Masters of darkness - Panasonic GH5S and Sony A7R III in extreme low light

[Andrew Reid]

Note 1: Download the video at Vimeo for pixel peeping, to bypass the streaming compression!

Note 2: My camera is running firmware version 0.2! If the performance changes when updated, I will re-do the test

Does the Panasonic GH5S finally achieve full-frame like performance at ISO 12,800 against the best of the competition?

Here I put the little nocturnal fox in the ring with the Sony A7R III, A7S II, Nikon D850 and the plain old 'non-video-orientated' GH5.

Read the full article

[Davide Roveri]

Very interesting comparison, thanks Andrew!! :-) 

The GH5s looks pretty good and the wider field of view is definitely welcome but what I think is really spectacular is the a7rIII! I wasn't expecting it to be so much better than my beloved rII and the fact that the full frame mode is as clean as the Super35 one is mind blowing! Couple that with an Atomos so you can record at 880Mbit/s 4:2:2 and it should be spectacular! The only thing missing is still 10bit and that's a pity but I want to make some proper test sometime to see exactly how much is an issue on a regular production...

What I'm thinking now is that Sony needs to make the a7SIII very spectacular spec-wise to be relevant otherwise I don't see why one would choose that over the rIII which is also a phenomenal still camera (unless you're going for the stupidly high iso scenario as you rightly said). Hopefully this will be the year in which we have an a7 camera with 10bit 4:2:2 internal recording (and IBIS), fingers crossed!! :-) 

 

[Andrew Reid]

Sony's new sensors have a very impressive pixel design when it comes to low light, the A9 is another one - was blown away how good it looked at high ISOs in 4K, considering it has none of the compromises of a 10 or 12 megapixel camera on the stills side. I actually would not mind that 24MP sensor in an A7S 3. It no longer needs to be just 12MP!

The pixel architecture, low noise readout, aluminium process instead of copper, less heat, oversampling, downscaling, BSI, all helps.

Sony also designed the sensor in the GH5S.

I hope Panasonic's Fuji partnership for an organic sensor bears fruit, as Sony definitely need better competition.

Canon 5D Mark IV is shit in low light for instance.

[Davide Roveri]

I couldn't agree more!! An a7SIII with 24MP would be a phenomenal hybrid shooter and as you said with the new tech I think they would be able to maintain pretty much all of the high ISO goodness of the first two models (which, if I'm not mistaken, still use "old" pre-BSI tech sensors).

NAB is not far away, let's see what Sony has in store... :-) 

 

 

 

[anonim]

Thanks for nice comparison and opportunity to download original futage.

I did very quick, and not at all processing power demanded, noise reduction node in Resolve 14 in the last ISO 6400 x400 GH5 example. Result is quite similar to GH5s's out of camera.

 

[ntblowz]

Thanks for the test Andrew!  I saw on youtube from this guy that A7SII still have lowlight adventage over A7RIII though, but yeah A7RIII's low light is much improved vs RII

 

I do like the wider view on the GH5s, I wonder how much my laowa 7.5mm turns into? 6.5mm?

This is a boring interior shot of the house to show FOV difference

 

[DBounce]

3 hours ago, Andrew Reid said:

Sony also designed the sensor in the GH5S.

I hope Panasonic's Fuji partnership for an organic sensor bears fruit, as Sony definitely need better competition.

Canon 5D Mark IV is shit in low light for instance.

Where did you hear this? According to Panasonic the sensor in the GH5s is a Panasonic design. Sony had nothing to do with it.

 

[Andrew Reid]

The spec of the sensor matches exactly the Sony Starvis chip. Not hard to figure out either that a marketing guy or sales live chat would want to pretend otherwise!

[Robert Collins]

The footage shows the 'A7rii' or the 'A7sii'? It is marked 'A7rii' in the video (apart from the Vimeo headline that mentions A7sii). But the commentary talks about A7sii v A7riii and the others.

[horshack]

The Sony A7R III has a 42MP sensor vs just 10MP for the GH5S which makes the similar low light performance in 4K video quite mind-bending. How did Sony manage it? 

