M43 High ISO

[Remo Nonaz]

I was pleased to see that Imaging Resource already has test images from the new Panasonic GX8. http://www.imaging-resource.com/PRODS/panasonic-gx8/panasonic-gx8A7.HTM. I downloaded some of the higher ISO RAW files and compared them to some test shots I have from my GH2. My initial reaction was, “Wow, this is terrific – at least two stops of improvement in high ISO conditions”.

I decided this might not be a fair comparison as the images I was using for the GH2 were my test images, not the Imaging Resource test scene. So, I searched around and found the older GH2 test images. http://www.imaging-resource.com/PRODS/DMCGH2/DMCGH2A7.HTM . These are a bit tricky to find as the GH2 is no longer listed on their main camera page – yea, I know, it’s old – and only JPG files are available, no RAW files. 

Attached are test shots for the two cameras at ISO3200 and adjusted in Lr. The GX8 is clearly the better image, if for no other reason than it is 25% bigger, but it is better in terms of clarity and grain as well.

Here is my issue: I’ve always considered my GH2 useless at ISO speeds above 1600. In fact, I won’t use 1600 unless I absolutely have to. The Imaging Resource test shots for both the GH2 and the GX8 at 3200 are very acceptable. Why? I think it is because when I need ISO1600+, I’m shooting a black cat in a coal bin, not a perfectly lit test stage. Why would I shoot a perfectly lit, static object, at ISO1600+? In good light M43 cameras shoot high ISO reasonably well.

But this is not realistic. When high ISO is needed, usually the lighting is pretty poor. How much improvement has been made with M43 imagers that can be seen in real-world, dark, difficult lighting situations? Is there a chance M43 will ever be able to shoot in low light or is a full-frame camera the only realistic tool for these situations?

[Telecaster]

Ah yes…real world handheld high-ISO pic taking. In film days I liked using Ilford's Delta 3200 at EI 1600 in low light. Any faster and the grain would tend to swamp image detail, which I didn't want. At the same time I liked using the same film in brighter conditions at EI 6400 or even higher. The idea with this was to accentuate the grain. Detail would hold up well enough.

Current m43 (GX7/8, EM-1/5ii), in my experience, works kinda the same way. But you can often use slower shutter speeds due to IBIS and/or OIS, which allows for lower ISOs. The increase in DOF for a given f/ value & field-of-view with m43's smaller sensor is usually a Good Thing in low light too. With the GX8 I've used ISO 3200 in dim lighting with consistently good (as defined by me, of course) results. ISO 6400 is IMO outside the sweet spot with tonally complex subject matter but still good otherwise. Keep in mind that with an f/1.4–2 lens this can allow photographing in conditions where it's hard to see (when not looking at the EVF) your subject matter. 

-Dave-

[Remo Nonaz]

I like your comment, "Keep in mind that with an f/1.4–2 lens this can allow photographing in conditions where it's hard to see (when not looking at the EVF) your subject matter."  I often wonder how many optical view finder fans understand that mirrorless cameras gain-up in low light and really improve your ability to shoot in dark conditions. I think it's one of their better, mis-understood features.

[Tony Ventouris Photography]

These are some shots with an E-M1 and the 40-150.  The dance photo is at ISO 4000.  The other three are at ISO 5000.  That's usually as high as I go.  Then again, I rarely if ever need to go higher than that.  Regardless what gear I have.  The frog was single-handheld holding the camera backwards to shoot inwards towards the frog which was underneath me.  Try that with a 300mm 2.8 lens and a canon body!  In fact I have a whole series of sharp images like that!  Unbelievable what the E-M1 can pull off. 

The dragonfly shot when zoomed to 100% shows distinction between all the eyes on each large eye.  I was surprised it resolved that at this high ISO.  All of these print crystal clear of noise.  Honestly, the way detail is retained, and the look of the grain...the images from the E-m1 at ISO 5000 are really not even a second thought.   

I really think people complaining about high iso is so overrated.  It opens up some interesting possibilities...but nothing that Ive seen dramatically changing the landscape of photography.  I think most people just want their snapshots in super dark restaurants and bars of friends to look like base ISO.

[Kiwi Paul]

These are some shots with an E-M1 and the 40-150.  The dance photo is at ISO 4000.  The other three are at ISO 5000.  That's usually as high as I go.  Then again, I rarely if ever need to go higher than that.  Regardless what gear I have.  The frog was single-handheld holding the camera backwards to shoot inwards towards the frog which was underneath me.  Try that with a 300mm 2.8 lens and a canon body!  In fact I have a whole series of sharp images like that!  Unbelievable what the E-M1 can pull off. 

