androidlad

3 hours ago, chrisE said:

Where did you get this information from? I'd be really interested in the specs for HLG to get a better understanding of the characteristics and how to grade it properly. Any white papers?

https://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.2100-2-201807-I!!PDF-E.pdf

As HLG is a relative standard just like BT.709, manufacturers can add their own customisation.

It's a minor glare caused by the metallic PDAF layer on top of the sensor, only happens when light hits the sensor in certain angles, mostly in backlit situation.

3 minutes ago, Attila Bakos said:

Where do you get the 22% from? I couldn't find it in the specs.

I also did my tests with the X-T3 when the update arrived. I used ETTR and placed highlights at the exact same level near clipping, then pushed shadows to the same level in post. Overall noise level was identical, and to my surprise I could get a little more detail out from the shadows of HLG even though they appeared more crushed. The difference is very little though, I wouldn't base my choice on it. I've chosen to stay with F-Log, because I learned to work with it in the past and I find it easier now.

From LUTCalc data.

HLG was not designed to be an acquisition format, the curve is extremely non-linear and hard to grade manually.

I think HLG does offer a bit more shadow detail due to its efficient use of all available bits, compared to F-Log's more compressed tonal range (elevated black level). 

I'm really starting to like HLG more, mostly because the new colorimetry, it's more clean and neutral.

9 minutes ago, DanielVranic said:

Not the biggest fan of what that LUT does to the highlights, I messed around a little bit with mine.

 

Arri3.cube

Yours compresses highlight to almost where skintone should be.

On properly exposed F-Log materials, the default F-log to Alexa LUT work really well, you can choose Amira option for a tad more contrast. If you think it does bad things to your highlights you might be overexposed, in that case you can generate LUT sets that compensate for over/underexposure.

1 minute ago, DanielVranic said:

I have never experienced it. I have also not used any LUT's to grade my XT3 stuff, just Resolve curves and 2020-709 color space.

With LUTCalc, what are you using? I tried to make the AlexaX2 from FLog one, and the highlights got NASTY!

The magenta bias shows up in large gradients like sky, even in 10bit 400Mbps. It's completely absent in HLG and the codec/bitrate are exactly the same

FLog_FLogGamut-AlexaX2_AlexaX2.cube

Just now, DanielVranic said:

Xt3 doesn't have the magenta FLog thing

 

It does.

51 minutes ago, mercer said:

Yeah I prefer the F Log as well. It’s way more dramatic looking. While the HLG looks like it could have been shot on a phone or a camcorder. Even the color tonality looks better with the F Log... in my opinion. 

Are there any HLG to F Log to Alexa LUTS floating around? If so, maybe that would have a better effect. 

I like HLG colour.

Yes HLG is less contrasty, but it can be easily added with some simple grading. I find that HLG completely removes the magenta bias that plagues F-log since it was introduced in X-T2. Also the roll off is even smoother. It's very Alexa like.

Went out and shot a quick comparison between X-T3 F-Log and HLG picture profiles.

Camera was set to:

Manual WB 5300K

H.265 LongGOP 400Mbps 3840x2160 25p

Sharpness -4, NR -4, Color -4 (adjustable for HLG only)

Exposure matched based on each profiles base ISO

F-log raw, base ISO 640, f/5.6, 1/1250s

F-Log with "F-log/F-Gamut to AlexaX2" LUT, generated with LUTCalc:

HLG raw, base ISO 1000, f/5.6, 1/2000s

HLG with "BT.2100 HLG to AlexaX2" LUT generated with LUTCalc:

Conclusion:

Fujifilm decided to "rate" HLG profile's base ISO at 1000, while the underlying analog gain is the same as F-log (if you compare F-log and HLG at base ISO without compensating for exposure, the highlight DR is exactly the same).

When you respect the base ISO and compensate for the exposure accordingly, HLG offers around 2/3 stop more DR in the highlights than F-log.

BT.2100 HLG maps 18% grey at 22% IRE, while F-Log maps at 46% IRE. The HLG image looks significantly "darker" than F-Log and requires an aggressive LUT curve to bring it back to BT.709.

Noise level in the shadow area is identical, despite the aggressive push in HLG's LUTed image.

Although both F-Log and HLG were normalised to the same target colour space AlexaX2, there's a visible difference in the tonal range (they were shot less than 5 seconds apart so it has nothing to do with ambient light level.)

Personally, for high contrast scenes I'd definitely use HLG over F-Log. Note that it's vital to get the exposure spot on with HLG in order to take full advantage of the expanded DR.

6 hours ago, liork said:

What color gamut does the Nikon Z shoot in 4K 10 bit? In Sony cameras, you can choose bt2020, but I did not see this option in Nikon menu.

N-log is fixed at BT.2020

http://download.nikonimglib.com/archive3/hDCmK00m9JDI03RPruD74xpoU905/N-Log_Specification_(En)01.pdf

8 hours ago, Alt Shoo said:

They also released a global shutter version which looks pretty promising!

They announced a global shutter version.

1 minute ago, Eric Calabros said:

And all automative sensors do some kind of pixel binning. Use this tool and set binning factor to 4 and see what happens:

https://www.microscopyu.com/tutorials/ccd-signal-to-noise-ratio

Nah. It's full pixel readout, no binning no temporal HDR trick.

Have a read at the IEDM 2018 Sony paper.

There's no temporal element to any of the HDR features in new Sony sensors, it just won't work for high speed imaging.

