SIGMA FP with ProRes RAW and BRAW !

[Llaasseerr]

10 hours ago, OleB said:

Indeed, I have at first used a vario ND to keep exposure identical and as this wasn't sufficient anymore I stopped down.

From my testings I will use the ISO 800 settings (so real 100), ISO 3200 (640) if I need more light, but I also felt comfortable to go to 6400 or even 12800. Any higher the noise would need more treatment in post than I like to do.

Next round of test files I will upload will be about increasing ISO without adjusting exposure. So you can see when the camera will clip ultimately so nothing can be recovered.

I have found that if I expose with the false colors and set the highlights to be at orange and avoid red, there is about 1 stop which can be recovered.

Your last round of test files was instructive in that I still found the clipping point in each clip, and it was uniform across all clips within a small level of variance. So I'm expecting with exposure differences in a new data set, that the clipping point will also remain uniform. Still, it's always helpful to have another tranche of test clips to play with, so bring it on!

I don't want to push it, but it would be interesting to have similar tests with internal DNG footage if you're up for it.

 

[OleB]

3 minutes ago, Llaasseerr said:

Your last round of test files was instructive in that I still found the clipping point in each clip, and it was uniform across all clips within a small level of variance. So I'm expecting with exposure differences in a new data set, that the clipping point will also remain uniform. Still, it's always helpful to have another tranche of test clips to play with, so bring it on!

I don't want to push it, but it would be interesting to have similar tests with internal DNG footage if you're up for it.

 

Files are uploading just now in the same dropbox. 🙂

The idea of the second test was to show what can be brought back and what is lost. Because with the clipping monitoring of the false colors in Ninja V, this should be already extremely close to real clipping. Which after my experience is the case.

For internal DNG things will get more confusing, like described also in the video you have found on YouTube, there the real clipping points will be all over the place. Have none of this material available currently, but might try this a little later as well.

 

[TomTheDP]

12 hours ago, Ryan Earl said:

This clip is 10 Bit.  It's hard to speculate about what's happening.  It seems like I should be able to dig out a little more.  I haven't encountered that in 12 bit but haven't shot any 10 bit to know if that's more common.

I suspected that. 8 bit and 10 bit have banding in the shadows where that isn't present with 12 bit. I'd use 10 bit only for more controlled situations. 

In the test I did 8 bit and 10 bit looked about the same in terms of banding. It was a rather extreme test however. 

[Llaasseerr]

9 hours ago, OleB said:

Files are uploading just now in the same dropbox. 🙂

The idea of the second test was to show what can be brought back and what is lost. Because with the clipping monitoring of the false colors in Ninja V, this should be already extremely close to real clipping. Which after my experience is the case.

For internal DNG things will get more confusing, like described also in the video you have found on YouTube, there the real clipping points will be all over the place. Have none of this material available currently, but might try this a little later as well.

 

Just tried the old link, but I'm only seeing the old files. Do you have a new link? 🙏

EDIT: my bad, downloading now 😅

[TomTheDP]

Here is the test I did. I can't remember which was 8 and which was 10 bit as they look almost identical. The 12 bit file is very apparent though, leagues better.

This was 100 ISO raised 5 stops in the resolve RAW panel. I used the BM color space.

[Llaasseerr]

1 minute ago, TomTheDP said:

Here is the test I did. I can't remember which was 8 and which was 10 bit as they look almost identical. The 12 bit file is very apparent though, leagues better.

Okay, wow - that's where the green comes from. Thanks for the test!

[Ryan Earl]

On 5/7/2022 at 1:01 AM, radar24 said:

Is this the recommended setting for ACES setting? my resolve 17.4.6 doesn't show "apply ACES reference gamut compress" checkbox though. I've been using P3 D60 Linear Raw tab to CST input to output Arri LogC since i recently bought fp. It seems like ACES workflow is whole another thing that I gotta learn about it.. 

I think i need to get more accurate monitoring. relying on internal LCD is not easy and fun for me.. are Ninja and VA the only options? I will keep using cDNG so I don't need BRAW or ProRes Raw recording abilities. using one of them would be so much better for accurate monitoring? I like the form factor of EVF but the price is at the same level with those monitors and using EVF also means I still rely on the internal LCD and i feel it wouldn't do any improvement on accuracy. having external monitor on fp was not an option I would think of when getting fp.. I thought I would be fine with the internal LCD from my experience with bmpcc og with loupe.    

The checkbox for 'gamut compress' shows up when you switch to the latest ACES version 1.3  It's supposed to help with highly saturated colors / light sources like tail lights I think.

