Panasonic Lumix DC-GH5S added to studio scene comparison

http://www.imaging-resource.com/PRODS/panasonic-gh5s/zfield-test-1-a7scomp25600-l.jpg

Hmmm... Why does the Imaging Resource still-photo comparison look WAY different than DPR's studio tests ?

I wonder if DPR had the GH5s set to use 2500 native ISO when they did their tests ?

9 months ago*permalink

1) I believe IR's image comparison is a GH5S video grab, compared with an a7S II JPEG.

2) You cannot set the camera to different conversion gain modes. If you shoot at ISO 160-640 it uses one mode and when you reach ISO 800, it uses the second. You cannot accidentally shoot in the wrong mode (though you can limit the camera to one of the modes and its resultant ISO modes).

9 months agopermalink

It's interesting that the video image quality would be so different from the camera's JPEG results, especially if the video image resolution is close to the dimensions of the JPEG image.

9 months agopermalink

karacop

Comment on comments: M4/3 guys MUST convince themselves they made the right choice, and FF people MUST defend the obvious: bigger is better. Boring.

9 months ago*permalink

kewlguy

But the fact remains the same, FF is indeed better ;)

9 months agopermalink

karacop

I wrote: "the obvious"

9 months agopermalink

Veselin Gramatikov

No big advantage for stills. Nearly the same iso performance at 10mp as GH5.
Still behind APS-C sensors like Nikon d500. Far behind any FF.

This is to be expected. Still cameras benefit themselves with high resolution but when we speak for video the things will change. There high megapixel sensors do not have any advantage even if they shoot with their whole resolution and resize to 4k. GH5S will have less and bigger pixels and faster readouts. This will contribute to less heat too. Also GH5s have dual readout sensor. With good processing there are will be 1 to 1.5 stop advantage in video. So far not huge difference to be expected. The sensor is still way to small compared with FF sensor.

9 months agopermalink

Full Stop

For RAW stills it looks like my GX85.
2 stops behind FF. No miracles here.
If the great results are only visible in video, Panasonic perhaps takes advantage of the time domain for better NR.
With enough processing power they could use 4 sequential frames in order to interpolate lower noise levels for every single frame.
For one output-frame they would need to calculate 4x10.5MPix=42MPix.
Seems doable.

9 months agopermalink

TheTrickster2k

When will you add it to the (probably more relevant) Video Comparison Tool?

9 months agopermalink

SandySandy

I Compared it to the GH5, and the GH5s seems to be MORE noisy. UNBELIEVABLE !!!

PLEASE test what camera photos you guys have uploaded, cause this comparometre seems to be seriously flawed.

9 months agopermalink

Where are you seeing that?

9 months agopermalink

SandySandy

Richard Butler
"Where are you seeing that?"
I saw it on the comparison chart on yout site. I guess, it may have to down ressing or something on those lines. For photo comparisons.

8 months agopermalink

SandySandy

There appears to be some issues with your samples. Anything about ISO 160 on the GH5s seems to have visible noise, and by ISO 3200, it is pretty noisy.

9 months agopermalink

adjp

GH5S crashed. Broked. It's just physics. Bigger sensor, more light ;)

9 months agopermalink

Oleg Ivanovskiy

Bigger sensor - more light - that's correct. But 1. you can open aperture 2 steps wider and still have good in-focus depth. 2. You can attach full-frame lens via Speedbooster, that gives you another step.
Also, while the RAW shows 2 stops in A7S II favor, you cannot shoot RAW in video mode. And for JPEGs 2 stops scale down to 1 stop.

9 months agopermalink

IamJF

Looks like 2 stops for me ...

9 months agopermalink

Roderick Balle

This camera is very good at what it's intended for - actually it is pretty much peerless. That doesn't stop the trolls from big silicon coming down to tell us how deficient it is in something it wasn't designed for.

9 months agopermalink

zxaar

Why do you care for what others have to say. Its your money and if you are convinced then it is all that matters.

Likewise if others are not convinced they won't buy it. In the end for a company sales matter so if a lot of people are convinced like you the company makes money. (Off which we get nothing)

9 months agopermalink

kewlguy

how much is this camera? Some may argue that it's more for video, but the studio comparison tells me it's an overpriced small sensor camera (for stills).

