GH4 GH5 beats a7S, a7R II, GH4, E-M1 II, X-T2 & X-T20 in Low-light / hi-ISO test
- Thread starter radtex
- Start date
joe12south
Veteran
I'm not sure why you say it "beats" them, but yeah, with a super fast speedbooster/lens combo you can reach into the dark.
filmguy123
Veteran
This isn't specific to this, but one thing I always find annoying about online tests is they are
1. so unscientific, or TOO scientific
2. so unartistic
Lens rendition is so important to a camera's look. Not just focal length, the actual glass. And glass behaves differently on all cameras (ie what aperture to set at - same for lens performance, or relevant to camera?)
Each camera has settings to dial it in that are unique to each camera. How can you really account for that? Do you set them all optimally, all equally, or all at default? It depends on who is reviewing.
And most of the time, they don't even tell you.
In this way, not scientific enough.
And then, on the other hand, the tests are TOO scientific. Color boards and static shots of a guy playing pool. Who gives?
This may just be personal preference, but I would MUCH rather see tests (and I have seen some before) that have a knowledgeable DP dial in each camera to what they consider optimal settings, optimal glass, etc. and film a one minute artistic piece on 3 cameras from a storyboard, largely the same.
Color grade it to taste on each camera.
And have thorough notes about what they did and why.
To me, this would be the most meaningful thing.
I don't mean this to be off topic from a high ISO test, I bring it because each camera has a sort of "look" there and it made me think how meaningless it really is without knowing every setting, detail, etc. and adjusting to put each camera in it's best light.
Personally if I want to see if a camera is good, I'd rather just watch a film made on it by someone talented and judge it from that, not randos on youtube doing quasi scientific tests.
1. so unscientific, or TOO scientific
2. so unartistic
Lens rendition is so important to a camera's look. Not just focal length, the actual glass. And glass behaves differently on all cameras (ie what aperture to set at - same for lens performance, or relevant to camera?)
Each camera has settings to dial it in that are unique to each camera. How can you really account for that? Do you set them all optimally, all equally, or all at default? It depends on who is reviewing.
And most of the time, they don't even tell you.
In this way, not scientific enough.
And then, on the other hand, the tests are TOO scientific. Color boards and static shots of a guy playing pool. Who gives?
This may just be personal preference, but I would MUCH rather see tests (and I have seen some before) that have a knowledgeable DP dial in each camera to what they consider optimal settings, optimal glass, etc. and film a one minute artistic piece on 3 cameras from a storyboard, largely the same.
Color grade it to taste on each camera.
And have thorough notes about what they did and why.
To me, this would be the most meaningful thing.
I don't mean this to be off topic from a high ISO test, I bring it because each camera has a sort of "look" there and it made me think how meaningless it really is without knowing every setting, detail, etc. and adjusting to put each camera in it's best light.
Personally if I want to see if a camera is good, I'd rather just watch a film made on it by someone talented and judge it from that, not randos on youtube doing quasi scientific tests.
Thomas Smet
Veteran
A focal reducer is one of the main key advantages to the m43 format. While m43 will likely always be slightly less sensitive than larger sensor cameras there is a lot that goes into sensitivity vs just the size of the sensor. The fact is the GH5 can already do some impressive low light shooting considering its sensor size and its secret weapon is the ability to use a focal reducer which more or less eliminates any low light advantage any other camera has excluding maybe the A7S which is a special snowflake.
We do need to factor in what a focal reducer can do in regards to low light because it is a realistic and intended mechanism for shooting with a m43 camera. m43 is basically a camera that allows us to get compact and eliminate any overheating issues but gain back what is lost by adding in a focal reducer. It essentially turns a m43 sensor into a super35mm sensor in every single way except for dynamic range. Not that a solid 12 stops is bad or anything.
So yes it is fair and realistic to think a GH5 with a focal reducer would match or beat most other cameras out there in low light. What sliver of an edge other cameras may have is for an unrealistic shooting situation and negated by all the other advantages the GH5 brings to the table.
We do need to factor in what a focal reducer can do in regards to low light because it is a realistic and intended mechanism for shooting with a m43 camera. m43 is basically a camera that allows us to get compact and eliminate any overheating issues but gain back what is lost by adding in a focal reducer. It essentially turns a m43 sensor into a super35mm sensor in every single way except for dynamic range. Not that a solid 12 stops is bad or anything.