I think transforming to Arri Log-C and applying the ARRI LOG-C to REC709 LUT looks good too BTW.

[Llaasseerr]

16 hours ago, OleB said:

Files are uploading just now in the same dropbox. 🙂

The idea of the second test was to show what can be brought back and what is lost. Because with the clipping monitoring of the false colors in Ninja V, this should be already extremely close to real clipping. Which after my experience is the case.

For internal DNG things will get more confusing, like described also in the video you have found on YouTube, there the real clipping points will be all over the place. Have none of this material available currently, but might try this a little later as well.

 

I have found consistency in DNG clipping points, but only in the narrow range I've been able to test with multiple samples (ISO 100, 125 and 160). I expect that disabling highlight recovery will give a better idea since the results can be variable, and also I expect it would compare better to the Ninja V output since there isn't any PRR highlight recovery available in software right now.

The main thing I've noticed is that the max highlight value in the DNGs scales as a factor of the increase in ISO, at least with 100-125, whereas with the Ninja V it remains constant.

I had a chance to look at your new clips. This time, I reverted to using Assimilate Play Pro since I figured out the mistake I was making, and I think it's more accurate at ingesting PRR than EditReady. Play Pro sets the highlight clipping a bit lower at 9.34 vs 10.35.

As expected, the highlight clipping point stays constant as the exposure and noise floor increase. To check the noise floor, I zero'd out the exposure in all clips. Once zero'ing out the exposure, it was apparent that ISO 3200/640 has a lower clipping point by 2/3 stop.

I'm uploading two contact sheets with all clips with invariant exposure. The second sheet shows the overbright highlights visible in the first few clips when the viewer has the exposure pulled down.

 

So my conclusions right now based on the PRR footage are:

ISO 6400/800 has less noise than ISO 100 +3 stops in post. So depending on how you're shooting, there may be an advantage there.

ISO 3200/640 has 2/3 stop less in the highlights. However it seems to me that if the exposure is bumped up +1 stop in post to match a 6400/800 clip, that the noise floor is the same but the highlights clip 1/4 of a stop higher (11.68 vs 9.34). I'm guessing that this may be because the higher base ISO has a slightly cleaner noise floor - not as dramatic as we have come to expect with other cameras, but maybe something.

 

[Llaasseerr]

7 minutes ago, Llaasseerr said:

Once zero'ing out the exposure, it was apparent that ISO 3200/640 has a lower clipping point by 2/3 stop.

 

 

Correcting something I wrote:

As expected, the highlight clipping point stays constant as the exposure and noise floor increase - except OOC, ISO 3200/640 has a lower clipping point by 2/3 stop.

[webrunner5]

Wow that green stuff is nasty! The footage breaks really hard.

[OleB]

On 5/17/2022 at 12:11 AM, Llaasseerr said:

So my conclusions right now based on the PRR footage are:

ISO 6400/800 has less noise than ISO 100 +3 stops in post. So depending on how you're shooting, there may be an advantage there.

ISO 3200/640 has 2/3 stop less in the highlights. However it seems to me that if the exposure is bumped up +1 stop in post to match a 6400/800 clip, that the noise floor is the same but the highlights clip 1/4 of a stop higher (11.68 vs 9.34). I'm guessing that this may be because the higher base ISO has a slightly cleaner noise floor - not as dramatic as we have come to expect with other cameras, but maybe something.

 

Great! Thank you for this in depth analysis. My last idea in regards to testing would be to take my color checker chart, a good and soft light source plus my Sekonic light meter and expose the camera as per what the meter says for the correct ISO. Maybe that would solve the question were middle grey is for the given ISO values?

In reality to match my shots I usually take the false color screen only and dial in exposure on how I would like the scene to look like afterwards get my talent into the shot and check in which IRE range the skin falls. If I know change angles I leave everything as it is. 

 

[OleB]

Files are uploading.

Have started with a Sekonic measurement of T2.1 on ISO 200 (ISO 1600 in camera). Increases in full ISO steps in camera afterwards, measurements in the light meter as per the table.

However importing in FCPX needs further adjustments. ISO 200 i.e. needs an exposure compensation of +1...

The more I measure, the more it is confusing to be honest. 

[Llaasseerr]

8 hours ago, OleB said:

Files are uploading.

Have started with a Sekonic measurement of T2.1 on ISO 200 (ISO 1600 in camera). Increases in full ISO steps in camera afterwards, measurements in the light meter as per the table.

However importing in FCPX needs further adjustments. ISO 200 i.e. needs an exposure compensation of +1...

The more I measure, the more it is confusing to be honest. 