9 months agopermalink

aloli

Pretty sure everyone can tell that this camera is more for video than stills.

9 months agopermalink

dombi

If you want a camera for stills, look elsewhere. As obviously this is a camera for video.

9 months agopermalink

Huge difference compared w test result from 1. imaging-resource:
http://www.imaging-resource.com/PRODS/panasonic-gh5s/zfield-test-1-a7scomp25600-l.jpg
2. DPR: https://screenshotscdn.firefoxusercontent.com/images/c0f320e1-7227-4955-bc44-6f0d9be16dbd.png

9 months agopermalink

bakanecko

cinema 5d found there is differences between the beta and production model which the production model have more noise than the protoype. Having said this, they will test much further.

9 months ago*permalink

Anulu

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

9 months agopermalink

I believe IR's images are a GH5S video grab (which includes temporal NR) alongside an a7S II JPEG.

9 months agopermalink

@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.

9 months agopermalink

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.

9 months agopermalink

Bear in mind that most cameras with Sony sensors released in the past couple of years also feature dual gain.

9 months agopermalink

Sidor

To me about the same as an Olympus e-M10. Maybe it works only for videos.

9 months agopermalink

Dante Birchen

Whoever is going to buy this camera for stills. Totally uninteresting how it performs. It's only feat photography wise is that it needs only half the pixels for the same IQ on high iso shots as the 20mp m43 sensor. So you'll get smaller files.

9 months agopermalink

So, purely a video camera then. This is as clean as N4/3 gets and for stills the 80D is beating it while the KP blows it away - for stills.

9 months agopermalink

eno2

The only thing the 80D is beating it's the dust. :)

9 months agopermalink

raw, 6400 ISO common size. No contest, and really no different from any other M43, which is also no surprise. Lowering resolution only reduces noise at the pixel level. It has very little benefit at the image level. But the GH5S certainly produces clean video, so maybe the reduced resolution achieves this by freeing up more processing power for noise reduction?

9 months agopermalink

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=25600&attr16_1=25600&attr16_2=25600&attr16_3=25600&normalization=print&widget=1&x=-0.7423182756248567&y=-0.6414868105515585

@whakapu

Are we looking at the same comparison?

Coz I agree there is no contest, GH5S blows Canon 80D out of the water starting from ISO 6400, while being maybe half a stop behind Pentax and Nikon. Incidentally, GH5S also blows any other MFT camera at the highest ISO settings.

9 months agopermalink

https://www.dpreview.com/news/2333575124/panasonic-lumix-dc-gh5s-studio-scene-posted

Well no we're not looking at the same thing. You're looking at 25.600. Nobody shoots at 25,600 except for experimental purposes. The dual gain has indeed kicked in by then but in the usable range - up to 6400 - the GH5S is still behind the worst APS-C sensor, closer in jpeg. It's a good performance but certainly nothing game changing about it for stills. And all that comes at the cost of ability to crop. It throws away the advantage crop sensors have of getting more pixels on subject. Essentially it's a quarter of an a7RII sensor.

9 months agopermalink

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.

9 months agopermalink

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

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.

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.

9 months agopermalink

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.

9 months ago*permalink

For the 3200 sample I'm seeing the Panasonic producing more green noise, the canon more magenta, at about the same level. It's called "photography" because it graphs light. I'm not sure what the term for ...darkness-graphy would be? Yes it does well in the dark but you're never going to get a pretty picture of the dark because its full of shot noise, and it's darkness. I might be missing your point but you're missing mine. There is no compelling reason to buy this as a stills camera. As a video camera it appears to compete remarkably well against full frame. As a still's camera you're throwing away a lot of resolution for negligible gain. There are other stills cameras that match or beat that ISO performance at the same or lower price. We don't yet know how ISO invariant it is but isn't 3,4,5 stops of that enough?

9 months agopermalink

Are you sure you switched to the high ISO mode when shooting the tests at ISO 3200 and above? Right now GH5, GH5s seem to have the same noise level at ISO3200 when resized to 'comp' or 'print', in RAW. And pls add the GH5s to the video comp tool too. TIA.

