These of you who often observe up on our lab checks are effectively conscious that we discovered a bug within the first manufacturing firmware of the brand new LUMIX S1II when utilizing 5.8K ProRes RAW with DR Increase “ON”. The merging of the 2 ISO values into one picture was not correctly applied. Now, Panasonic has launched new firmware that primarily addresses this bug. Curious to see the outcomes? Then learn on for our 2nd a part of the Panasonic LUMIX S1II Lab Take a look at…
In our first lab check of the brand new LUMIX S1II (head over right here), we might already see the potential of the brand new implementation of Dynamic Vary Increase mode with the interior compressed codecs and ProRes RAW in APS-C mode. Now, the time has come to judge the full-frame 5.8K ProRes RAW DR Increase mode.
The LUMIX S1II in our CineD Studio. Picture credit score: CineD
The rolling shutter values haven’t modified with the brand new firmware implementation; therefore, we’ve got to say the largest draw back of the full-frame DR Increase “ON” implementation: as a result of two totally different ISO values are joined into one picture, read-out instances are greater than 2x slower than when turning DR Increase “OFF”. We’re speaking a few rolling shutter worth of 27.5ms, which is moderately poor, to be trustworthy.
Nevertheless, it’s a frequent theme with dual-gain output sensors, as we’ve got seen comparable conduct with the Canon C70, as effectively, for instance (lab check right here).
Earlier than we start, a giant “thanks” to my colleague Florian Milz, who was a giant assist in capturing and analyzing this second spherical of testing.
Dynamic Vary at ISO 1000 (DRB “ON”) utilizing 5.8K ProResRAW
As traditional, we used Ultimate Reduce Professional to develop the ProRes RAW information into V-Log earlier than operating them by IMATEST. If you’re not conscious of how we check dynamic vary, please head over right here first.
Taking pictures our Xyla21 chart, we get the next waveform plot:
Waveform Plot for five.8K ProRes RAW DR Increase “ON” at ISO1000. Picture Credit score: CineD
We are able to see a stable 13, if not 14 stops above the noise ground with an extra fifteenth and sixteenth cease contained in the noise ground – very spectacular!!
IMATEST confirms this outcome:
IMATEST outcome for five.8K ProRes RAW. Picture credit score: CineD
We’re getting 12.9 stops at a signal-to-noise ratio (SNR) of two, and 14.2 stops at SNR = 1. Additionally, we are able to see an extra 3 stops above the blue “14.2” line within the center diagram above – these needs to be useful within the latitude check additional under.
These are very spectacular outcomes, particularly for a RAW codec! Additionally, the “Noise spectrum” graph on the decrease right-hand aspect reveals wholesome values of amplitudes even at excessive frequencies, therefore a really detailed image with out a lot inside noise discount.
Publicity Latitude of the LUMIX S1 in 5.8K ProResRAW with DR Increase “ON”
Now, if all goes fantastic, we should always be capable of see the total energy of the 12-bit RAW codec in our latitude check now. As a fast reminder, latitude is the potential of a digital camera to retain particulars and colours when over- or underexposed and pushed again to a base publicity. This check may be very revealing, because it pushes the entire picture pipeline of any digital camera to its absolute limits, not simply within the highlights however primarily within the shadows.
As traditional, the latitude check is carried out in DaVinci Resolve, with the complication that there is no such thing as a assist for ProRes RAW but. Therefore, we’d usually use the RAW Converter app, which might transcode the PRR information to Cinema DNG, a format that can be utilized in DaVinci. Nevertheless, this app didn’t acknowledge the digital camera (but), therefore it didn’t transcode the information. Subsequently, we developed the information in Ultimate Reduce Professional to 12-bit ProRes XQ 4444 information (and in addition adjusted the publicity with the publicity slider after which the ISO), after which imported them into DaVinci Resolve.
Our studio base publicity is (arbitrarily) chosen as having an (ungraded) luma worth of round 60% on the brow of our topic on the waveform monitor – on this case, my colleague Johnnie:
From right here, we are able to overexpose by 5 stops:
The purple channel on Johnnie’s brow is on the cusp of clipping however nonetheless intact.
