Noli

Hi all,
since I have had a lot of requests to do so, I have created a YouTube series about the LUTs and my SIGMA fp workflow.
Maybe that is a little more accessible than just a white paper (although there is now a v3 with updated LUTs included as welll)
Here is the link to episode 1. Latest episode is 5. Everything for free of course. 🙂
Maybe you enjoy watching this series. 🙂 
If you have questions, let me know.
Ole
 
 

Hi all,
have attached a chart that sums up my essay about the fps ISO behaviour.
Concentrated only on the native ISO circuits instead of the ones with further analogue gain, which I would not recommend using.
Rating of the native ISO value for the 1st circuit are per my definition in accordance to the 2nd circuit stops distribution.
The chart is valid for RAW (CINE) recordings of the Sigma fp.
If you decide to go through the ARRI Rec709 LUT, than you should consider using the fp in the bluely highlighted ISO range to get the rolloff.

@TomTheDP
 
Take a look at this essay below, used the same settings with your ARRI footage and they are extremely close now. Thanks again for giving access to that footage. 🙂
 
Hi all,
I have some very exciting news for you. After a couple more testing sessions I have figured out how to match my ProRes RAW workflow coming from V-log ending in the ARRI Rec709 LUT as well for cDNG!!
Assume this workflow can also be compiled into a monitoring LUT to be uploaded on 3D compatible monitor. The highlights will still be shown as clipped with more than +3 stops over middle grey. But that could be remedied by securing highlights setting the camera to ISO 100 and afterwards switch back to ISO 800. They will be left intact as the camera is invariant in that range and you can see what is happening in the shadows.
Necessary steps are the following:
1. RAW settings & CST to ARRI LogC

2. CST (non industry standard) bringen the colors into range where they belong

3. ARRI Rec709 Classic LUT

4. Comparison shot of my PRR workflow footage: same same 🙂 

Hi all,
I want to come back on the topic of ISO behavior of the fp. Have tested the camera further over the last weeks and found something interesting.
Think I understand now how the camera works in detail. 
Between ISO 100 and 800 the camera is working in CINE EI mode, with a native ISO 500 following the same rating logic as they have rated ISO 3200 (about 5.42 stops above and 7.04 stops below 18% grey). That means the clipping point is as well the same. Do not understand why they have chosen to declare ISO 100 as the 1st native ISO for video mode. Rather untypical for a cine camera.
When checking the false colors on the Ninja V in native (V-Log) you can see the clipping point jump up and down accordingly while selecting the ISO values mentioned above. 
That means your native selections for ISO should be 500 and 3200 if you do not want to use Cine EI. Nothing else. Just imageing the camera has to be loaded with film stock. You can choose a lower or higher sensitivity. Than compensate too much light with a ND filter, to little by letting in more light. 
ISO values 1000 to 2500 are achieved with analogue gain and will sacrifice image quality. Same applies for all values as from ISO 6400 and above. 
However if you want to use the benefits of Cine EI you can use this knowledge to your advantage as well.
Lower the preview ISO below 1st base ISO 500 you will get more dynamic range shifted to shadows, shift it upwards to ISO 800 you will receive more highlight details (good for rolloff on a bright day). With the 2nd base ISO 3200 you have the option to get some more highlight headroom (to be precise 2/3 of a stop) if you go up to ISO 5000.
Have attached the chart provided by Sigma for visual reference. 
That mixture of using a CINE EI and (analogue gain) ISO is rather special. Some cameras give you the option to use either or, but no camera I know is doing the same as the fp.
Hope this will proof to be a value for some of you to squeeze the best quality out of the fp. 🙂 

I must say I love this little camera. Especially since the LUTs are now workable. Have attached you a couple of pictures, am currently recreating a scene together with my wife and child. Guess some of you will recognize the movie? 🙂 
Lit by candles only. Lens used DZO Vespid 50mm at T2.1 and fp set to ISO 3200. 

Once it is ready I will post a link here as well so that you can judge the image quality. But do not expect great acting, we are both not even bad actors, more like no actors at all. 😄 
 

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.
 