Thanks, I just took a look at these. In my reading, broadly they are the same exposure except that the ISO 1600 file is ~.25 stop lighter than the others. 

As for the overall exposure, I'm seeing that they are ~2.5 stops overexposed (a little more for 1600). There is an overall blue tint which I'm somewhat taking into account, meaning I'm doing a bit of a neutral grade to get the RGB channels closer to 0.18 in addition to the exposure adjustment.

I should make a note that to adjust the exposure I'm using an Exposure node in Nuke marked in stops, and processing in linear space. This is as opposed to the weird eyeballing using LGG wheels that youtubers and CineD typically do in their latitude tests.

How about you try metering based on the Atomos rating? In that case then theoretically it should be 3 stops overexposed instead of 2.5, but maybe just see how it goes at 3 since that is what they are rating it as with their different ISO numbers compared to internal.

I would be interested to see if there's more of a predictable logic with exposing the internal DNGs.

[kye]

I really hope that once you guys have figured it out you'll post a summary...   you lost me a few pages ago!

[OleB]

15 hours ago, Llaasseerr said:

How about you try metering based on the Atomos rating? In that case then theoretically it should be 3 stops overexposed instead of 2.5, but maybe just see how it goes at 3 since that is what they are rating it as with their different ISO numbers compared to internal.

What do you mean with the Atomos rating? I am metering to match the false colors. however that is hard to do. 
 

i.e. the white patch of the color checker has to be between 90-100 IRE. Middle grey 40-50 IRE and black at max 10 IRE. 

Did another round and tried as follows. Set the min ISO where the false colors are working so ISO 800 and opened up the lens completely. Tuned the light so that the white patch in the false color screen was just below orange so 93 IRE. 
To get an exact match for the t stop with the light meter I needed it to be on ISO 125. Now I left the light setting identical and increased ISO and compensated this by closing the aperture. In the light meter I increased the ISO setting until I got the same aperture reading. For 1600 that was ISO 400. 

So far so good but bringing these files in FCPX I could only bring white and middle gray into the correct range by pulling down the highlights. If I bring down exposure overall all middle grey is off. 

Have created presets now for every full ISO step which gives quite good results however I totally understand that people will be confused by this procedure. So am I…

Do not understand what is so difficult in the firmware to match camera and meter ISO values and the info which is send to the Ninja V exposure metering tools…

[Llaasseerr]

5 hours ago, OleB said:

What do you mean with the Atomos rating? I am metering to match the false colors. however that is hard to do. 
 

i.e. the white patch of the color checker has to be between 90-100 IRE. Middle grey 40-50 IRE and black at max 10 IRE. 

Did another round and tried as follows. Set the min ISO where the false colors are working so ISO 800 and opened up the lens completely. Tuned the light so that the white patch in the false color screen was just below orange so 93 IRE. 
To get an exact match for the t stop with the light meter I needed it to be on ISO 125. Now I left the light setting identical and increased ISO and compensated this by closing the aperture. In the light meter I increased the ISO setting until I got the same aperture reading. For 1600 that was ISO 400. 

So far so good but bringing these files in FCPX I could only bring white and middle gray into the correct range by pulling down the highlights. If I bring down exposure overall all middle grey is off. 

Have created presets now for every full ISO step which gives quite good results however I totally understand that people will be confused by this procedure. So am I…

Do not understand what is so difficult in the firmware to match camera and meter ISO values and the info which is send to the Ninja V exposure metering tools…

Before I say anything else, I want to clarify that when I mentioned a blue tint in my last post I did not mean from the camera - just that there was blue light in the scene that made the colorchecker skew towards blue, which is fine.

What I was referring to as metering for the "Atomos ISO rating" is that you said you metered based on ISO 200 instead of 1600, for example. Given the amount of overexposure, I suggest trying to meter for the Atomos ISO value.

if you just forget about false colour, what works for me with other cameras is to just take a meter reading and adjust the exposure based on that (with ND or stop down lens). Then I check in linear space in Nuke and it should be at around 0.18 in each channel of RGB (depending on scene light tint).

Without Nuke, what I suggest is encode the linear image to a known log curve where you know the code value or IRE for middle grey, then check the raw log image to see that it's a close match. So the log encoding would need to be documented, which for example some Blackmagic ones are not.

It seems the false colour "north star" would be to establish if the recent v4 internal version works predictably, if that actually measures the scene radiance as captured by the sensor. The grey card should hit green on the false colour as a sanity check for you while shooting.