9 months ago*permalink

When switched to 'comp' that is the case with nearly all cameras.

The exact same situation is true of the A7r II vs. the A7s II. There is no real advantage with stills when downsampled to the same output resolution.

Unless of course you're shooting ISO100,000+. But no image shot at that ISO will be a good one, anyway, so it's purely a measurebation competition.

9 months agopermalink

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

9 months agopermalink

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.

9 months agopermalink

@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.

9 months agopermalink

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).

9 months ago*permalink

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'.

9 months agopermalink

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.

9 months ago*permalink

zzzxtreme

I have seen still grabs of iso 100k videos from a7s slog2 which looks darn nice. In fact, in low light I shoot still jpegs using overexposed +2 slog2 profile and do the editing in davinci resolve.

9 months agopermalink

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

9 months agopermalink

jst13

Could be. This is a fotography forum and most of us are only looking for stills. Also the comparison Tool is for stills we discuss about. I see no video comparision in this forum up to now. We discuss what we see and for still it is nothining special.

9 months ago*permalink

...But if the camera was designed to be a 4K video camera from the ground up, it's going to score quite poorly when measured using still-photo tools.

9 months agopermalink

The A7SII performs much better when shooting still photos at high ISO speeds, but the GH5s is a camera that was designed for 4K video work so the results look very different when comparing 4K video at speeds up to 25K ISO.

9 months ago*permalink

sbu

A lot of armchair skeptics here who hasn't really understood what this camera is optimized for, lol

9 months agopermalink

piratejabez

It handles the tricky green foliage/fluff rather well.

9 months agopermalink

Sezano

I don’t do video but..from this exercise, and for my purposes A smallish 35 mm f1.7 lens welded lens into camera (point and shoot) with a m4/3 sensor would be the best use of this medium.

Jus sayin, and maybe off topic.

9 months agopermalink

Sezano

35mm equivalent that is

9 months agopermalink

Roland Schulz

Hmm, similar noise to a6500, but a6500 has MUCH more resolution so noise reduction in post will blow GH5s like the a6500 already does at base ISO which is also possible to be set 2/3 stops lower with the Sony. Not what many had expected.

9 months agopermalink

Entropy512

The one problem for both the S2 and A6500 in video is that they don't offer 10-bit recording either internally OR externally, which may limit the ability to do NR in post compared to the GH5S internal temporal NR.

That said, if the noise amplitude is sufficient in the A6500, then likely 8 bits should be enough as long as you're not excessively aggressive with grading.

It's pretty clear Sony doesn't need to do anything "special" as far as sensor tech to answer the GH5S, but they desperately need to deploy BIONZ implementations that support 10-bit H.264. It's unlikely they can retrofit the A6500 and A7S2, but maybe there's a small possibility they can at least deploy 10-bit external.

9 months agopermalink

Robert Schroeder

And there goes the myth of the magic sensor. It is now obvious that the Panasonic engineers have some new noise-reduction tricks up their sleeves when it comes to video, but for stills, especially in RAW, the sensor looks exactly like a new Four Thirds sensor should be expected to look compared to a two-year-old 24x36 sensor – slightly less than 2 EV steps worse.

9 months agopermalink

Well yeah.

The A7S sensor has mythology surrounding it that is completely unfounded, too.

For stills, you're better off downsampling low-light shots from the A7r II. Same noise, more detail.

9 months agopermalink

Entropy512

I think the original S sensor was back when Sony hadn't deployed gapless microlenses AND they were using older manufacturing technologies that had poor area efficiency at fine pixel pitch.

The S sensor was also the first one to deploy dual-gain.

Gapless microlenses, improved manufacturing, and dual gain have become standard in all Sonys announced since (and including) the A7R2. The R2 has gapless + BSI, the A6300/A6500 are FSI on a newer manufacturing process and I'm fairly certain are using gapless microlenses - at least it's basically on par with the R3 in terms of performance per unit area despite not being BSI and having a finer pixel pitch.

9 months agopermalink

dansclic

Comparing with Sony A7 II , you see on Panasonic shots a lot of circles on the color circles/targets which do not appear on Sony shots, why that ?

9 months agopermalink

Max Iso

Moire.