 
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.

 
 
 

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.

 
 
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.
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.
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. 
 
 
 
 

Just a note IMO about linear Raw vs Log encoding. Let's keep in mind that log encoding was developed by Kodak for Cineon scanning of film negative, and is basically what was passed on to digital cameras. It's not an inferior image encoding method developed for hybrid 4k mirrorless cameras. The LogC encoding on the Alexa is probably the most prominent log curve for a digital camera to be very close to Cineon, and Red and Sony eventually capitulated to that.
It's not inferior to linear raw, unless the individual log curve implementation by that camera manufacturer is left wanting. It's a more efficient encoding of highlights and shadows because it gives less code values to highlights, rather than half of a linear image for the brightest stop. It also allocates more code values to the shadows than linear raw. I would rather take 12 bit log ProRes RAW from the Sony cams/Ninja V than 12 bit linear DNGs, but obviously that's a much more expensive solution than the fp - like 5x the price.
The reason that log footage is often inferior in the prosumer cams is because it's probably chroma subsampled, denoised and DCT compressed. ProRes 4444 log from the Alexa Classic is a thing of wonder though.
If you apply the inverse transform of the log encoding curve, then your image is back in linear space and if you factor out the ways the image has been decimated, then it is very close to an original raw recording.
 

You're right, they are defined and that's what allows us to transform in and out of them to other colour spaces. It's not like we have to work in linear floating point in ACES gamut, Rec2020 or Alexa wide gamut, but to me working in ACES is a kind of lingua franca where the mathematics behave in a simple predictable way that is the same as the way exposure works in the real world. And under the hood, the Resolve colour corrections are still applied in a log space (ACEScc or ACEScct). I'm not saying it's perfect, but it makes a lot of sense to me.
The caveat though is that more often than not, an image that is represented as sRGB or Rec709 has had a linear dynamic range compressed into the 0-1 range not just by doing a transform from linear to sRGB or linear to Rec709, because that would cut out a lot of highlights. So there's some form of highlight rolloff - but what did they do? In addition, they probably apply an s-curve - what did they do? Arri publishes their logC to Rec709 transform which is via their K1S1 look, so this is knowable. But if you transform logC to Rec709 with the CST it will do a pure mathematical transform based on the log curve to the Rec709 curve and it will look different. 
So basically, to say an image is sRGB or Rec709 isn't accounting for the secret sauce that the manufacturer is adding to their jpeg output to most pleasingly, in their mind, shove the linear gamma/native wide gamut sensor data into the "most pleasing" sRGB or Rec709 container. Sorry if you knew all that, I'm not trying to be didactic.
Just as a side note, Sigma apparently didn't do this with the OFF profile, which is both helpful and not helpful (see below).
 
