monitor with the Ninja V in Vlog, but record DNG raw internally or to the SSD. Currently it's either/or. Because obviously through the Ninja V we get all the highlights visible. Basically it adds the log monitoring capability that Sigma did not add. ProRes Raw is a good option though, but for example as opposed to DNG I don't think you get any highlight recovery.

[Llaasseerr]

@OleB I atted you for my last post but it wasn't highlighted, so you may not have seen it. I was having issues exposing, but I realized I made an error with my ND filter. All good now. I was able to expose 3200 by setting the light meter for 640. Thanks for your detective work.

Seems to work fairly well with getting middle grey at close to 42 IRE on the Vlog display on the Ninja, then I can flick on a viewing LUT. I could also do it directly through the LUT. I just need to confirm in Resolve where middle grey falls on the waveform.

[OleB]

@Llaasseerr I am now learning to use the camera in the V-log way (with and without a LUT).

What I simply do not understand is that if the camera records RAW data, ISO value is basically just metadata. That being said since the fp has dual iso it must have two native EIs. The DR of the camera is always the same no matter the ISO setting both native EIs. 

Have seen couple of videos explaining that in a dark environment you should use a low EI value and in bright a higher EI. They try to explain that with selecting different ISO values, you are just shifting the middle grey point.

On one hand it makes sense to let as much light into the camera in a dark environment as possible, and in a bright environment do the opposite. But yet I am not 100% sure I understand what they mean.

For the fp I have noticed that with false colors turned on (and the V-log LUT) only as from ISO 320 you are touching 100 IRE. So I assume that is in reality the native ISO value of the sensor. Or is that already wrong?

Maybe you are willing to explain so maybe I can understand this?

[Llaasseerr]

6 hours ago, OleB said:

@Llaasseerr I am now learning to use the camera in the V-log way (with and without a LUT).

What I simply do not understand is that if the camera records RAW data, ISO value is basically just metadata. That being said since the fp has dual iso it must have two native EIs. The DR of the camera is always the same no matter the ISO setting both native EIs. 

It is basically just metadata except for when it switches from base ISO of 100 to 3200. There is also a difference in DR between different ISOs because some of them are clipping the highlights at the same value even as they raise middle grey. I also noticed there seems to be some good subtle in-camera work taming the noise floor with, for example, ISO 800 vs if you shot ISO 100 and boosted by +3 stops.

6 hours ago, OleB said:

Have seen couple of videos explaining that in a dark environment you should use a low EI value and in bright a higher EI. They try to explain that with selecting different ISO values, you are just shifting the middle grey point.

From my tests with DNG, it's clear to see the advantage of ISO 3200 over ISO 800 when it comes to an improved noise floor, at the expense of about 2/3 stop less highlights. I would need to double check the behavior with the Ninja V though. It is nowhere near as dramatic as the higher base ISO in something like a Sony a7sIII/FX3/FX6 though.

Here's an example using 12-bit DNG. On the left is shadow detail at ISO 3200 and on the right is ISO 800, both exposed for middle grey -2 stops (to protect more highlights) with the Sigma kit lens which is very contrasty. I'm pushing the exposure up in the extreme so you can see how the ISO 3200 is holding the color in the shadows, not clipping at black and also not contaminating with green.

What interests me about 3200 besides shooting in under lit scenes, is using it with wide contrast scenes if I underexpose it by 1-2 stops. So I can take advantage of the better noise floor while protecting highlights. Otherwise 800 (base ISO 100) is a good choice.

 

6 hours ago, OleB said:

On one hand it makes sense to let as much light into the camera in a dark environment as possible, and in a bright environment do the opposite. But yet I am not 100% sure I understand what they mean.

 

It's about setting a usable key/fill ratio. In an uncontrolled setup where you're not lighting from scratch it would be about ideally adding enough fill to keep shadows from being at zero and enough diffusion to tame extremely harsh bright light.

6 hours ago, OleB said:

For the fp I have noticed that with false colors turned on (and the V-log LUT) only as from ISO 320 you are touching 100 IRE. So I assume that is in reality the native ISO value of the sensor. Or is that already wrong?

I don't see why that would be the case.

 

 

 

[OleB]

2 hours ago, Llaasseerr said:

I don't see why that would be the case.

I came to this because if you select something lower than ISO 320 you will reach sensor clipping before 100 IRE in the LUT. Do not know if this is intentional or I am misunderstanding something though...

