Panasonic S1 4K 10bit video mode to be present at launch with Hybrid LOG Gamma

[Andrew Reid]

Overlooked in the first articles about this camera is the 4K H.265 mode of the Panasonic S1. Similar to the Fuji X-T3, this will allow 10bit internal recording in HEVC with launch firmware version 1.0 - no need to wait for the paid 10bit update.

3840 x 2160 is available in 10bit at 24p, 25p and 30p with no crop - it is full frame.

The later paid update is for a higher bitrate 10bit 4:2:2 mode and V-LOG, but in the meantime 10bit H.265 provides the silky smooth colour of 10bit with HDR and much smaller file sizes.

Read the full article

[currensheldon]

Asked this on another thread, but Andrew I'd be curious to hear your thoughts. 

I am actually curious of which would result in a better image (with everything else being equal).

- 10 bit 422 LongGOP at 150mbps or in this case

- 10 bit 420 LongGOP H265 at 72mbps (S1)

- 8 bit 420 All-I at 480 mbps (EOS R)

Especially if we assume they are all the same camera withe these shooting options. 

Even with a lower bit-depth, my knee jerk reaction would be that 6x or 3x the bit rate would result in a fuller, more gradable image than the 10-bit image. I remember all the hullaballoo around the GH5's 150mbps mode not really being "10 bit" because you have to have a data rate of so much to really get the advantage of the bit-depth (something like 300-400 mbps, if I remember correctly). 

[Simon Young]

I’m mostly concerned about the internal noise reduction, I really hope it’s not as aggressive as the one seen in the a7III at iso 6400 and above, even when dialed all the way down. If the S1 retains more detail  but with slightly more noise in the aforementioned iso:s it’s an instabuy for me.

[Andrew Reid]

1 hour ago, currensheldon said:

Asked this on another thread, but Andrew I'd be curious to hear your thoughts. 

I am actually curious of which would result in a better image (with everything else being equal).

- 10 bit 422 LongGOP at 150mbps or in this case

- 10 bit 420 LongGOP H265 at 72mbps (S1)

- 8 bit 420 All-I at 480 mbps (EOS R)

It depends on more than just the codec specs so it's tricky to answer that especially as I don't yet have an S1 to compare to the EOS R.

It depends also on what aspects of the image you are measuring. I'd expect for instance 10bit to have an advantage in dynamic range but ALL-I would give an advantage in motion cadence. Tricky to say what is "overall" the most cinematic.

[PannySVHS]

Please full VLOG curve and gamut with possibility for NR dialed off and ALL Intra for HD and 4K!

[thebrothersthre3]

So people think motion cadence is noticeably better in all-i on the GH5 vs Long GOP? In theory it is, but maybe not in practice. 

[KnightsFan]

1 hour ago, currensheldon said:

I am actually curious of which would result in a better image (with everything else being equal).

I've done a number of codec tests, some of which I posted on EOSHD. With H.264, you can easily find a situation where an IPB encoding performs as good or better than All-I at 5x the bitrate. Static shots, for example. As you move the camera more and more, the IPB advantage goes away.

1 hour ago, currensheldon said:

I remember all the hullaballoo around the GH5's 150mbps mode not really being "10 bit" because you have to have a data rate of so much to really get the advantage of the bit-depth

I do not buy this. In my tests, I have found that increasing bit depth can increase quality without increasing the bitrate. The more information the encoder has to work with, the more efficiently it can decide what to keep, what to throw out, and what to fudge. The result: you actually need less data to get a better image when using 10 bit. This is true whether the source is 8 or 10 bit. But to answer your real question:

2 hours ago, currensheldon said:

Even with a lower bit-depth, my knee jerk reaction would be that 6x or 3x the bit rate would result in a fuller, more gradable image than the 10-bit image.

I think that, controlling for all other encoder settings, your knee jerk reaction is correct--but in your examples the difference of All-I vs IPB isn't being controlled, and will be the biggest factor depending on the amount of motion in the scene.

 

2 hours ago, Andrew Reid said:

Similar to the Fuji X-T3, this will allow 10bit internal recording in HEVC with launch firmware version 1.0 - no need to wait for the paid 10bit update.

Yeah, I get the impression that the publicity for the 10 bit 422 paid update made some people miss the fact that even without the update, the S1 shoots 10 bit 420 out of the box, internally.

[sanveer]

My favourite dynamic range measuring site(?), claims that it has 12.2 Stops of Dynamic Range in HLG. You could probably get another 0.7-1 Stop with the VLog (non L). 

As well a perfectly usable ISO of upto 30000. Which is huge for both. 

Autofocus is as consistently terrible as always. But I guess you can't be superb in 3 fields. 

IBIS is also very impressive apparently. 

https://***URL not allowed***/panasonic-lumix-s1-review/

 

I wish it was a little smaller and $2000. That way 2 of these wouldn't pinch so much. 

[deezid]

3 hours ago, sanveer said:

My favourite dynamic range measuring site(?), claims that it has 12.2 Stops of Dynamic Range in HLG. You could probably get another 0.7-1 Stop with the VLog (non L). 

As well a perfectly usable ISO of upto 30000. Which is huge for both. 

Autofocus is as consistently terrible as always. But I guess you can't be superb in 3 fields. 

IBIS is also very impressive apparently. 

https://***URL not allowed***/panasonic-lumix-s1-review/

 

I wish it was a little smaller and $2000. That way 2 of these wouldn't pinch so much. 

12.2 stops is actually really impressive. Best camera in its class then.

