Slow-mo shootout - which camera gives the most detail at 120fps?

[Andrew Reid]

The loss of detail is due the pixel binning method used on the sensor to increase the readout speed. It's not due to bitrate. 80Mbit is a LOT for H.265 1080p. That is equivalent to 160Mbit 1080p on the GH4!

[Shield3]

On 8/9/2017 at 3:27 PM, Deadcode said:

Thank you for the test, very informative.

Did any of you noticed that the 1Dx Mark II's motion "candence" seem like it's much more fluid than the others? It's like 30p.

That's because it is 30p; a 1dxII conforms to 30p in camera when shooting 1080p120 (4x slowmo).  No you cannot change it to 24p either.

 

 

 

[Mark Romero]

Because I don't always trust my eyes, I hope you can all help me out here confirming either way:

- The a6500 looks to be the softest (unsurprisingly), right???

- The RX10 III looks to have the most noise, right????

-The GH5 and the Leica seem to be the sharpest???

- the a7S II colors are very... ummm... "Sony Green"???

Is this what other people are seeing too???

[SMGJohn]

I wish you did a test with Samsung NX1 using bitrate hack seeing how it does improve 120fps 1080p footage substantially. 

[aestheticprofessional]

I would love to see where the A9 falls into this group. Full frame without the crop - hopefully similar to the 1Dx or Leica. Anyone have any thoughts?

[markr041]

Sony FS5/FS700 in burst mode 12bit RAW->ProRes 422 HQ 10bit 4K DCI 120 fps no crop gives best quality, but you need an external recorder. Burst mode limits mostly not an issue if interested in slow motion, since that translates to 16 seconds or so of slow mo (enough to get to the boring threshold). This is a constraint. But this is about quality. 

This test video  has both 4K 60P and 120P shots (slowed down), to compare quality:

 

[Amro Othman]

On 8/13/2017 at 7:58 PM, Mark Romero said:

Because I don't always trust my eyes, I hope you can all help me out here confirming either way:

- The a6500 looks to be the softest (unsurprisingly), right???

- The RX10 III looks to have the most noise, right????

-The GH5 and the Leica seem to be the sharpest???

- the a7S II colors are very... ummm... "Sony Green"???

Is this what other people are seeing too???

 

I think the first shot with the A6500 was a bit out of focus- because the second shot at higher ISO seemed much sharper to me.

[markr041]

Slow Motion 4K 120P Experiment Falls Flat

Wanted to shoot preparing a bubbly soda in 4K 120p slowed down and up close. Unfortunately, the soda was flat, and so is this video.

End trigger worked well, though - nailing the exact action. And skin tone and detail are excellent.

[kidzrevil]

On 8/13/2017 at 11:59 AM, SMGJohn said:

I wish you did a test with Samsung NX1 using bitrate hack seeing how it does improve 120fps 1080p footage substantially. 

I think he only tested cameras they actually still produce 

[Kisaha]

@kidzrevil 

NX1 has been tested, above a couple of cameras, certainly above a6500 which is dead last.

And I quote

"Canon 1D X Mark II
7/10 – Quite soft with aliasing issues – could be better but nice dynamic range

Samsung NX1
7/10 – High detail but high aliasing. I do like it though.

Sony A7S II
6/10 – Soft with a crop, not exactly a stella performance

Sony A6500
5/10 – The softest yet also the one with the most aliasing, so clearly the worst of the bunch"

The question of @SMGJohn was about moded NX1, that as we know can increase bitrate to at least 160Mbps stable, on H265 (since 2014, as Andrew mentions in the same article). Normal PRO bitrate is 80Mbps.

[kidzrevil]

@Kisaha oh gotcha, I figured he didn't have a hacked one on hand. He seems to have been avoiding installing the hack on his hardware

[jcs]

On 8/13/2017 at 3:29 AM, Shield3 said:

That's because it is 30p; a 1dxII conforms to 30p in camera when shooting 1080p120 (4x slowmo).  No you cannot change it to 24p either.

 

 

 

You can get to 24p from 30p by retiming to 80% (24/30 = .8) in Premiere etc. for 120/24 = 5x slomo.

We recently shot 1DX II 120fps (PDAF worked well). Without pixel peeping didn't see any resolution reduction vs. 60p (full frame, no crop for both cases). Provides enough resolution/detail for close up and medium shots. For wides best to use 4K60p if detail is important for the shot.

[omega1978]

 

Slowmotion on NX1 with bitrate hack 140mb.

Not bad, a bit of aliasing

[markr041]

9 hours ago, omega1978 said:

 

Slowmotion on NX1 with bitrate hack 140mb.

Not bad, a bit of aliasing

This is a pretty picture, but as a "test" this is the worst kind of example - most of the frame is OOC (bokeh is pretty nice) and there is hardly any detail to anything. Let's see people moving around in a complex setting. If pouring water (soda) into glasses is the main use of slo-mo, then this is right on...

[Mattias Burling]

On 2017-10-03 at 3:26 AM, markr041 said:

Slow Motion 4K 120P Experiment Falls Flat

Wanted to shoot preparing a bubbly soda in 4K 120p slowed down and up close. Unfortunately, the soda was flat, and so is this video.

End trigger worked well, though - nailing the exact action. And skin tone and detail are excellent.

 

27 minutes ago, markr041 said:

This is a pretty picture, but as a "test" this is the worst kind of example - most of the frame is OOC (bokeh is pretty nice) and there is hardly any detail to anything. Let's see people moving around in a complex setting. If pouring water (soda) into glasses is the main use of slo-mo, then this is right on...

 

[markr041]

4 minutes ago, Mattias Burling said:

I guess you missed this one:

My criticism extended to my soda pouring video (did I not mention soda?), which is a parody of silly glass pouring videos, like the one I criticized, get it?

[markr041]

Camera RAW 12bit 4K DCI->Shogun ProRes 422 HQ 10bit 4K DCI. Sony 55-300mm Alpha lens.

[Jonathan Warner]

I have a question about the mention of the Sony data rates.  You point out the normal 120p HD recordings are at 100mbps, whereas S&Q 120p is at 60mbps, and you say therefore the normal recordings are better because they are at the higher bitrate.  However, because the S&Q conforms the recording to 24fps in-camera, doesn't that mean that the 60mbps is on the conformed 24fps (which works out at 2.5mb per frame, as opposed to 1.2mb per frame for the normal recording) , or is the compression performed before the conforming?  I'd be grateful if you could confirm.  Thanks!

[hmcindie]

You are correct Jonathan, it is indeed displayed as the "bitrate after conforming" so it's actually measured for the 24fps conform.

[Orangenz]

The article says RX100V, the video title is RX100IV, and there doesn't seem to be a clip of either in the video?