9 months agopermalink

Yep, moire. Also, it absolutely is visible on A7 II, even on A7R II and even on Hassy X1D. It's a troubling consequence of Bayer colour filter array employed by the sensors on those cameras.

The only cameras not suffering from this is Leica M Monochrom, which does not distinguish colours, and Sigma cameras that use Foveon sensors that do a full colour capture (for Quattro, it's "kinda") at each pixel.

9 months agopermalink

zakk9

The RAW files look terrible, the jpegs look great. What's up with that?

9 months agopermalink

Great JPG engine? Sony and Panasonic really made progress with them

9 months agopermalink

ecka84

That's impossible :)

9 months agopermalink

Marc Ramirez

Is dual iso used here?

9 months agopermalink

Yes, there is no option to turn it off

9 months agopermalink

AmateurMike

Sensitivity only as good as the Sony a6500, but the a6500 downscales video from 6k to 4k, so will be sharper. Definitely not at the level of the a7sII.

9 months agopermalink

The 4K video from the A6500 is dramatically sharper than the video from the A7s II. Does that make the A7s II "definitely not at the level" of the A6500?

9 months agopermalink

AmateurMike

I meant the GH5s is not clean at the same ISOs as the A7sII, and also not as sharp as the a6500.

9 months agopermalink

noisephotographer

Still worse than Sony or Fujifilm 24 Megapixel Aps-c sensor.

9 months agopermalink

Not for 4K video it isn't.

9 months agopermalink

ecka84

1 stop behind APS-C.
2 stops behind FF.

9 months agopermalink

Are we looking at the same images? I'm seeing 0.1 stops behind APS-C and barely more than 1 stop behind FF. Definitely not 2 stops.

9 months agopermalink

@tkbslc
Actually similar performance to a GX85
https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=daylight&attr13_0=panasonic_dcgh5s&attr13_1=panasonic_dmcgx85&attr15_0=raw&attr15_1=raw&attr16_0=6400&attr16_1=6400&attr171_1=1&normalization=compare&widget=585&x=0.13586413586413576&y=-0.9658034939908312
-1stop APS-C
https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=daylight&attr13_0=panasonic_dcgh5s&attr13_1=fujifilm_xt20&attr15_0=raw&attr15_1=raw&attr16_0=6400&attr16_1=12800&attr126_1=1&attr171_1=1&normalization=compare&widget=585&x=0.12685437760288043&y=-0.9668375089468237
-2 stops FF
https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=daylight&attr13_0=panasonic_dcgh5s&attr13_1=sony_a7riii&attr15_0=raw&attr15_1=raw&attr16_0=6400&attr16_1=25600&attr126_1=1&attr171_1=1&normalization=compare&widget=585&x=0.12685437760288043&y=-0.9668375089468237

9 months agopermalink

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

you should be comparing with the low lighting setting when you want to see low light performance. In the dark, there's almost no difference vs APS-C at ISO 12800: 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=-0.12558828390491722&y=-0.6316638774509042

We don't typically need ISO 12800 when things are well lit, so the daylight simulation setting is a bit irrelevant.

9 months ago*permalink

noisephotographer

In low light at Iso 25000 the GH5S seems to have an advantage over Aps-c cameras. https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=lowlight&attr13_0=panasonic_dcg9&attr13_1=panasonic_dcgh5s&attr13_2=fujifilm_xt20&attr13_3=sony_a6500&attr15_0=raw&attr15_1=raw&attr15_2=raw&attr15_3=raw&attr16_0=25600&attr16_1=25600&attr16_2=25600&attr16_3=25600&attr126_0=1&normalization=compare&widget=1&x=0.1653437475993795&y=-1.030827300766805

9 months ago*permalink

ecka84

@tkbslc
Well, yes, the ISO12800 is kind of irrelevant, because I don't use it in low light even on FF. But the a7sII gives a very similar result at -2 stops.

@noisephotographer
Are there any video oriented 10mp APS-C cameras to compare with? They (crops) are all really bad at 25600.

9 months agopermalink

Podz

@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)

9 months ago*permalink

Obviously a 24mp sensor is going to beat a 10mp one for detail!