Regarding false colour on the Atomos and in general, I'll admit I have not been tracking exactly what you've been doing as it sounded a bit complicated. This may just be my preferred workflow and others might not like it, but I generally ignore manually reading IRE values.

If I was going to consider using the Atomos false colour at all, I would instead aim to know that the false colour display is designed specifically for the image that you're monitoring. But again, there are many variables here since we don't know what the encoding of the OFF image is that is being sent by the Sigma to the Atomos, although I speculate it's the middle range of the linear scene capture that is then transformed to Rec709. But then you mention you're monitoring in PQ, so I don't know if the false colour is accounting for that transform, either. It seems that there are several undefined variables at play due to the undefined transforms.

You might also find it helpful to see how the Falsify plugin in Resolve interprets the image:

https://xtremestuff.net/falsify-false-color-transform-released-for-davinci-resolve/

Without having the camera in front of me, I would guess that the OFF output displayed is Rec709, so maybe first take PQ out of the equation and see if it works for you.

Also just a note that it seems to me that for a ProRes RAW recording from the Ninja V where the camera manufacturer does not have a defined log curve/gamut, that Atomos adds Vlog/V-gamut metadata tags. But I don't think it's monitoring in that space on the unit. I think it just means that by default, when you're in some software later that it will read those tags and transform the linear/native sensor gamut to Vlog/V-gamut. However I've been overriding this and exporting as Alexa LogC/AWG. Overall, it's pretty confusing.

 

 

 

[Llaasseerr]

If anyone can upload some 12-bit DNGs shot through the ISO range with shadows and something bright enough to clip the sensor, that would be great!

I think the following: 100, 200, 400, 640 (if that's possible), 800, 1600, 3200, 6400, 12800. 

Bonus for a grey card, but not essential.

@OleB both your lightbulb tests were great, but maybe what would be easiest for a DNG test is to replicate the second test with the exposure increase via the ISO change, and without the compensation of stopping down the lens. It then cuts out one more variable.

[OleB]

Ok so I went back to the drawing board.

If you monitor regular ProRes 4:2:2 8 Bit (non RAW) with the Ninja V, the ISO values and meter tools seem to work absolutely correct. Clipping occurs at 100 IRE and false colors. 

The change is happening with the switch to external RAW. If you monitor "native" the clipping points are as follows (approximately, have noted them down from the Ninja V waveform). Picture is very flat and log like. Black point for all ISO is at approx. 10 IRE.

ISO 100 - 65 IRE

ISO 200 - 75 IRE

ISO 400 - 80 IRE

ISO 800 - 90 IRE

ISO 1600 - 90 IRE

ISO 3200 - 85 IRE

ISO 6400 - 90 IRE

ISO 12800 - 90 IRE

ISO25600 - 90 IRE

False colors will show the ranges accordingly to those max values, but not scaled. So ISO 100 leaves you with white being in grey (59-77 range).

Rec709 mode and PQ modes meter as from ISO 800 and upwards as like I have described earlier.

Have read a little more about ProRes RAW. What I found out is that RAW is always linear HDR, no matter how you monitor it. So basically when you bring it in a Rec709 timeline you have to bring back the overshooting dynamic range. At least FCPX is not applying any conversation, it is importing it as linear RAW. You can however transform the very same into LOG (different manufacturer options) and then from that LOG to a certain LUT you like to use.

Maybe that is the explanation why it is best to monitor that linear RAW best in an HDR mode, as this will show you as much information as the monitor can show will reviewing. 

What I do not get however is why the ISO range is acting strangely in RAW mode when monitored PQ. That could be a bug no body liked to address at Sigma/Atomos, or it is because they had to do it like this for their internal RAW recording. I do not know. 

 

[Ryan Earl]

3 hours ago, Llaasseerr said:

If anyone can upload some 12-bit DNGs shot through the ISO range with shadows and something bright enough to clip the sensor, that would be great!

I think the following: 100, 200, 400, 640 (if that's possible), 800, 1600, 3200, 6400, 12800.

Are you asking for the exposure to adjust with the ISO change?  Increase ND for example?  I can add some DNG clips, I wanted to check this out for myself, but don't have a Ninja V to compare DNG To ProRes RAW. 

[Llaasseerr]

1 hour ago, Ryan Earl said:

Are you asking for the exposure to adjust with the ISO change?  Increase ND for example?  I can add some DNG clips, I wanted to check this out for myself, but don't have a Ninja V to compare DNG To ProRes RAW. 

I'm thinking of making this as simple as possible, just an ISO change. Just so that there's a bright enough source that will clip the sensor at all those ISO settings, and also that there's enough shadow detail to challenge the noise level a bit.