I stated before, that the work Adobe did with CinemaDNG and the way a tool like Resolve and a few more specialized command line tools like oiiotool and rawtoaces interpret the DNG metadata, is based around capturing and then  interpreting a linear raw image as a linear-to-light scene referred floating point image that preserves the entirety of the dynamic range and the sensor wide gamut in a demosaiced rgb image. Not that Cinema DNG is perfect, but its aim is noble enough.
What you're describing with shooting a chart and devising a profile is what the DNG metadata tags are meant to contain courtesy of the manufacturer (not Adobe), which is why it's a pragmatic option for an ACES input transform in absence of the more expensive and technically involved idea of creating an IDT based on measured spectral sensitive data. If you look in the DNG spec under Camera Profiles it lays out the tags that are measured and supplied by the manufacturer.
Separate from that is the additional sauce employed in the name of aesthetics in Lightroom, and above I described the process by which the manufacturers add a look to their internal jpegs or baked Rec709 movies. And what does it matter if there's something extra that makes the photos look good? Well I'd rather have a clean imaging pipeline where I can put in knowable transforms so that I can come up with my own workflow, when Sigma have kind of fucked up on that count.
What I will say about sigma's OFF profile that was introduced after feedback, is that as best I can tell, they just put a Rec709 curve/gamut on their DNG image almost as malicious compliance, but they didn't tell us exactly what they did. I mean in this case, they did not do any highlight rolloff, which in some ways is good because it's more transparent how to match it to the DNG images, but also unlike a log curve the highlights are lost.
So the most I've been able to deduce is that by inverting a Rec709 curve into ACES that I get a reasonable match to the linear DNG viewed through ACES, but with clipped highlights. But for viewing and exposing the mid range, it's usable to get a reasonable match - ideally with another monitoring LUT applied on top of it for the final transform closer to what you would see in Resolve.
And I absolutely don't want to say that we all must be using these mid-range cameras like we're working on a big budget cg movie, thus sucking the joy out of it.  But like it or not, a lot of the concepts in previously eye-wateringly expensive and esoteric film colour pipelines have filtered down to affordable cameras, mainly through things like log encoding and wide gamut, as well as software like Resolve. But the requisite knowledge has not been passed down as well, as to how to use these tools in the way they were designed. So it has created a huge online cottage industry out of false assumptions. Referring back to the OG authors and current maintainers pushing the high end space forward can go a long way to personal empowerment as to what you can get out of an affordable camera.

Hi all,
I have invested much time to investigate on the monitoring and ISO behavior of the fp. Finally in combination with the Ninja V I have found a working solution for me which gets both the monitoring picture and the metering right.
Basically after reading through your posts I understood that the fp is ISO invariant and has only two native ISO values available for metering and recording. (ISO 100 & ISO 3200) Others are digitally amplified. 
False color screen with the Ninja V is working on both of these ISO values to meter the correct exposure. However as a lot of you have mentioned the Rec709 monitor view is only showing the correct picture in ISO 100. The 3200 setting will be overexposed and useless.
After several tests I have figured out that if you set the Ninja V to PQ the ISO 3200 screen will look correct. Different story for ISO 100, but this can be tricked if you select ISO 400 in camera. Screen will look exactly the same now for both ISO values. Metering for ISO 100 is unaffected, since the false color mode is only showing the values for ISO 100.
Since the recorded ISO values of the Ninja V are pushed into Final Cut Pro X you will get an overblown picture in Rec709, BUT and this is really nice, if you select ISO 100 instead of ISO 400 and ISO 640 instead of ISO 3200 everything will look as expected. Correctly metered and captured. Remember, in the additional manual of the fp in regards to the ISO behavior, Sigma is mentioning ISO 100 and ISO 640 for photos...
If you need to work with Zebras the correct settings are for ISO 400 => 55% and for ISO 3200 => 60% if you want to avoid anything in the 100% range. You could go slightly higher as the fp has some more headroom over 100 IRE.
Quite a lot of hassle if you ask me, but once you have got used to this you will get a nice picture on both. 
 
Hoping this will help some of you 🙂
 

I just updated to V3.0 and wow, what a great update. Every button on the camera is now customisable.
With my setup above, changing any settings in the quick menu means moving the VA out the way. But now I can have ISO on the front dial, shutter on the left and right buttons, volume on up and down. Then all the buttons along the bottom can be a whole host of over functions.
To top it off, it also gives real 24p UHD cDNG. Jeez.
https://www.sigma-global.com/en/cameras/fp/?tab=support&local=firmware

I saw his first part video when it came out and liked his theories, then the last Jedi came out, I watched it... by the time the second part of his comparisons came out, I liked Yedlin a lot less 😉

I would say it isn't. OFF changes the color matrix in the DNG to sRGB while all other modes retain the camera native RGB color space. The other color modes have no effect of any kind in raw but OFF changes the DNG metadata which then affects the color in Resolve, for example.

