Bigger sensor = Better Low Light Performance? Maybe Not

Do you upgrade to a full frame sensor because you want better low light performance? Maybe, it is not always the best choice.

Camera gear used to film this video:
Panasonic Lumix S5 | Lumix S 24mm f/1.8

King Icon by Chanut Is Industries

[AustralianImage]

I've tried to explain this many times in the past to stills photographers looking at going into FF, but it's difficult to convince them that the gains aren't always there. Unfortunately many get fixated on the shallow depth of field over all other considerations. And then it becomes a completely different story when it comes to video. Try maintaining focus with moving subjects, even if you have the best AF camera you can get. I saw the results that a wedding videographer took of my niece's wedding a while back, where I nailed focus using manual focus with my BMPCC4K, but theirs was all over the place with their Canon 5D Mk IV. In a three minute long highlight video, more than half of their video was constantly out of focus and/or hunting.

[gregor]

Fully agreed with your observations. I shoot both M43 and FF. My Nikon gives me roughly a 2 stop advantage over my Pana. All above ISO3200 I really don't care much.
However, it's not only ISO performance what counts to me. Dynamic range is crucial too. Also lens design resolving 24 MP is a plus when printing (I neglect multi shot scenarios at this point). So again it comes down to personal preference.
Your video makes it very clear not to underestimate M43 as it has it's place amongst many pro and enthusiast photogs. Thumbs up

[RoadFromNowhere_SE]

I’ve been frustrated with my full frames at times, back when the max ISO was 3200 and you’d still have noise reduction work to do in post. Nice having the S1H now. Natural light is about the only time one would run into this, but it’s important for many to understand what you are saying.

[Ulrich.Bierwisch]

I fully agree. I'm pretty happy with the low light performance of the GH5 with the 12mm/1.4. I get kind of the same results as any FF camera with a 24mm/2.8 and 2 stops higher ISO. This is OK for me in any situation where I can see enough with my own eyes. I even can buy Panasonic's F1.7 lenses that would put me almost on par with F2.8 zooms from 20 to 100mm.
The advantage for the FF-world is, that you can buy lenses that go a few stops further if you can live with the extreme shallow DOF. The envelope gives you more room in this direction.
The even bigger advantage for the FF system is at daylight. M43: 12mm/1.4 ISO 200 is the same as FF: 24mm/2.8 ISO 800. If there is to much light, M43 needs ND filters while the FF-system can go down 3 stops to ISO 100. This is where the 14 stops DR are coming from and you can't get them with M43. M43 is always like ISO800 or higher on FF.

PS:
Everything will change when computational multi shot systems are coming. Live ND from Olympus is a first step but the sensors need to get much faster readout.

[marcosalamon]

I completely agree with you. This is almost the same reason in my opinion little sensor are really good for street photography. As a street photographer I wish a camera with 1 inch foveon sensor and 28mm equivalent fixed lens. Bye from Italy.

[EugeneMaynard]

Each system has its place with its own advantage and disadvantage. The specific situation that you describe is case in point, that is why I shoot with cameras with various sensor sizes but if one has to choose just one then it is a difficult choice!😉

[stevebonn]

You make some very valid points. However, for moving wildlife in low light, micro 4/3 often has a disadvantage. You can reduce shutter speed and increase the aperture to some extent, but often one ends up increasing ISO and getting more noise. To some extent, denoising software can ameliorate the noise. At the longer distances of wildlife, it is often possible to shoot at a wider equivalent aperture and still get sufficient depth-of-field. Therefore, it would help further if more f/2.8-f/4.5 affordable telephoto zoom options were available in micro 4/3. For me, this disadvantage is manageable, given that I have the benefit of lighter and less expensive lenses, and that I'm often photographing either static scenes or in brighter daylight.

[imagenatura]

It's important to say sensor size doesn't affect aperture, depth of field, bokeh, focal length, shutter speed, ISO, or almost anything other than field of view and vignetting. I believe the difference in real-world usage is due to heat, pixel sizes, and read times.

[imagenatura]

I don't like equivalency videos, but this one comes at it from a shooter's point of view. I'm sure those FF shooters won't like it though. After all, they just sunk a tonne of money into their system to get better low-light performance. It must suck to hear they threw a bunch of money down the toilet.

[carriagel]

OK, but we don't have on market any m43 with ibis and dual native iso.. if you compare gh5 to s5.. what results you will get.

[enricolorenzani]

Great in dept review!!! I've ended up with the same set up for my S1R ( 24mm+50mm 1.8) I traded the 24-105f4 for them, I think ever if it's an excellent zoom prime lens will tame Better the amount of insane details from the s1r. the 24-105 was great at 24mm and I cannot deny the versatility, but for a landscape photographer like me this s 24mm 1.8 does even better although been very light and compact.

[Stephen.Bingham]

Bravo - real world low light performance is usually mainly determined by the size of the hole on the front on the camera, not the size of the sensor. Larger sensors have the advantage of being able to record more light - which results in higher dynamic range as long as you don’t mind the slower readout times of high precision analogue to digital conversation electronics.

[AbrahamLatchin]

Richard, you have done the test I have always wanted to see. Almost all video comparisons are performed at ISO to ISO comparisons. This then represents very different shots. Unlike in stills where shutterspeed is a variable, video generally have fewer SS options as they change the aesthetic of the video. What this seems to show is that the marginal benefit to a smaller sensor in stills is actually a greater gain in video.

The best way to look at it is light first, not sensor size. Can I add more light to improve my IQ (either by living with wider apertures or adding actual light to the scene) Once you hit base ISO area you might start to see the gains more clearly of the larger sensor. But it doesnt happen the way people think.

Edit note: Even with AF I dont often shoot moving/handheld video at lower than f2-f2.5 so using the widest FF apertures would result in a mess ;)

[lightlivingktt]

Brilliant and this is great way to break this down to people. I figure this out through trial and error when going back and forth between my full frame Nikon and crop sensor Nikon mirrorless cameras,

[ggdfggdfgdffgfddg]

Ну тут вы немного не правы. На системе 4/3 светосила=полному кадру, другое дело боке не равно.

[vladislavihl]

I tried out my Lumix G9 with a Sigma 16mm f1.4 lens against my S5 with a kit lens filming indoors in low light. The S5 had ISO6400 but I’m not sure what the sensitivity the G9 had, I suppose it was ISO 1200… Anyway, both footages had unacceptable level of noise but the footage out from the S5 had significantly less noise.
Generally speaking, I find mft better for pure filming. But for a hybrid shooter like me (90% stills, 10% video) the FF may make more sense.

[rickkoloian]

Richard, outstanding, your reviews are objective, logical & informative. Thanks for posting,

[Interceptor]

It is worth mentioning that while there are two stops of gap in terms of depth of field usually there are 1.5 stops or less in terms of low light performance. So your lumix 10-25 1.7 does better than the S5 with the 20-60mm lens across the focal range. The benefit of larger sensors are in very low level of light (think astrophotography) with focus at infinity when you shoot wide open or in scenes with a lot of dynamic range where the benefit is on the highlights. This is a well known fact since a long time so nothing new here

[MarkoKoskenoja]

That was an interesting comparison 🤔 I'm not sure myself but I trust you are correct with your assertions.

[yaupie]

What you said is logically correct. But I don't know how often a FF shooter would want not so shallow DOF which could readily achieved by M43 gear. As you use both M43 and FF cameras, perhaps you can share more about your own experience. (P.S. I don't use any FF camera)