Temporal Noise reduction most likely. And that would be from Panasonic's ASIC processing of the video rather than a property of the sensor itself.

[Andrew Reid]

1 hour ago, Robert Collins said:

The footage shows the 'A7rii' or the 'A7sii'? It is marked 'A7rii' in the video (apart from the Vimeo headline that mentions A7sii). But the commentary talks about A7sii v A7riii and the others.

Caption error in the video, should be A7S II

[Emanuel]

7 hours ago, Davide Roveri said:

(...)

Couple that with an Atomos so you can record at 880Mbit/s 4:2:2 and it should be spectacular! The only thing missing is still 10bit and that's a pity but I want to make some proper test sometime to see exactly how much is an issue on a regular production...

Take a look on this: 

sample extracted from (minute 28:30):
https://vimeo.com/114978513

Bit rate is more crucial than anything else...

[Robert Collins]

7 hours ago, Davide Roveri said:

Couple that with an Atomos so you can record at 880Mbit/s 4:2:2 and it should be spectacular! The only thing missing is still 10bit and that's a pity but I want to make some proper test sometime to see exactly how much is an issue on a regular production...

I could be wrong but I think the maximum external recording for the A7riii is 147Mbit (compared to 100Mbit internal.)

[Emanuel]

12 minutes ago, Robert Collins said:

I could be wrong but I think the maximum external recording for the A7riii is 147Mbit (compared to 100Mbit internal.)

A pity, if so... a7RIII performance in this comparison test is simply a good revelation. Impressive. Seems this still stands though:

 

[Robert Collins]

BTW I thought I would throw in the A7riii and A7rii comparison here because the difference is so striking (in FF video). I am not sure I see or understand where Sony found 3 extra stops of lowlight performance out of the same sensor,

 

[Trek of Joy]

Great test Andrew, these videos really help with buying decisions. The more I see the harder its getting to resist the a7r3 - and I say that as a a7r2/s2 owner. Hopefully we see the a7mk3 and a7s3 soon, I could live with 24mp. Thanks.

Chris

[Davide Roveri]

42 minutes ago, Robert Collins said:

I could be wrong but I think the maximum external recording for the A7riii is 147Mbit (compared to 100Mbit internal.)

Robert I don't think that is the case, as per Sony specs the HDMI output of the camera is uncompressed therefore the final bitrate is dictated by the encoding you use (the maximum you can get on the Atomos is Pro Res HQ which at 30p in QFHD has a data rate of 882Mbit/s plus you get the benefit of the 4:2:2 colour sampling)

58 minutes ago, Emanuel said:

Take a look on this: 

sample extracted from (minute 28:30):
https://vimeo.com/114978513

Bit rate is more crucial than anything else...

This is very interesting, thanks Emanuel! :-)

[jonpais]

Back to the sensor controversy for just a moment: if pressed, do you think for an instant that Panasonic would admit that their Leica Summilux 12mm f/1.4 was designed by Sigma? ?

[Robert Collins]

16 minutes ago, Davide Roveri said:

Robert I don't think that is the case, as per Sony specs the HDMI output of the camera is uncompressed therefore the final bitrate is dictated by the encoding you use (the maximum you can get on the Atomos is Pro Res HQ which at 30p in QFHD has a data rate of 882Mbit/s plus you get the benefit of the 4:2:2 colour sampling)

You could well be right. I wasnt speaking from experience but from what I had read in Gary Friedman's 708 page epic on the A7riii. 'The maximum bitrate you can choose for SD-stored 4K is 100 Mbps, whereas for external recorders you can record your video at about 147 Mbps.'

He does occasionally get things wrong.

[Emanuel]

53 minutes ago, Davide Roveri said:

Robert I don't think that is the case, as per Sony specs the HDMI output of the camera is uncompressed therefore the final bitrate is dictated by the encoding you use (the maximum you can get on the Atomos is Pro Res HQ which at 30p in QFHD has a data rate of 882Mbit/s plus you get the benefit of the 4:2:2 colour sampling)

This is very interesting, thanks Emanuel! :-)

I've posted this test in a few threads for about two years and a half now since these test results have popped up. You can also find a good discussion here: