GH5 GH5s Has More Dynamic Range Than GH5 in All Profiles...

TheDingo

Veteran
Check out this new GH5s review video from The Camera Store TV YouTube channel. At around 6:35 Jordan talks about the extra dynamic range that can be seen in GH5s footage. What's also interesting is that the HLG profile records more dynamic range than the V-Log L profile, as it appears that the V-Log L profile is chopping off a little bit of the dynamic range.

 
Check out this new GH5s review video from The Camera Store TV YouTube channel. At around 6:35 Jordan talks about the extra dynamic range that can be seen in GH5s footage. What's also interesting is that the HLG profile records more dynamic range than the V-Log L profile, as it appears that the V-Log L profile is chopping off a little bit of the dynamic range.


This makes sense based on what we know about v-log L. We suspected that HLG might be able to achieve a little bit more dynamic range. If the 4k Pocket does in fact use the same sensor and they claim 13 stops it makes sense that the GH5S should also be able to pull off that 13 stops unless BMD is greatly exaggerating. Based on this video it sounds like the sensor may in fact be higher than 12 stops.

It really is sad that the GH5S didn't get the full V-log but at 13 stops it might mean full V-log really requires at least 14 stops and so Panasonic had to use V-log L. Just guessing here.

Thanks for sharing the video.
 
I measured (conservatively) the DR of GH5s HLG to be 11.67 stops, which correlates to the 11.5 stops mentioned in the Camera Store video. But I should point out I measured 12.67 *levels*. In other words, the first level is the base level 0, and I didn't count that as a stop, because the semantics of a stop is a doubling of light, which doesn't happen until you've taken your first step from the base level. I think some are counting the base level as a stop, in which case if you do, 12.67 rounded up to 13 equals the claim made for the BMPCC which is almost certainly using the same Sony Starvis 4/3 sensor. I have not measured DR for V-Log/L, but my expectation is that HLG would be the greater DR between the two.

The base level I defined as the point where I could just barely make out the white patch against a black background, just above the noise ceiling on the WFM. I was not using absolute black as the base level, there had to be a visible image.
 
Last edited:
The base level I defined as the point where I could just barely make out the white patch against a black background, just above the noise ceiling on the WFM. I was not using absolute black as the base level, there had to be a visible image.
The base level for dynamic range should actually be the point of full saturation of the sensor, as that's the only absolute fixed reference level on the db scale. DR then extends downward from full saturation, light intensity halving with each stop, until signal level equals noise floor. Of course, that's the opposite of Panasonic and Sony's DR marketing charts, which are manipulated to make it appear that increased DR = more highlight detail. But aside from V-log exposure calibration, the highlights are always there, it's the shadow detail that opens up as DR is increased.
 
The base level for dynamic range should actually be the point of full saturation of the sensor, as that's the only absolute fixed reference level on the db scale. DR then extends downward from full saturation, light intensity halving with each stop, until signal level equals noise floor. Of course, that's the opposite of Panasonic and Sony's DR marketing charts, which are manipulated to make it appear that increased DR = more highlight detail. But aside from V-log exposure calibration, the highlights are always there, it's the shadow detail that opens up as DR is increased.

Only count to the last stop *before* it clips. That's the most conservative number and always a consistent baseline from the bottom; the image has to be visible and it has to be above the noise ceiling. It removes that subjective element of "how much noise can you tolerate for the measurement." It's an arbitrary point, but consistently applied. Either way counting up or down should give the same number if the practice is consistent. In the case of counting up and measuring the result 11.67 (using 1/3 stop increments) and because I stop counting at the last stop before it clips, it means the actual DR could have been something more than 11.67 but for sure less than 12. Totally agree with your point that it's the shadow detail where higher DR actually opens up.

Counting up from the baseline this way is a consistent method to obtain accuracy to -0, +1/3 stop.
 
And because some will count the levels as stops, then 12.67 levels rounded up is 13 stops. Unless they are tested together using the same method, a number like that just can't be reliably compared.

Analogy: 100 watts per channel into 8 ohms @ 1khz with 10% THD is a lot less conservative than saying 15 watts per channel into 8 ohms 20-20 kHz both channels driven with 0.05% THD, but could reflect the same actual power capability. We just don't have standards like that for ferreting out DR claims.

I'm curious what number I will get using the same method to test my Sony PMW-F55 which has a claimed 14 stops of DR.
 
I think the BM Pocket 4k will come out ahead as you can shoot in raw. I can definately pull more from RAW stills then I can from my video on the GH5.
 
This. There's no video mode on the GH5(s) that retains all of the DR of the sensor. (Interesting - and disappointing) that HLG retains more than V-LOG L. Panasonic really s*** the bed with V-LOG L. An all around disappointment since introduced on the GH4. Their stubborn refusal to make a profile that unleashes the GH-line frustrates me to no end. In a world of LUTs, who gives a crap if a curve is "compatible" with a Varicam?

Also, don't forget that there are techniques for squeezing more DR/less noise out of the same sensor. Heat management is one, and the BMPCC 4K has a fan. Black Magic's DR claims have always been in the realistic ballpark in the past. Given DaVinci's NR capabilities, sometimes even conservatively rated.
 
This. There's no video mode on the GH5(s) that retains all of the DR of the sensor. (Interesting - and disappointing) that HLG retains more than V-LOG L. Panasonic really s*** the bed with V-LOG L. An all around disappointment since introduced on the GH4. Their stubborn refusal to make a profile that unleashes the GH-line frustrates me to no end. In a world of LUTs, who gives a crap if a curve is "compatible" with a Varicam?

Also, don't forget that there are techniques for squeezing more DR/less noise out of the same sensor. Heat management is one, and the BMPCC 4K has a fan. Black Magic's DR claims have always been in the realistic ballpark in the past. Given DaVinci's NR capabilities, sometimes even conservatively rated.

Haha
The best thing is, every V-Log L alteration since the GH4 has a completely different color science (Gamut), lol
There's nothing compatible about it. They're just handicapping it (as well as forcing strong processing on their prosumer cams), to differentiate from their higher end tiers like the EVA-1 with real V-Gamut and V-Log and an organic looking image. The GH5s sensor is capable of delivering up to 14 stops of dynamic range!
 
Only count to the last stop *before* it clips.
The point where the sensor saturates is "stop 0", all stops below that are numbered negatively. With a straight-line LOG proflie, there is no such concept as "the last stop before it clips". Saturation is the absolute 0db reference point, stop -1 is 50% as much light intensity, and so on.
 
The point where the sensor saturates is "stop 0", all stops below that are numbered negatively. With a straight-line LOG proflie, there is no such concept as "the last stop before it clips". Saturation is the absolute 0db reference point, stop -1 is 50% as much light intensity, and so on.

Lpowell, I got the same number as the Camera Store reported. What are you trying to say, that I didn't?
 
Back
Top