The 605K e- FWC at base is simply achieved by hexa-gain readout. Period.

IMX490, an 1/1.55" automotive sensor, achieves up to 760Ke- with only 3um photosite.

9 hours ago, thephoenix said:

can we have the "eterna look" on the screen and record flog ?

 

No it's just a neutral BT.709 look with "natural live view" feature, with 2.0 firmware the camera can send Eterna look via HDMI and record F-log internally (or the other way), but there's a bug at the moment that applies a 1.18x crop on all 4K frame rates.

10 minutes ago, jagnje said:

Does this camera do off speed playback? And if so, does it play it out the HDMI?

I forgot to check this while I had it at home. I wonder if you can do what I used to do while I had the fs5, where I would record a super slow motion, play it back at 24 fps and record the playback to the shogun, to get prores on the spot.

 

Yes, only in FHD high speed recording mode. But the HDMI output resolution is fixed at 4K30p, regardless of what frame rate you set in camera. Also the playback mode displays some UI element but disappear as soon as playback starts, it'll need to be manually edited out.

20 minutes ago, Azagey said:

Can anyone let me know if it’s possible to input a LUT in camera for F-log monitoring without needing an external recorder? For example, can I record F-log internally to the SD while simultaneously displaying the Eterna LUT on the back screen? Thanks!

You can't put any LUT into the camera, but there's a feature called "Natural Liveview" that works as a view assist, it displays a BT.709 image on screen but records F-log.

26 minutes ago, Eric Calabros said:

There are pro shooters who can match Canon still raw to Nikon raw, to the level that nobody recognize which is which.

The main reason its a bit difficult is that these companies don't share any data about spectral response of the CFA used on the sensor. 

They don't, but I did.

User interface is much much better on the Portkeys.

5 hours ago, Eric Calabros said:

605000 is too good to be true for this pixel size. Probably its a base-only HDR trick. 

20 minutes ago, wolf33d said:

Yeah this would mean 17.5 stops of DR  at a base iso of 12-15. Too good to be true unfortunately. 

I am dreaming of it though, a 36mpx A7S III with all of this new codec and DR and video modes. Problem is it would kill all Sony cameras including the A7R III. 36mpx is close to 42 and if it has better DR and better everything well... And the A7R III is a year old so hard to believe it will be renewed at the same time. 

Maybe an A7S III mid year and the A7RIV in October 2y after the RIII then. We'll see, hard to tell now 

The base FWC is true, it's achieved through multi-gain readout using 18-parallel ADC, similar to how ALEXA expands DR by using dual-gain architecture.

45 minutes ago, Eric Calabros said:

The only "native" one here is ISO200. 

The correct terminology here is “base".

More details about 36MP IMX435 Exmor RS sensor:

6 native ISO:

ISO50 = Conversion Gain -18dB, EI +12dB, DR 17.4EV
ISO200 = Conversion Gain 0dB, EI +6dB, DR 15.4EV
ISO800 = Conversion Gain +18dB, EI 0dB, DR 13.4EV

ISO12800 = Conversion Gain +36dB, EI +6dB
ISO102400 = Conversion Gain +54dB, EI +6dB
ISO1638.4K = Conversion Gain +72dB, EI +24dB

A7SII sensor IMX235 only has 2 native ISO at ISO100 and ISO3200

24 minutes ago, frontfocus said:

i‘d love to see a video of that, neber have I seen 10 bit hevc encoding lightning fast on pc components

Just exported a 5min sequence with all X-T3 4K 100Mbps HEVC clips to 7Mbps 10bit HEVC using "hardware encoding" option in Premiere. The export took 7min03sec, using Intel Iris Plus 640 iGPU.

4 hours ago, frontfocus said:

editing is fine with new hardware. Any 7th gen Intel cpu and modern gpus work fine. 
But 10 bit HEVC export is not hardware accelerated. Forget about that and think about other ways of delivery. It's just too time consuming. 

Personally I'd go for an iMac if you can find one for a good price or, if you go for a macbook, take the vega graphics. A 13" macbook pro or mac mini with e-GPU might be a great tool too. 
 

no waveforms on the camera. Focus peaking: set the camera to manual focus. In the video menu look for MF assist and set it to focus peak highlight. You can change the color and intensity 

In Premiere CC 2019, with compatible Intel CPU or Nvidia GPU/NVENC, 10bit HEVC decoding and encoding are lightning fast.

Major bug in X-T3 2.00 firmware update: when selecting SD: film simulation, HDMI: F-log, or SD: F-log, HDMI: film simulation. A 1.18x crop is applied on all 4K framerates.

8 hours ago, thebrothersthre3 said:

Gonna get it today. Mainly interested in HLG and if its any better at high ISO's or with DR. 

It's a bit less noisy than F-log (mainly because the shadow/black level are not as high), DR is exactly the same.

Minimum ISO 1000 matches F-log ISO 640, so it does not look like actual analog gain, it's really just an EI rating to encourage slight underexposure.

Brand new PDAF design from Sony Semicon, significantly faster and more accurate AF performance with zero impact on pixel fill factor.

Works reliably up to -5EV, f/11.

5 minutes ago, Andrew Reid said:

I'm confused.

Is just the sensor colour sampling bayer raw, or entire codec is RAW data in that mode?

How can HEVC support RAW video?

HEVC Version 2 supports Bayer RAW encoding.