[Andrew K White]

@Llaasseerr Hi, I've read this discussion from the beginning trying to grasp my head around coloring with the FP. I'm curious about your statement that you think the false colors on the camera can show the clipping levels for the sensor itself, since that isn't my experience when I view them. For example, at ISO 800 the false colors on the camera show clipping (whereas for the same scene at ISO 100 the false colors do not show clipping), and I'm able to bring the highlights into viewable range in post after recording at ISO 800.

[Llaasseerr]

56 minutes ago, OleB said:

I came to this because if you select something lower than ISO 320 you will reach sensor clipping before 100 IRE in the LUT. Do not know if this is intentional or I am misunderstanding something though...

There's no rule that a LUT has to necessarily show sensor clipping at 100 IRE, especially at a low ISO. If you look at the linear clipping value at low ISO especially on the Ninja V which does not use highlight reconstruction, it's a very low value and gives the impression that the fp is bad with highlight detail.

So if you transform that low value to Vlog then apply a Vlog to Rec709 curve, I can't imagine it would get anywhere near 100 IRE, but it would also not be indicative of what is the native ISO.

Depending on how aggressive the highlight rolloff is on a log to 709 LUT, it can be hard for anything to get to 100 IRE so viewing in log might be a better way to measure max IRE from the sensor.

 

[OleB]

Ok thanks. That seems logical. 
 

However we now have established that ISO100-800 is the same in regards to sensor clipping. 
 

If you check the fp dual iso manual. The most balanced amount of stops above and below middle grey is with ISO 800. comparing this chart to the one from ARRI Alexa is very similar with the Alexa having more steps. But ARRI gives 800 as native value. 

As from my understanding than both cameras could have an EI of 800 as base and decreasing ISO relocates the stops more to the dark side. With the Alexa you can do the opposite as well, while with the fp you will not get more steps available for the highlights than with ISO800. 
 

 

 

[OleB]

To follow up on this topic. I am now working the intended way with the V-log conversion and LUT. From what I see the color saturation is about 50% off. Meaning not saturated enough.

Fortunately I have a Leica SL2-S available so that I can check how the color checker should look like after converting L-log to Rec709. 

If I go the way with the V-log files and their LUTs it is not a one click solution, but at least it is close after pulling up the global saturation.

Guess that is what we have to live with, because there is no Sigma log and no proper Rec709 LUT.

[OleB]

Experimenting with Resolve right now to figure out why the above is happening. Not very familiar with the software, but what I did, I imported a PRR clip to FCPX and exported it into ProRes 422 HQ without any adjustments.

Now I have imported the clip to Resolve and put the V-LOG to Rec709 LUT in the last serial node. Before that node I have created a color space transform. Now testing various settings. 

Could it be, that there is some sort of input gamma issue and because of that the colors are not saturated enough?

[OleB]

Okay here comes the result.

Fiddling with the saturation knob is not correct at this stage. I seem to have missed one important step, which I have now found after reading through Panasonic website. The PRR RAW file needs to be brought into the V-Gamut first before selecting the Rec709 LUT. They have a technical LUT for this to download.

Steps:

1. RAW to LOG convert in FCPX info tab set it to Panasonic V-LOG

2. Add 2 custom LUTs at the end of the editing chain.

3. Select Panasonic V-LOG RAW to V-Gamut LUT <= This step brings the colors back to where they should be 🙂 

4. Select Panasonic V-Gamut to Rec 709 

 

FINALLY!!

 

[Llaasseerr]

On 6/29/2022 at 4:51 PM, Andrew K White said:

@Llaasseerr Hi, I've read this discussion from the beginning trying to grasp my head around coloring with the FP. I'm curious about your statement that you think the false colors on the camera can show the clipping levels for the sensor itself, since that isn't my experience when I view them. For example, at ISO 800 the false colors on the camera show clipping (whereas for the same scene at ISO 100 the false colors do not show clipping), and I'm able to bring the highlights into viewable range in post after recording at ISO 800.

Maybe what you are referring to is that with the Ninja V, Atomos decided to encode the linear raw image as Panasonic V-log, which does bring the highlights into a viewable range as a log image and would allow false color monitoring if the false color tool is expecting a Vlog input. So basically they are making up for one of the camera's deficiencies.

I need to check more carefully but I'm pretty sure the internal image as displayed on the camera is clipping the highlights, at least in "OFF" color mode which is what I set it to. Because all it's doing is applying a Rec709 curve with no highlight shaping to a linear raw image and that causes anything above a certain level to get clipped.

I hope the false color tool in the v4 firmware can be used to accurately check middle grey - but not highlights. This is assuming that the false color tool does read values directly from the sensor, which someone else mentioned.