So yes it is fair and realistic to think a GH5 with a focal reducer would match or beat most other cameras out there in low light. What sliver of an edge other cameras may have is for an unrealistic shooting situation and negated by all the other advantages the GH5 brings to the table.
filmguy123
Veteran
I agree Thomas.
And not just focal reduced lenses, but also great glass such as the Voigtlander kit (workable at 0.95 and sharp at 1.2) among some other options.
And not just focal reduced lenses, but also great glass such as the Voigtlander kit (workable at 0.95 and sharp at 1.2) among some other options.
I personalty am not saying anything - that title is an quote of the testers title from the utube page...
Bern Caughey
Veteran
+1
joe12south
Veteran
Agreed, and that flexibility has made the GH4 usable for me in scenarios I would have had to give up with my native glass, but it's not the same value prop as a camera that has a sensor that is more sensitive. There scenarios in which you can reach comparable quality are more limited. While I'd still pick the GH5 for overall value, you don't need a special combo of adapter and lens to get better low light performance out of an A7SII.
So, I fundamentally disagree with the basis for this test. Test the low light response (output) of the sensors with matching input.
Northrup's comparisons between cropped-sensor and full-frame cameras are highly suspect. He applies different settings to each camera, and justifies this deceptive practice by citing mathematical "equivalence" conversions. For years, he has used DxO Lab's proprietary "perceptual megapixels" marketing hype to claim that a kit lens on an APS-C camera will produce "better results" than a pro full-frame lens.
At 3:49 in the video linked above, Northrup shows a side-by-side comparison between a GH5 using a Metabones Speedbooster versus a Sony A7R II using a conventional lens. But that's not all that's different - the GH5 is set to ISO 3200 while the A7R II is at ISO 12800! Why? Because... mumble mumble... the GH5 is technically gathering a little more light... especially since the A7R II has some pixel binning. What he means by "a little more light" is that the GH5 has a speedboosted f1.8 lens while the A7R II lens is f2.8. And what a surprise, the Sony looks noisier than the GH5!
TL;DR: Bulls*%t.
At 3:49 in the video linked above, Northrup shows a side-by-side comparison between a GH5 using a Metabones Speedbooster versus a Sony A7R II using a conventional lens. But that's not all that's different - the GH5 is set to ISO 3200 while the A7R II is at ISO 12800! Why? Because... mumble mumble... the GH5 is technically gathering a little more light... especially since the A7R II has some pixel binning. What he means by "a little more light" is that the GH5 has a speedboosted f1.8 lens while the A7R II lens is f2.8. And what a surprise, the Sony looks noisier than the GH5!
TL;DR: Bulls*%t.
Last edited:
It appears that he's trying to equate the DOF between the two formats which is legitimate, except that he's giving the GH5 a 2/3 F-Stop advantage with the SpeedBooster ( the effective F-Stop is f/1.1 when it should be f/1.4 to match the DOF ), which is a significant cheat to his test.
Still it's interesting to see that the GH5 with the 2/3 F-Stop cheat performs as well as it does, definitely FAR better than the GH4 ever did.
It's not just a speedbooster advantage. He's comparing the GH5 at ISO 3200 to the A7R II at ISO 12800. Not even close to a fair comparison.
alaskacameradude
Veteran
Well, to be fair, this is NOT some 'special advantage' of the m43 format....it's more of a flange depth thing
isn't it? For example, Super 35 (bigger chip than the m43) has exactly the same thing....'focal reducers/speed boosters'
to make them even more sensitive and better in low light......as long as you are using the E-mount, which in a similar manner
to the m43 mount, has a very shallow flange. Something like the Fs5 or Fs7 can use a Speedbooster, it's not some key
advantage that the m43 format has. I think because of the shallow flange, you have the space to have the 'focal reducer' and
still fit something like a FF Canon L lens on it.
Focal reducers on Micro Four Thirds are in no way an advantage for Micro Four Thirds. They simply make up for how small the four thirds sensor is, and imperfectly so, with added distortion, chromatic aberration, and vignetting. Put the same taking lens on an equivalent APS-C camera and everything about the image will be the same or better. Larger pixels have an inherent advantage for dynamic range, devoting a smaller portion of their die area to things like wires.
A focal reducer makes a lens perform better on a Micro Four Thirds sensor than the same lens would on the same camera without the focal reducer. That's the best thing you can say about focal reducers.