The dragonfly shot when zoomed to 100% shows distinction between all the eyes on each large eye.  I was surprised it resolved that at this high ISO.  All of these print crystal clear of noise.  Honestly, the way detail is retained, and the look of the grain...the images from the E-m1 at ISO 5000 are really not even a second thought.   

I really think people complaining about high iso is so overrated.  It opens up some interesting possibilities...but nothing that Ive seen dramatically changing the landscape of photography.  I think most people just want their snapshots in super dark restaurants and bars of friends to look like base ISO.

My experience is similar, I use whatever ISO is needed to capture the shot at the shutter speed and aperture required and decide if the shots are worth keeping in post processing. I've taken countless shots at ISO's up to and even beyond 5000 and with a bit of careful noise reduction in LR the image quality is fine.

This macro is at ISO 3200 and no noise and plenty of detail.

P7030115

[Bob Rockefeller]

A formal comparison that I have not seen, but would like to, would be between perhaps an Olympus E-M1 and a Fuji X-T1 with hand-held "fast" prime lenses in the same lighting conditions.

The X-T1 will have to shoot at a higher ISO, all other things equal, to keep the shutter speed high enough for a sharp image. The E-M1 will benefit from its IBIS.

Does the X-T1's better noise characteristics at higher ISOs "beat" the E-M1's image at its lower needed ISO?

[Tony Ventouris Photography]

Low noise is low noise.  They all are so similar.  If the xt1 is in higher noise, it will look like higher noise, it's not cleaner than the em1 at low noise.  I think I actually have a staged set of comparisons from outdoors.  I'll have to dig through my archive.  I've also done this with the em1 and canon full frame, as well as a p45.  At base ISO, clean is clean.  Honestly, noise doesn't affect the image as much as we make it sound. 

[Bob Rockefeller]

Low noise is low noise.  They all are so similar.  If the xt1 is in higher noise, it will look like higher noise, it's not cleaner than the em1 at low noise.  I think I actually have a staged set of comparisons from outdoors.  I'll have to dig through my archive.  I've also done this with the em1 and canon full frame, as well as a p45.  At base ISO, clean is clean.  Honestly, noise doesn't affect the image as much as we make it sound.

I guess what I'm really thinking about here is if the E-M1 made an image that with "modest" noise reduction made a "good" image with "retained details" at ISO = Y, at what ISO would the X-T1 be able to make a similar image? ISO = Y x 2? Y x 3?

[Bob Rockefeller]

Another angle for interesting work would be a "noise reduced" ISO comparison to get at the question of how easy/hard is it to clean up a high ISO image from each sensor. Does a program such as Lightroom or Capture One do a better/worse job at cleaning up noise while retaining details from each sensor? Maybe the E-M1 has more noise, but it is more easily corrected?

[Tony Ventouris Photography]

The xt1 would need virtually the same settings to make a similar image.  If you were to adjust its ISO down for example, then shutter speed or aperture would have to adjust to reclaim the same apparent light capture.  But then it is no longer a similar image since depth of field changed or speed of capture changed which can introduce motion or stop motion.  In the real world, if I have a certain aesthetic I need and set my camera settings to achieve that, I would have to do the same on both cameras. 

Noise will be virtually a non issue.  There will be a small difference after 1600.  Half a stop, to a stop at most.  Most situations I bet you won't even notice that if the light and exposure were solid.

In Lightroom I never use more than a +25 or a +50 (if I pushed exposure or boosted shadows after the fact) luminance noise reduction on an ISO 5000 file.  I also have specific sharpening settings.  The e-m1 files over sharpen way quickly with similar settings for other cameras. 

[Bob Rockefeller]

The xt1 would need virtually the same settings to make a similar image.  If you were to adjust its ISO down for example, then shutter speed or aperture would have to adjust to reclaim the same apparent light capture.  But then it is no longer a similar image since depth of field changed or speed of capture changed which can introduce motion or stop motion.  In the real world, if I have a certain aesthetic I need and set my camera settings to achieve that, I would have to do the same on both cameras. 

Noise will be virtually a non issue.  There will be a small difference after 1600.  Half a stop, to a stop at most.  Most situations I bet you won't even notice that if the light and exposure were solid.

In Lightroom I never use more than a +25 or a +50 (if I pushed exposure or boosted shadows after the fact) luminance noise reduction on an ISO 5000 file.  I also have specific sharpening settings.  The e-m1 files over sharpen way quickly with similar settings for other cameras.