Now let’s underexpose by stopping down the iris of our ZEISS Compact Prime 85mm T1.5 in one-stop increments till T8, after which doubling the shutter worth. All of the information are then normalized again once more to the bottom publicity ranges (utilizing the publicity slider and altering the ISO setting when growing the information in Ultimate Reduce Professional).
Normally, the ProRes RAW information have a fantastic noise which appears excellent to my eye, and at 3 stops underexposure and pushed again to base, for the primary time, some fantastic noise begins to seem within the picture:
We’re already at 8 stops of publicity latitude (5 over, 3 beneath, and pushed again to base). That is the purpose the place sometimes full-frame shopper cameras begin to break up.
Not the LUMIX S1II! The 4 stops underexposed, pushed again to base picture appears like this:
The fantastic noise, which is now everywhere in the picture, can simply be eliminated in DaVinci Resolve 19:
Fairly amazingly, at this level we’ve got reached 9 stops of publicity latitude and the picture nonetheless appears fantastic! No horizontal or vertical strains, no bigger blotches of chroma noise. Even colours are utterly intact – that’s the facility of 12-bit RAW!
Can we push it to 10 stops? The one full-frame digital camera that was able to doing so was the ARRI Alexa Mini LF (lab check right here) and the Sony BURANO 8K, which achieved near 10 stops (lab check right here) – let’s take a look:
Noise discount yields the next:
Wow – now, eradicating luma and chroma noise in post-production is troublesome with out affecting picture sharpness an excessive amount of, however this picture remains to be usable! The sensor efficiency is certainly on ARRI Alexa Mini LF degree (however there’s an Achilles heel of 27.5ms rolling shutter although, whereas the Alexa Mini LF exhibited 7.4ms)!! Additionally, colours basically are nonetheless intact, though the picture has a slight greenish solid now.
Then let’s transfer additional to 11 stops:
Now, noise is in all places and corrupting the picture. Let’s see what noise discount can do – utilizing now three frames of temporal NR and ULTRA NR in Resolve with greater luma values, which at the moment are affecting the small print of the picture:
This nonetheless appears considerably OK! Nevertheless, three frames of temporal NR are borderline for a transferring picture as ghosting begins to seem. That is once more a case the place the static picture above appears higher than the transferring picture, which additionally reveals bigger blotches of chroma noise hovering round within the picture. Additionally, the shadow aspect of Johnnie’s face is corrupted by chroma noise that can not be eradicated with out destroying all the small print within the picture.
Abstract of the Panasonic S1II LUMIX Lab Take a look at, half 2
The DR Increase “ON” mode with the interior 12-bit 5.8K ProRes RAW implementation in Panasonic’s new LUMIX S1II digital camera has surpassed all my expectations. Not solely does it present excellent dynamic vary values for a RAW mode (right here, sometimes, DR values are decrease as little inside noise discount is occurring within the RAW picture pipeline), nevertheless it reveals huge potential within the latitude check. From our base publicity, you possibly can overexpose 5 stops and underexpose 5 stops, offering a stable 10 stops of publicity latitude with wiggle room in the direction of 11.
For the primary time, we’ve got a shopper full-frame digital camera that’s on par with the ARRI Alexa Mini LF (lab check right here) in our standardized latitude check. The LUMIX S1II reveals a really strong picture pipeline with a fantastic, pleasing noise sample that survives after noise discount when heavy shadow lifting is required. All of that with out giant colour shifts. Therefore, outperforming all different shopper full body cameras, but in addition skilled ones in utterly totally different value brackets just like the URSA Cine LF 12K (lab check right here), or the RED V-Raptor [X] (lab check right here), or the Sony BURANO (lab check right here).
Life is all about compromise; therefore, this comes at a value. This might be the sluggish readout pace, leading to 27.5ms rolling shutter for full-frame mode (with DR Increase “ON” – “OFF” yields a very good 12.7ms) – the excellent in-camera IBIS can, nonetheless, mitigate a lot of the destructive rolling shutter results for lots of capturing eventualities. An excellent compromise might be the APS-C mode for ProRes RAW. The complete sensor potential cropped with a rolling shutter of 18.7ms, which is far simpler to deal with for extra demanding capturing eventualities.
Have you ever shot with the brand new LUMIX S1II? What are your experiences? Tell us within the feedback part under!