Actually, I think it is. If the GH5 is shooting f/2.0 at 800 ISO, then the DOF match for the A7S is f/4.0 at 3200 ISO. Given that the FF camera has to use a F-Stop that is 2 stops smaller to obtain matching DOF, how else are you going to get correct exposure ?
If you keep the same ISO then the FF image will be under-exposed by 2 F-Stops because it is using a F-Stop that is 2 stops smaller than the m4/3 camera to obtain the matching DOF, so the FF camera is forced to use an ISO that is 2 stops higher to obtain correct exposure.
I don't see a problem here. Unless you throw out the matching DOF and use the same aperture for both cameras, how are you going to get the correct exposure with the FF camera ?
Thomas Smet
Veteran
It is an advantage because you have the best of both worlds and a user can choose how they use it. They can use lenses as is or with a focal reducer. That to me is a clear advantage because it gives choices to the user. A FF camera is more or less stuck with that lens system and using those lenses as intended. No advantage can ever be squeezed out of them. With m43 a user can choose how they use the camera depending on their low light needs. A FF user pays for a low light advantage they may never use or need.
Low light is also extremely over rated and about as useful at judging camera quality as the unemployment rate used as the sole judge of an economy. It is a easy benchmark we have clung onto but not everybody needs low light. Most productions should never make use of anything higher than ISO800 no matter how clean it may look to the naked eye.
I agree Tony Northrup has some very odd methods of judging cameras and I rarely ever agree with anything he puts out there. I'm not even saying I agree with his methods in this latest comparison. I'm just saying that when one factors in a m43 camera they also need to factor in that is can use a focal reducer and can have access to very fast lenses like f0.95 primes. Tony also pointed out the fastest zoom the Sony can use is a f2.8 and that needs to be taken into consideration as well. A GH5 with a focal reducer and a f1.8 zoom is already 2 stops better than a naked GH5 compared to a Sony camera with a f2.8 zoom. That really does eliminate most if not all of that low light advantage. Of course it depends on the shooter and the situation but if one wants/needs to use a zoom lens then the GH5 can have certain advantages that people don't always take into consideration. Especially when you consider that GH5 setup will never overheat.
Especially if people think that great low light performance is somehow going to replace having to light your shots.
joe12south
Veteran
Something Tony has done quite well on multiple occassions is to explain how multiple different factors (not just crop factor) have to be considered to match a full-frame look.
But, respectfully, what does DOF have to do with low light sensitivity? When performing tests, it's important to not conflate too many different factors.
Thomas Smet
Veteran
Agreed. I have seen some A7S shoots that are impressive from an engineering perspective that they can see in the dark but I would never call them art or professional in any way. Visually they had about as much creativity as a snail. The film look has never been about seeing in the dark but using light to sculpt and define a scene with texture and depth. The more we cling to low light as a crutch the more we destroy that concept. It is impressive the A7S can see in the dark but if that means the image is flat dull and lifeless and any sources of light in the shot go nuclear then what is the point? A lack of noise in a shot doesn't mean it is instantly good. Just means it has less visible noise but it can still be equally ugly.
I see larger sensor cameras as having one advantage and only one advantage over a m43 sensor and that is dynamic range. That is something you can never fake/recover/interpolate. Not that 12 stops is horrible at all if done right.
joe12south
Veteran
Absolutely, but sensitivity *is* one of several key measures, and it is counter-productive to argue that it is anything but a weakness of the M43 format. The GH5 had to make herculean steps in terms of image processing in order to compensate. (And it does so amazingly well, IMHO.) But a camera with a better sensor could have spent that processing power on something else, so there is a price to pay.
I also find the argument that if you light intentionally (as we all should, even if we're using the sun) that there is not a benefit to more sensitive sensors. Ultimately, such a technology choice translates to creative choice. For example, if camera A needs a bunch of lights just to bring up the ambient levels to an acceptable point to then begin creative modeling, that camera has a time and money *tax* that camera B - which can work with the existing ambient levels - does not incur. The creative intent can be obtained easier and cheaper. That's nothing but a good thing.
A great example, I think, is the amazing cinematography on "Hell On Wheels". It was almost exclusively shot with natural light and practicals, but it is some of the most thoughtful lighting you'll ever see. It simply would not have been possible (at least not without being a grainy mess) with film.