Where I'm going is to drive the shutter speed on the E-M1 down, taking advantage of the IBIS, and allow the ISO to stay lower. The X-T1 is reputed to have "much better" high ISO performance, so neutralize that advantage with lower shutter speed.

[Mousecop]

As usual, it depends on personal preference, and the specific use in question.

The 20mp sensor doesn't offer any substantial improvements over the 16mp sensor. You get a little more detail, and... that's about it.

The 16mp sensors did offer a nice jump over the 12mp's. I'd say 2 stops sounds about right.

I have no problems shooting up to 1600 (RAW) on a regular basis. I believe it reduces the DR a little bit, but not usually enough to cause a major issue. I can tell it's a little digital-grainy, but only when looking at 100%.

I've gotten pretty good images up to 6400 for web / casual use, and AFAIK it doesn't look good beyond 6400. As others have noted, it depends a little bit on your skills at NR.

There's no question that a 35mm sensor will offer better low light performance, and always will. At the moment, I believe it's around 2 - 2.5 ev, more with a speciality low-light camera. But, there are other considerations, such as stabilization; lens size and speed; are you stopping motion; how long is your lens, and so forth.

[Bob Rockefeller]

I have no problems shooting up to 1600 (RAW) on a regular basis. I believe it reduces the DR a little bit, but not usually enough to cause a major issue. I can tell it's a little digital-grainy, but only when looking at 100%.

Which camera do you shoot?

[Tony Ventouris Photography]

I do have some files of the Fuji at ISO 5000.  They are not of the same scene as the Olympus, but similar.  On comparison... There is no way I could say the Fuji is much better.  To me, the noise difference is negligible in real world use and situations share ideas shoot that ISO.  It's just not that different.  In fact, the xtrans  noise profile also looks more film like, similar to the em1.  Honestly, to me reviewing these files now after a while, I feel it's a wash and for me if I had to make the purchase decision over again, ide probably still go Olympus.  The Fuji is great, but it has limitations in operation for me compared to the em1 body. 

[Bob Rockefeller]

I do have some files of the Fuji at ISO 5000.  They are not of the same scene as the Olympus, but similar.  On comparison... There is no way I could say the Fuji is much better.  To me, the noise difference is negligible in real world use and situations share ideas shoot that ISO.  It's just not that different.  In fact, the xtrans  noise profile also looks more film like, similar to the em1.  Honestly, to me reviewing these files now after a while, I feel it's a wash and for me if I had to make the purchase decision over again, ide probably still go Olympus.  The Fuji is great, but it has limitations in operation for me compared to the em1 body.

I think I'm beginning to agree - the X-T1 is not a great deal better at noise than the E-M1 at a given ISO. I shot some boring images of a stack of books in my office in the evening at ISOs 800, 1600, 3200 and 6400 on both cameras. The X-T1 might be slightly better by the time I got to 6400, but not a lot and not even a whole stop.

In low light, I am convinced that an M.Zuiko f1.8 lens shot using IBIS has less noise that one shot by a Fujinon f1.4 if both are at a shutter speed that provides a focused image.

[scooby70]

I liked the beams under the pier and shot this with my GX7 at 14mm, f8, 1/30 and ISO 25,600. I've just used the same noise reduction and wider settings I'd use on any other shot. It's ok for screen viewing, noise and detail look ok and I think it'd make a decent sized print.

I could have used a wider aperture and brought the ISO down but I wanted deeper depth of field.

[Robert Falconer]

I think I'm beginning to agree - the X-T1 is not a great deal better at noise than the E-M1 at a given ISO. I shot some boring images of a stack of books in my office in the evening at ISOs 800, 1600, 3200 and 6400 on both cameras. The X-T1 might be slightly better by the time I got to 6400, but not a lot and not even a whole stop.

In low light, I am convinced that an M.Zuiko f1.8 lens shot using IBIS has less noise that one shot by a Fujinon f1.4 if both are at a shutter speed that provides a focused image.

Sorry to say, but at ISO800 and above, the Fuji starts to pull away a bit; it's about a full stop by the time you get to ISO3200 in my experience, a finding also more or less born out in this careful comparison...

https://photographylife.com/olympus-om-d-e-m1-vs-fuji-x-t1-iso-comparison

The laws of physics do kick in eventually, I'm afraid.