[Llaasseerr]

9 hours ago, OleB said:

Okay here comes the result.

Fiddling with the saturation knob is not correct at this stage. I seem to have missed one important step, which I have now found after reading through Panasonic website. The PRR RAW file needs to be brought into the V-Gamut first before selecting the Rec709 LUT. They have a technical LUT for this to download.

Steps:

1. RAW to LOG convert in FCPX info tab set it to Panasonic V-LOG

2. Add 2 custom LUTs at the end of the editing chain.

3. Select Panasonic V-LOG RAW to V-Gamut LUT <= This step brings the colors back to where they should be 🙂 

4. Select Panasonic V-Gamut to Rec 709 

 

FINALLY!!

 

 

Right, you get a desaturated image with a wide gamut recording. So in addition to the log transform, you need to do a gamut transform to make it look correct for your viewing device (probably a rec709 or sRGB screen).

If you break it down, the log transform is a 1D curve (can be a 1D LUT). The gamut transform is a matrix transform. Both these operations can be collapsed into a LUT. So when Panasonic provides a Vlog to Rec709 3D LUT, what it's really doing is concatenating both those operations into one.

I am not 100% clear on this, but Atomos seem to have established a native gamut for the Sigma recording that is different to Vgamut. But if I recall, you can transform the raw file in FCPX immediately to Vlog/Vgamut in the inspector before doing anything else. After you do that, you can apply the LUT and you won't get the saturation issues, which is actually just a gamut mismatch.

Hopefully the images then match on both the Ninja V and the FCPX timeline.

[OleB]

10 hours ago, Llaasseerr said:

 

Right, you get a desaturated image with a wide gamut recording. So in addition to the log transform, you need to do a gamut transform to make it look correct for your viewing device (probably a rec709 or sRGB screen).

If you break it down, the log transform is a 1D curve (can be a 1D LUT). The gamut transform is a matrix transform. Both these operations can be collapsed into a LUT. So when Panasonic provides a Vlog to Rec709 3D LUT, what it's really doing is concatenating both those operations into one.

I am not 100% clear on this, but Atomos seem to have established a native gamut for the Sigma recording that is different to Vgamut. But if I recall, you can transform the raw file in FCPX immediately to Vlog/Vgamut in the inspector before doing anything else. After you do that, you can apply the LUT and you won't get the saturation issues, which is actually just a gamut mismatch.

Hopefully the images then match on both the Ninja V and the FCPX timeline.

V-log yes. You choose that one in the info panel, but the gamut needs to be converted from RAW. They are not providing a LUT which does everything in one step. Instead they wrote this manual: https://av.jpn.support.panasonic.com/support/global/cs/dsc/download/lut/s1h_raw_lut/index.html

This is what I have applied and it works a charm. 

Even if you add one further LUT and replace the step v-log to Rec709 with v-log to Arri LogC and afterwards the Arri LUT 709. 

Capable little camera and now a consistent workflow without manual knob turning. 🙂

 

[Llaasseerr]

On 7/7/2022 at 3:35 AM, OleB said:

V-log yes. You choose that one in the info panel, but the gamut needs to be converted from RAW. They are not providing a LUT which does everything in one step. Instead they wrote this manual: https://av.jpn.support.panasonic.com/support/global/cs/dsc/download/lut/s1h_raw_lut/index.html

This is what I have applied and it works a charm. 

Even if you add one further LUT and replace the step v-log to Rec709 with v-log to Arri LogC and afterwards the Arri LUT 709. 

Capable little camera and now a consistent workflow without manual knob turning. 🙂

 

Okay, that's interesting. I looked on youtube and I saw that the FCPX inspector only offers the Vlog conversion whereas with Sony it offers both the log and gamut conversion together. Too bad. I've been using Assimilate Play Pro to convert to ProRes4444 log and luckily they handle the gamut mapping accurately.

That LUT is designed for PRR footage from the BS1H/S1H/S1/S5/BGH1/GH5S/GH6. So the transform will not be accurate from the native raw gamut of the Sigma FP. But if the gamut transform is not known by FCPX,  then maybe this is a "close enough" workaround.

There's already an "intermediate gamut" tag for Vgamut that was inserted by Atomos if you choose Vlog for monitoring on the Ninja V, and it seems Play Pro is able to use that to accurately transform PRR to Vgamut. I'm guessing Atomos worked out the native gamut to Vgamut transform when they included that tag. I downloaded Play Pro for free when I bought the Ninja.