9 months agopermalink

caravan

Looks about average…No pixels were harmed.

9 months agopermalink

Anulu

Imaging Resource said Gh5S is the new king of low light...
I don't know what they are smoking but i want to :D
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=12800&attr16_1=12800&normalization=full&widget=585&x=0.15501165501165493&y=0.424854417048693

9 months ago*permalink

LJ - Eljot

Both shot at f/5.6 and 1/40 s at 12800 ISO, but one with 43.27mm image circle and one with 21.6mm. The GH5s is shot with a much higher debth of field.

9 months agopermalink

IR are talking about 4K video recording, NOT still photo performance. Take a look at actual VIDEO tests and you will see things are very different compared to still-photo performance.

9 months agopermalink

https://goo.gl/uAhq1e

https://goo.gl/wQRzVi

9 months agopermalink

OhWeh

I did compare the GH5S with the E-M1 MkII in RAW. I used the "Comparision" mode, which means, that both images are shown in the same size.

For me it seems, that up to ISO 25600 the E-M1.2 has slightly but definitely less noise in "Daylight Simulation" and vice versa (especially blue noise) in "Low Light".

So I can use my E-M1.2 and downscale the images to get the same HighISO-quality.

9 months agopermalink

The "Daylight Simulation" is not very useful to anticipate real-world results. Unless you generally shoot during the day at ISO 25600 with a 10-stop ND on your camera?

The "Low Light" sim is the only one worth using, imho. Most direct correlation to my experiences actually shooting different formats in low light.

9 months agopermalink

zeratulmrye

nothing unexpected

I failed to see its value compared to GH5, 240FPS vs 180FPS?

9 months agopermalink

citrate

Better noise reduction in video?

9 months agopermalink

It's not only noise reduction. The way the data is read out from the sensor is different for stills and video. So is processing of that data.

And check out shadows in low light scene at high ISO. The difference there compared to GH5 is very much obvious.

9 months agopermalink

Albert Valentino

Makes me feel good a about still using 2013 technology in the form of a Olympus EM1 mark i. I do not shoot video and shoot mostly at ISO 200, rarely going over ISO 800. So no plans to upgrade. These days it’s really okay for most of enthusiasts to skip a generation of technology, especially if you are happy with the results your getting 😃

9 months agopermalink

FLruckas

Try 10 year old technology...
I use FF a lot and it's a 5dii...
Video gear is a different story....
I'm with you on the skipping generations...
:=)

9 months agopermalink

LJ - Eljot

The E-M5 is 1.5 y older and is basicly the same as the E-M1 in IQ. The E-M1 has other advantages though.

9 months agopermalink

Ravilo

Call me blind, but I see about 1 1/2 stop difference in noise from GH5s to A7S II. Not bad at all, considering 4x smaller sensor!

Shame it has no IBIS, though.

9 months agopermalink

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=25600&normalization=compare&widget=585&x=-0.1211135173217083&y=0.49073140083995287

Yep, that's about what I'd say.

RAW still comparisons are also never favourable to these video-focused cameras. The 42MP A7r II looks at least as good, if not better, than the 12MP ultra-low-light focused A7s II until ISO 12800. It's only arguably at ISO 25600 that there's any reason to pick the A7s II, and only at 51200 when the argument becomes definitive...but still less than half a stop difference to pick between the two!

It's pretty clear that Panasonic has invested their resources in the video codecs to make the sensor look good in near total darkness.

If only because no one views 4K video on a 1080p monitor at 100% crop the way we do with still photos...

9 months agopermalink

TrojMacReady

Depending where you look, up to 2 stops, which is exactly what to expect for a 4 times smaller sensor with similar efficiency per unit sensor area.

9 months agopermalink

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

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.

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

9 months ago*permalink

TrojMacReady

That still looks to be roughly 2 stops, but I wouldn't judge noise in corners, because that's where light falloff (and possible under the hood correction) of lenses comes into play too.

This is the darkest bit and for all practical purposes, noise is similar at these (2 stop different) settings:
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=25600&normalization=compare&widget=585&x=-0.13327428744095426&y=-0.8670804085221829

9 months agopermalink

LessMirrored19

Sigma sd quattro H image is so clean and sharp at iso 100
What's the type of magic sigma did with sd sensor ????

