Panasonic Lumix DC-GH5S added to studio scene comparison

Because in iR are noise reductened jpg's. Look how blurry the details on gh5s jpg's because of the agressive NR

@Butler I saw someone else on eoshd speculated GH5s improvement is from "temporal NR". I disagree. If that's the case, the same NR technique can be applied to all other (recent) Pany cameras and get similar results. The whole point of using the expensive Dual Native ISO technology is that the noise level at ISO2500 is similar to what's at ISO400, and at ISO 25600 (4.5 stops above 2500) similar to ISO 8000. That should be available in still mode too, unless the Pany chooses not to implement it. Or the DPR tests were done in high ISO w/o selecting the High ISO mode. That's something you can help to confirm. TIA.

Per my other reply to your other comment, you don't get a choice of which sensor gain mode is used for each ISO. You can't accidentally shoot ISO 800 without engaging the higher gain mode.

Also, you're looking at the low light version while I was looking at the normal light version. Yep, the 80D does particularly badly in that one at 25,600.

Ok, let it be your way. ISO 3200, GH5S is significantly cleaner than 80D in the shadows, and only slightly behind the Nikon and Pentax.

https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=lowlight&attr13_0=canon_eos80d&attr13_1=pentax_kp&attr13_2=nikon_d7500&attr13_3=panasonic_dcgh5s&attr15_0=raw&attr15_1=raw&attr15_2=raw&attr15_3=raw&attr16_0=3200&attr16_1=3200&attr16_2=3200&attr16_3=3200&normalization=print&widget=1&x=-0.7767140564090806&y=-0.4809009934909214

And you are completely missing the significance that those high ISO samples demonstrate. You absolutely do not need to shoot at ISO 25600 to benefit from that. You can shoot at ISO 1600 and judging by those samples here, you will be able to raise shadows A LOT more than with 80D. And rising shadows isn't some esoteric thing that no photographer ever does. It's one of the most common things to do and a bit reason to shoot RAW.

And yes, I was looking at low light version. That's the whole point of this camera. And when you are shooting in low light conditions, guess what? Yeah, you will be dealing with low levels of light and probably a bad light at that.

But this advantage will pop up in normal light as well as soon as you start raising shadows. Try shooting a high dynamic range scene and you'll see what I mean.

Besides, DPR's studio scene is very flat. Even the low light version hardly stresses any modern camera. >90% of the time you will be dealing with scenes of much greater dynamic range and this comparison tells you very little about that aspect. Yet you are acting as if it were all there is to it.

See my new post. The results here definitely are off.

BostonC - you do not get to choose ISO modes. ISO's 160-640 are shot using the first mode with increasing amounts of analogue gain. Then at 800 the camera switches over to the second mode, then starts to apply increasing amounts of analogue gain.

There's no user option to continue to use mode 1 or to jump to mode 2 early and 'pull' exposure. You can limit the camera to the ISO settings that use mode 1 or the settings that use mode 2, but that's all.

@Butler I saw your response in the other thread. Sry missed it earlier. Actually you can choose which mode to shoot. Under the 'Custom', you'd find this submenu: https://screenshotscdn.firefoxusercontent.com/images/0c574e8b-88b2-4e2b-91e2-105e1314b6fd.png
Say it again, DPR is wrong equating Pany 'Dual Native ISO' w Sony's 'Dual Gain' that's acquired from Aptina. They're very different. Sony does it automatically in the background, but for much limited step-up in ISO, typically at 800 while Pany does it much higher, at 2500 as in EVA1 or 5000 as in VariCam.

My original answer included the statement: "You can limit the camera to the ISO settings that use mode 1 or the settings that use mode 2, but that's all." I was referring to that menu option when I said that.

'Low' limits the camera to ISO 800 and lower, 'High' limits the camera to ISO 800 and above. In principle this means it might be possible to shoot ISO 800 in both the low and high gain modes, but that it.

Every part of Panasonic's description is consistent with Sony's implementation of Aptina's DRPix dual gain mode. All the diagrams, all the behaviour. All the specs and the appearance of the GH5S's sensor are consistent with it being a Sony IMX294 chip, which would also point towards it being dual gain. The switchover point is ISO 800 (in Standard Photo Mode, which is 1600 in VLog-L mode).

I think you're talking about a different submenu for selecting ISO upper limits, that has 2 entries for GH5s, as shown below:
https://screenshotscdn.firefoxusercontent.com/images/d2c4f895-e308-48bb-8f22-aa577240bb51.png
After spending two days going through Aptina's Dual Gain whitepapers, I'm certain it's different from Pany's DNI. The former has just one circuit but each photocell has a transistor that does the switch, while Pany's DNI has 2 separate circuits that can't be dynamically switched. Adding to the confusion, Pany's implementation of DNI is similar to Arri Alexa that's unfortunately is also called 'Dual Gain'.