[deezid]

Well, seems like Panasonic went back to it's temporal noise reduction issues they had with the GH4 lol
The GH5 doesn't show this problem anymore since version 2.0
Other candidates with lots of (temporal) noise reduction artifacts: Sony A73/A7R3, Z Cam E2 (update incoming to fix), X-T3 (not as bad), Canon XC10/XC15

 

[mat33]

I was considering the S1 but its a bit of a strange release.  How many advanced/pro photographers are really going to dump their Canon/Nikon/Sony cameras/lenses for a Panasonic, the big 3 all have pretty compelling hybrid mirrorless offerings now and while the S1 is a mighty fine camera, does it do enough to get people to switch and commit to the new system?

[thebrothersthre3]

3 minutes ago, mat33 said:

I was considering the S1 but its a bit of a strange release.  How many advanced/pro photographers are really going to dump their Canon/Nikon/Sony cameras/lenses for a Panasonic, the big 3 all have pretty compelling hybrid mirrorless offerings now and while the S1 is a mighty fine camera, does it do enough to get people to switch and commit to the new system?

Well the hope is probably once the lenses get off the ground new photographers might start using them.

[Video Hummus]

8 hours ago, Andrew Reid said:

3840 x 2160 is available in 10bit at 24p, 25p and 30p with no crop - it is full frame.

Man, this is a nice camera from Panasonic with a lot of potential. I hope they get a chance to develop it more. Great IBIS, easy handling ergo, great stills and lots of video options. But it’s biggest problem is that it doesn’t have Canon/Nikon on the front of the camera.

The glass is expensive, but so is Canon RF glass. If it has similar quality what’s the problem? Also, it’ll have cheaper Sigma options.

I still think Sony’s new eye detect hybrid af (A9) is going to eat this things lunch though. I also think we haven’t seen the A7Siii precisely because of this camera. They wanted to see how Panasonic priced it and what software they put in it. They obviously knew what the hardware was capable of.

I expect the see the A7siii priced similar or slightly higher than the S1R.

[thebrothersthre3]

The A7S2 should be superior if its a video focused camera. Everything is setup for Sony to knock it out of the ballpark, they've waited everyone out. It needs Sony Venice though and they need to get a handle on overheating.

That said if the S1's start showing up used for $2000 it will be hard to deny. The firmware update is expensive but the built in HLG should suffice, though I'd prefer 10bit H264 as its easier to edit.

[DBounce]

1 minute ago, thebrothersthre3 said:

The A7S2 should be superior if its a video focused camera. Everything is setup for Sony to knock it out of the ballpark, they've waited everyone out. It needs Sony Venice though.

They won’t... they must protect the FS lineup. Fuji and Nikon are the new disrupters.

[mat33]

Its just strange they made a stills focused full frame G9, when they are loved for their video focused GH5. I would think a full frame GH5 with 13-14 stops DR would sell like hot cakes at S1R prices, and bring a lot of L-glass sales.  Do they sell that many EVA1s that they have to protect this, given that none of their glass is being used on their 'pro' bodies and a S1R + 24-105 + 50mm brings them more $$$ than a EVA1?

[androidlad]

2 minutes ago, mat33 said:

Its just strange they made a stills focused full frame G9, when they are loved for their video focused GH5. I would think a full frame GH5 with 13-14 stops DR would sell like hot cakes at S1R prices, and bring a lot of L-glass sales.  Do they sell that many EVA1s that they have to protect this, given that none of their glass is being used on their 'pro' bodies and a S1R + 24-105 + 50mm brings them more $$$ than a EVA1?

To get beyond 12 stops DR the sensor ADC has to be at least 14bit or dual ADC readout. The power consumption and heat management are simply not feasible in a mirrorless stills camera form factor.

[mat33]

I don't know about any of that ADC stuff but it seems to be getting 12.2 stops in HLG, so hopefully full v-log will add some extra DR and its a pretty big mirrorless form factor for heat management...

[Cliff Totten]

Cinema5D is reporting the S1 to test at about 12 stops of dynamic range using HLG gamma. This actually matches the Sony FS7 tests with SLog-3 gamma. It's hard to know if Panny will allow it's VLog upgrade to gain any more than 12 stops. I suspect they might lock it at 12 stops to give the EVA-1 a 1 or 2 stop dynamic range advantage. Who knows?

There is one very depressing thing that Cinema5D found out....NOISE REDUCTION IS HEAVY AND +/- NR CONTROLLS HAVE ZERO AFFECT!!

Ouch!....that one REALLY stings! If Panny has disabled the noise reduction controlls than we have no way to draw it back and save ourselves from heavy in-camera noise reduction.

As we all know,....noise reduction is a 100% DESTRUCTIVE PROCESS....once it's applied, it's artifacts can never be removed in post.

Im hoping their firmware just had a bug that Panny will fix soon.

[androidlad]

15 minutes ago, Cliff Totten said:

Cinema5D is reporting the S1 to test at about 12 stops of dynamic range using HLG gamma. This actually matches the Sony FS7 tests with SLog-3 gamma. It's hard to know if Panny will allow it's VLog upgrade to gain any more than 12 stops. I suspect they might lock it at 12 stops to give the EVA-1 a 1 or 2 stop dynamic range advantage. Who knows?

There is one very depressing thing that Cinema5D found out....NOISE REDUCTION IS HEAVY AND +/- NR CONTROLLS HAVE ZERO AFFECT!!

Ouch!....that one REALLY stings! If Panny has disabled the noise reduction controlls than we have no way to draw it back and save ourselves from heavy in-camera noise reduction.

As we all know,....noise reduction is a 100% DESTRUCTIVE PROCESS....once it's applied, it's artifacts can never be removed in post.

Im hoping their firmware just had a bug that Panny will fix soon.

It's unlikely to go beyond 12 stops in V-log, this is a stills-oriented camera that operate using 12bit ADC in video mode. Heavy NR also tricks the image analysis software for cleaner shadows (more absolute DR in shadow but not very usable).