9 months agopermalink

It's a Foveon sensor. An entirely different kind of sensor that uses 3 coloured layers rather than colour sub-pixels.

It has terrible sensitivity (i.e. ISO performance), but miraculous resolution, since each pixel has total R, G, B colour information, whereas all other sensors need to average out and interpolate that colour information from neighbouring pixels to "guess" what colour each pixel actually is.

9 months agopermalink

Mike FL

Other than too much hype for high ISO performance, I really do not see any meaningful difference than older MTFs such as Olympus EM5.M2.

9 months agopermalink

AT ISO 6400 the 7 year old sensor in the GX85 looks almost the same compared to the new GH5s. Great Achievement!

9 months agopermalink

I see no improvement at all. It's still 2 stops behind FF sensors, just as all previous sensors 1/4 the size of FF.

9 months agopermalink

I see, you say that two stops gives four times more noise.

9 months agopermalink

I wasn't saying that here, but essentially yes. If there is 2 stops less (i.e., 1/4 as much) light the amplification needs to make the signal 4 times as high, which makes the noise 4 times as high as well.

9 months agopermalink

golfhov

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

9 months agopermalink

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

9 months agopermalink

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.

9 months agopermalink

https://goo.gl/uAhq1e

https://goo.gl/wQRzVi

9 months agopermalink

golfhov

miki
here is ISO 3200 vs ISO 12800 that is a multiplier of 4x.
https://www.dpreview.com/news/2333575124/panasonic-lumix-dc-gh5s-studio-scene-posted?comment=4400691967
That is two stops because each time it is double. The next stop would be 8X even though it is only one more stop. Simple concept. If you double the size of the sensor(using similar technology) then the next size should actually have the exact same noise level BUT since it is 100% bigger the effects should be 100% less. Comparing MFT to APSC or APSC to FF generally shows that to be true. Move the cursors by one stop and the results end up similar

9 months agopermalink

SaltLakeGuy

at higher ISO's there is a considerable greater detail with the Sony. I think we can all move on and wait for another camera to come along truly ground breaking

9 months agopermalink

Bhima78

Interesting. This camera isn't better at high ISO in stills than the EM1ii at all (a bit worse-solid half-stop behind. Set it to Comp)... but somehow it competes with the A7Sii in video high ISOs (6400 and 12800).

9 months ago*permalink

Rensol

You are not getting more details here compared to the old E-M1 at ISO 25600 (on this test scale).
So image quality is NOT ANY BETTER, It is not a photo camera by any means.
This is purely Video camera fit inside photo camera body in hope that some photographers will pick it up (which would not happen should it be released as a video camera).
Smart move by Panasonic

9 months ago*permalink

Rich Evans

Outstanding achievement and progress for the m43 sensor that now it is only 2 to 3 stops behind a full frame. This is truly the most that has even been milked out of this sensor in terms of higher iso.

9 months agopermalink

Yes. That is normal. Since a m43 sensor is four times smaller than a FF sensor, its noise level is only two times behind FF, definitely not three. Nothing special here, pure laws of mathematics/physics.

9 months agopermalink

Miki, its noise level is 4 times (=2 stops) behind FF.

9 months agopermalink

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 :-)

9 months ago*permalink

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

9 months agopermalink

Miki, let's just make it simple and say "2 stops more noise". :)

9 months agopermalink

citrate

I was expecting more improvement than this when comparing to GH5/G9 in raw.

9 months agopermalink

otto k

Why? Video is benefiting from state of the art noise reduction, RAW files - not really.

9 months agopermalink

GinoSVK

That's one giant step for Four Thirds, one small leap for industry.

9 months agopermalink

Comparing the G9 to the GH5s at ISO 6400 i would give the edge to the G9.
The GH5s really is no stills camera

9 months agopermalink

LessMirrored19

Micro43 will be micro43 forever lol

9 months agopermalink

https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=lowlight&attr13_0=panasonic_dcgh5s&attr13_1=panasonic_dmcgf6&attr15_0=raw&attr15_1=raw&attr16_0=6400&attr16_1=6400&normalization=compare&widget=585&x=-0.8885281385281386&y=0.18262916615041522

Sure...