I'm talking about the menu option you originally posted the screengrab of. I checked against the camera and it works just as I described in the article and just as I described above.

I've also compared the Aptina DRPix white paper to the simplified briefing presentation from Panasonic. Nothing in Panasonic's description, either in the presentation or my notes from the meeting contradict the idea that it's the Sony implementation of DRPix. "Two circuits" is exactly the sort of over-simplification that I was corrected for using in our writeup.

The rather simplified Arri writeup suggests its two paths are used in parallel, whereas the Panasonic version is switched, upsteam of the variable gain amplifier: per Panasonic's diagram.

@tkbslc
You should also compare other parts of the scene:
https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=lowlight&attr13_0=panasonic_dcgh5s&attr13_1=fujifilm_xt20&attr15_0=raw&attr15_1=raw&attr16_0=12800&attr16_1=12800&normalization=compare&widget=585&x=-0.11411071567321578&y=0.6663078718044029
(1 stop advantage for the Fuji)

and detail (for stills):
https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=lowlight&attr13_0=panasonic_dcgh5s&attr13_1=fujifilm_xt20&attr15_0=raw&attr15_1=raw&attr16_0=12800&attr16_1=12800&normalization=full&widget=585&x=-0.29958014597293164&y=-0.06648072004372048

Yeah, it's +2 stops in the areas that have a lot of light. Those aren't the areas that you'd expect to have intrusive noise. Strangely enough, it's the darkest shadows where the GH5S does best, shrinking the lead to just over 1 stop.

https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=lowlight&attr13_0=panasonic_dcgh5s&attr13_1=sony_a7sii&attr15_0=raw&attr15_1=raw&attr16_0=6400&attr16_1=12800&normalization=compare&widget=585&x=-0.767898767898768&y=0.7383223887994054

That's a very counter-intuitive result for me. Usually the deep shadows are the worst.

Miki. A stop is double. So two stops would be a multiplier of four. Three stops would be a multiplier of eight. Etc.......that is how stops work

The GH5s was designed for VIDEO work not Still Photo work. Things look very different when you compare 4K video performance.

We are talking about noise level, and it is not doubles with the total amount of light, it is increased with the ratio of the square root of the total amount of light. This is the reason that, overall you have to double the size of a sensor to get 50% noise benefit. With doubling you will get double the amount of light but the noise ratio is only 50% better. At least this is what I've learned from Great Bustartd's articles.

Here are two 1080p JPEGs taken from the digiDirect GH5s test video. The original video was shot in 4K, but I do not have access to the 4K footage. You can see for yourself that the GH5s was designed to shoot 4K video and NOT still-photos...

https://goo.gl/uAhq1e

https://goo.gl/wQRzVi

Marcus, my understanding was that the noise level decreases with the square root of the total amount of light; I might be wrong. Maybe, Great Bustard chime in here.
Oh, sorry, I think you are right! Or, not? Oh my. I am confused :-)

Yeah, it's a solid 1.5 stops behind the A7s II at high ISOs. Still a pretty good result, really.

The GX85 does not have a 7-year old 16MP sensor. That would be the GF6, which is specifically the example that I posted to show the improvement. That sensor debuted in the Panasonic G3 in mid 2011.

The GX85, by contrast, has a significantly improved 16MP, that generation of which debuted on the GX7 mid 2013, so 4.5 years.

It absolutely, quite easily compares to the X-T20. It is, being charitably, 1/3 stop behind. But it's very difficult to tell with X-Trans sensors, because the demosaicing does significant chroma noise reduction in Adobe Camera Raw, which DPReview uses for its comparison. What you are seeing what you say it is worse is colour noise, not luminance noise.

https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=lowlight&attr13_0=panasonic_dcgh5s&attr13_1=fujifilm_xt20&attr15_0=raw&attr15_1=raw&attr16_0=6400&attr16_1=12800&attr171_1=1&normalization=compare&widget=585&x=-0.6874791874791876&y=0.6927270474538473

And many people still buy Canon APS-C cameras...

@Astrotripper
I get your point. But we need to differentiate two things: signal to noise ratio for standard grey and dynamic range.
And the G5S creates this contradiction. The SNR is not better than other, even older, M43 sensors. It is also one clear stop behind the APS-C competition and two stops behind FF.

The dynamic range on the other hand exceeds all other M43 sensor on the market. (At high ISO) and as you explained rivals the A6500.

What does this mean in real life, at high ISO? Properly exposed subjects will show a similar amount of noise compared to other M43 sensors and more noise than APS-C sensors. But shadows and dark parts of your image will show less noise compared to other M43 cameras. Does that matter? Depends on your subject.

This is Stupid Photographer descript about Cinemotographer tool,LOL.
if you are doctor you talk Measuring Tape is garbage ha ha ha.
Apple or Orange,