Hi, everybody. I'm the author of this modification.
It is currently only available in China, and will be released in November with a self-assembly kit that you can install yourself.
The installation isn't complicated. You just take off the screen and don't have to take the device apart. The whole process takes about 20 minutes.
No parts are damaged and can be restored to factory state.
For $249, it's available through paypal for two to three weeks of shipping and $30 to $40.
I would declare a lower value of the item to reduce the tax.
Contact me at power_cheung@139.com

In the interest of clarification I'd like to see if a few observations are correct. I am talking about DNG forward. It has been my understanding that "base ISO" is determined  by the least noise vs the highest dynamic range. Sigma has proscribed that the base ISO is 100.  I think we have established that the FP uses a linear curve. There are +/_ 12.5 stops of range.  If shot at base ISO then neutral grey should be at the middle of this. The camera's waveform and zebras are correct with this. In my understanding, changing the ISO on camera or in post results in changing the amount of information available  above or below that grey exposing properly-AKA grey is grey. If we choose, or "pretend" that the ISO is 800, we say "I choose grey will be here", whites will still clip at the same amount of light with this much light reaching the lens. We will have less information to convert the above grey-to-blown-out to the +/_ 6 stops of rec 709, but more in the below-grey-to black, which will still go black with the same lack of light amount reaching the sensor.  When changing ISO on camera The FP changes the waveform and clipping point zebras when you change the ISO! As if it is looking at an initial transform at 100, and then you are boosting that image thus indicating clipping where there is none. This makes on the fly protection of highlights and checking of black levels problematic! 
I thought maybe, use 100 ISO to check clipping, set exposure, then switch to what ISO you might choose to monitor and eventually use as the working ISO to expose grey properly. When shooting, then remember, (or make a sticker for the camera!) how many stops above or below your chosen ISO there is to that blown point or black. A spot meter is a pretty good tool to have at this point. Not great for moving fast. Or start to develop your eye. faster, but more accurate than you might think! 
This business of 3200 is a real conundrum though. If I read Rawshooter's generous contributions right, the very metering is wrong, and math with a base ISO check or a spot-meter is the only way through. And all monitoring in camera is going to be off by two stops at proper exposure. Damn. So to move quickly, dial in 3200 + look at recommended exposure, then open up two stops, deal with a blown out monitor image use it as a framing reference. Spot meter the highlights (or gage with your eye haha) to know where the image will blow out.
There is a fundamental question that is kind of pivotal to this and my understanding as a question though.  If I have a camera that has a base ISO of 100, and it has a 5 stop dynamic range (for ease of math), If I choose (shooting raw) to set the ISO at 1600 (one stop of information before blow out), when choosing a rec 709 transform, will the remaining stop of information above middle grey be parsed into the  3 stops rec 709 sees as above grey to white? Here is where I am sure I sound stupid because a stop is a stop, double the light. 
Here's a bit for thought... or telling me to shut up and that's your last glass of wine Mr. I have gotten used to looking at a good monitor and knowing what I see is what will be delivered. It is a pretty seductive thing. I shot film though for 20+ years and I see a lot of parallel with shooting RAW. When shooting film you used a meter and knew the filmstock, what it could handle, how and when it would blow out, how it would handle underexposure (black pixels matter yo! haha) You did not see the printed image for a couple of days (quicker if you were shooting for the studio!)I used a meter and knew the sensitometry of any given film stock, knew what it would matriculate to when graded. Raw seems like this. We look at a proxy image awaiting development. That sounds like knowing the sensor. I got to work with some greats, they could tell an exposure without a meter by eye. We used to play a game when I was a camera assistant on big pictures, and try to guess the shooting stop.  After a while, we nearly always got it. Practical experience of the sensor, no matter whether it's film or digital is key. Knowing the tools is the mark of a cinematographer. I'd like us to have a clear idea of how this camera deals with light, put it in in the back of our minds and then go shoot wicked cool stuff with it. 
 
 

Looks like some people need to learn what’s censorship and what’s standing up against immorality, racism, nationalism and worse.

Hello.  Here is the official answer from Sigma about 0.1 second shutter lag:
 
”We are so sorry that your fp is showing a lag between the moment of release and the actual picture. We are aware of this, and unfortunately, this is how fp works at this point. This is a display lag happening between the sensor and LCD.
 