But check this out:

4.5 years is not that long.

And it's basically identical to the Canon 7D II, which is still somehow a current model:

https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=lowlight&attr13_0=panasonic_dcgh5s&attr13_1=canon_eos7dii&attr15_0=raw&attr15_1=raw&attr16_0=6400&attr16_1=6400&normalization=compare&widget=585&x=-0.48451548451548465&y=-0.8746128112997149

Just 1 stop behind a Canon 1DX or Nikon D4...

9 months ago*permalink

https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=daylight&attr13_0=panasonic_dcgh5s&attr13_1=panasonic_dmcgx85&attr15_0=raw&attr15_1=raw&attr16_0=6400&attr16_1=6400&attr171_1=1&normalization=compare&widget=585&x=0.13586413586413576&y=-0.9658034939908312

@Androole
Why pick the worst examples?

It is not better than the cheap GX85, which has the 7 year old 16MP sensor

https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=daylight&attr13_0=panasonic_dcgh5s&attr13_1=fujifilm_xt20&attr15_0=raw&attr15_1=raw&attr16_0=6400&attr16_1=6400&attr171_1=1&normalization=compare&widget=585&x=0.12685437760288043&y=-0.9668375089468237

It also can't even remotely compete with modern cheap APS-C cameras, like the X-T20

Also if you compare the 42MP FF A7R III you can see that MFT ISO 6400 is still only as good as FF ISO 25600

9 months ago*permalink

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

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.

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

9 months ago*permalink

@panther fan

By looking just at those samples and ignoring everything else, you are basically guaranteed to arrive at dubious conclusions. The DPR studio comparison scene tells you very little about many aspects of image quality. Stuff like:
- dynamic range
- tonal range
- pattern noise (raising shadows or in general pushing the image up)
- amp glow (long exposures)
- hot pixels (long exposures)

You demonstrated this by claiming that this is no different than GX85. I just clicked your link and switched to low light scene setup, and the difference between those two models is very much obvious and easily 1 stop.

I then switched to a6000, and GH5s is clearly better. Then a6500, and they are basically identical. Above ISO 6400, I really don't see any noticeable difference between GH5S and the best of APS-C cameras. I bet you would not be able to tell the difference between them in a blind test (Fuji notwithstanding, because X-Trans).

9 months agopermalink

And comparing to Fuji is just plain silly. If you'd apply similar chroma noise reduction to Panasonic's image, it would be no worse than the one from Fuji.

9 months agopermalink

@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.

9 months agopermalink

agachart

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,

9 months agopermalink

Slaginfected

@panther fan: This "less" noise becomes easily visible down to the point that this even shows in the studio test scene. For example, the GH5 shows no blacks at ISO6400 "low light" scene, because the noise floor is getting into the visible parts of the image already (i.e. would need corrections of some sort). It will show in the post-processing capabilities, colors, tonalities etc.

However, all this mumbo-jumbo is something close to noone talks about, because the main decision factor for high ISO (supposedly) is noise, or, if we want to be more precise, "grain", because most of the noise we perceive and call noise is shot noise, whereas these mumbo-jumbo things are results of limitations in the SNR/DR department, i.e. read noise.

Shot noise pretty much depends on sensor size, i.e. should be about the same for all m4/3rd, but the read noise isn't. Together with the saturation capacity this is also the reason why the GH5s can do 14 bit and not just 12 bit ...

9 months agopermalink

CallumG

Good detail considering the MP's and sensor size. Shame that it's buried under all that noise.

9 months ago*permalink

Still doesn't compete with APS-C and is far behind FF. But that was to be expected. There is no miracle camera

9 months agopermalink

Entropy512

Yup. There were pretty strong hints at temporal noise reduction in the first low-light shootout (such as the city lights behind the Christmas tree at around 2:00 in DPR's video), this makes it clear that's how the GH5S managed to keep up in the first video tests.

That said, unfortunately the A7S family is still limited to 8-bit for internal (and I think even external) recording, which limits the amount of NR you can apply in post for video. :(

Sony REALLY needs to move to 10-bit for future releases.

9 months agopermalink