[AlterEgo]

Sorry to say, but at ISO800 and above, the Fuji starts to pull away a bit; it's about a full stop by the time you get to ISO3200 in my experience, a finding also more or less born out in this careful comparison...

https://photographylife.com/olympus-om-d-e-m1-vs-fuji-x-t1-iso-comparison

The laws of physics do kick in eventually, I'm afraid.

the laws of physics as opposite to your "experience" are :

1) S/N above deep shadows is guided by sensor size difference, E-M1 for example has 1.9 crop (Panasonic sensor) - not 2 crop, so it is quite less thant 1 stop... ~2/3rds... but even with Sony sensors (smaller than Panasonic) it is not a full stop as the difference in sensor size (the area from which the image is formed) is not 2 times.

2) S/N in the deep shadows (affects DR) illustrated by BClaff's tests @ http://www.photonstophotos.net  ... the new Sony 20mp M43 sensor (first appearance in Panasonic GX8) simply beats or matches 16mp veteran X-T1 ( http://www.photonstophotos.net/Charts/PDR.htm#Fujifilm%20X-T1,Panasonic%20DMC-GX8 ) ... till Fuji will switch to a new 24mp sensor from Sony they can't pull away.

3) Fuji does not have any real gain analog or digital past a nominal ISO ~1600 - they just write a tag to raw file as an instruction to a raw converter to apply a hidden expo-correction, so it makes no sense to compare raw from other cameras @ higher ISOs - raw shooter shall select the optimal ISO for each camera (even different nominal ISOs) and equal exposure (exposure time and aperture)...

===

"Official Fujifilm brand ambassador" = Fuji marketing grass makes a good smoke... get a calculator

[BJL]

Sorry to say, but at ISO800 and above, the Fuji starts to pull away a bit; it's about a full stop by the time you get to ISO3200 in my experience, a finding also more or less born out in this careful comparison...

The laws of physics do kick in eventually, I'm afraid.

It has been said hundreds of times before, but here goes:

The laws of physics also say some things about lenses and keeping things in focus.  Using high ISO speeds only makes sense when there is a need for a high enough shutter speed, and to get equal shutter speed at equal ISO speed, a larger format needs to use a larger focal length at the same aperture ratio, which means a bigger effective aperture diameter (entrance pupil size), and less DOF.  Such a lens is likely to be bigger and heavier, and maybe more expensive (but lens pricing can be mysterious!).

So equal ISO speed comparisons only apply when
(a) the larger format offers a lens of equally low minimum f-stop
(b) the lower DOF of that lens wide-open is acceptable
(c) the size, weight and cost of this lens is acceptable.

In the end, it is the ability to use this bigger lens that is the key to the better low light performance, which it achieves by gathering more light from the subject in the same exposure time through its larger entrance pupil.

On the other hand, in any situation where aperture choice is limited by the need to keep the subject in focus -- as in all those close-up examples above -- comparisons at equal ISO speed are irrelevant.  Instead that Fujifilm X-T1 will need to use an aperture ratio and ISO speed about 2/3-stop higher.  Similarly when considerations of weight limit the longer focal length lens needed with a larger format to a higher minimum f-stop.

[Robert Falconer]

the laws of physics as opposite to your "experience" are :

1) S/N above deep shadows is guided by sensor size difference, E-M1 for example has 1.9 crop (Panasonic sensor) - not 2 crop, so it is quite less thant 1 stop... ~2/3rds... but even with Sony sensors (smaller than Panasonic) it is not a full stop as the difference in sensor size (the area from which the image is formed) is not 2 times.

2) S/N in the deep shadows (affects DR) illustrated by BClaff's tests @ http://www.photonstophotos.net  ... the new Sony 20mp M43 sensor (first appearance in Panasonic GX8) simply beats or matches 16mp veteran X-T1 ( http://www.photonstophotos.net/Charts/PDR.htm#Fujifilm%20X-T1,Panasonic%20DMC-GX8 ) ... till Fuji will switch to a new 24mp sensor from Sony they can't pull away.

3) Fuji does not have any real gain analog or digital past a nominal ISO ~1600 - they just write a tag to raw file as an instruction to a raw converter to apply a hidden expo-correction, so it makes no sense to compare raw from other cameras @ higher ISOs - raw shooter shall select the optimal ISO for each camera (even different nominal ISOs) and equal exposure (exposure time and aperture)...

===

"Official Fujifilm brand ambassador" = Fuji marketing grass makes a good smoke... get a calculator

See, here's the thing. I don't use a "calculator" to evaluate my images, and neither do my clients. My eyes, the final results, and my client's expectations guide my evaluation.

In the field, Fuji IQ beats Olympus at high ISO. It is what it is. No need to be rude about it.