I will forward your message to our engineer team, and they will check if there is something they can do to fix this. However, at this moment, we have to ask you to accept that this is how fp works.”
So maybe they will come up with something...

Just a cheap chinese DMW-DCC8 coupler.

Sigma also reads this topic for firmware feedback and I think they are doing a great job improving the camera
Lets stop talking about black magic as a company in this sigma topic.
. 

Hey Andrew I'm actually happy with all the technical talk here, because with the input technical users give to sigma,
It is improving things in the firmware, like the flickering and the exposure preview problem.
Please let us bitch a bit more until those issues are all sorted. 😎
And yes lets make separate tread about resolve DNG no problem with that.
 
 
 

Your 640ISO doesn't flicker. That's correct right? I need to redo mine to double check that 320 was doing it all the time but 400 was the trigger factor on my camera.
I think there are ways you can get around purple/green and most of them, like BMD Film, involve crushing the shadows and possibly desaturating them,  so they're not so obvious. @Lars Steenhoff suggested incorrect black levels but it's not that. I suspect ACES is crushing a bit and an IR Cut filter may leech some colour out overall, you could desaturate the shadows in Resolve.
The blotchiness is a factor of 10 bit shadows, each one of those blotches is a single shadow value. In most codecs and compressed formats you would never see this because noise is breaking it all up! But uncompressed shows you what's really there and of course you can add noise to break up the blotches should you want too.
I don't want to keep repeating because i will sound like a broken record but if you take a 12 bit file and a 10 bit file that has the same grey point then what you will see in terms of actual finite values are:
Stop 0 : 1 value
Sopt 1 : 2 values
Stop 2 : 4 values
Stop 3 : 8 values
Stop 4 : 16 values
Okay, if 10 bit and 12 bit have the same 18% exposure brightness level (they do, in the DNG) what the 12 bit is giving you are 2 more stops of shadows. So a value in 10 bit which is at stop 2 is 4 values but that same stop in 12 is 16 values. This is why, when you compare 10 bit to 12 bit (look at my lens images) the shadows in the 12 bit are not blotchy, but a lot more tonality. Now if i pushed 12 bit a further two stops then i would see similar blotchiness but the colours are more accurate (not green/magenta)
I hope that makes sense to anyone still wrapping their head around things....
cheers
Paul
 

Yes, I am on the latest firmware. The notes only mention that they fixed the first frames being enlarged.
I have 3D printed a cheap viewfinder attachment that works with a 3.2" Display Loupe you can find on ebay for 20€. I don't know how long it will last but right now it sits pretty tight. It locks into the small vent holes at the top and you can still use all the buttons.
If anyone is interested, message me and I can send you the blender file.
 
 

The fp has some interesting ISO behaviors, certainly!

I've been using it at lower ISOs even in the dark and then pushing things 2.7 stops without major issues. It's truly a form of night vision. Shooting wide open at F1.4 helps as well, but consider this example of a quick night shot I did:
First shot is at ISO 320 pushed 2.7 stops, second shot is ISO 2500 as-is, third shot is ISO 320 pushed 2.7 stops again, and the last shot is ISO 2500 as-is.

The original video file straight out of Resolve can be found here to avoid seeing the recompression macro blocking artifacts from YouTube:
https://www.youtube.com/redirect?q=https%3A%2F%2Fdrive.google.com%2Fopen%3Fid%3D1sEtAlhfZOjfHO7R7XmXrJJvMUiJqRAHC&event=video_description&v=vYpPZPCWo5g&redir_token=-6uwx0ZKXpomIasN7CDyU0Ps-m98MTU4NDAzNTQzNkAxNTgzOTQ5MDM2

Hey y'all! After a long time playing and testing, I made a flowchart that guides you through the process of buying scopes. This is mainly aimed at beginners, but I'm hoping it can at least be fun for experienced shooters too. I'd love to hear your thoughts!
http://www.tferradans.com/blog/?p=15998
Thanks!