SEL50F14Z high ISO noise

[Susurs]

Hi!

I’m kind of confused...

I used to use SEL55F18Z with my A7R3 and I loved it. I could take pictures @ ISO 1000 in dark environments and got vietually no visible noise (see the attached image).

Now I’m testing SEL50F14Z and I cannot get rid of ISO noise even @ ISO 80 in daylight (compressed .ARW) on the same camera, same settings.

What is/could be going on?


The first image indoors is with 55mm @ ISO 500

 

[mollyc]

View attachment 828783 View attachment 828781 View attachment 828782 Hi!

I’m kind of confused...

I used to use SEL55F18Z with my A7R3 and I loved it. I could take pictures @ ISO 1000 in dark environments and got vietually no visible noise (see the attached image).

Now I’m testing SEL50F14Z and I cannot get rid of ISO noise even @ ISO 80 in daylight (compressed .ARW) on the same camera, same settings.

What is/could be going on?


The first image indoors is with 55mm @ ISO 500

It's hard to say where your screen shot is from since the windows kind of all look the same. But at f/1.4 I'd be inclined to say that the image isn't noisy per se, but that screenshot area is just OOF a bit. I think if you'd shot with a higher ISO and f/stop you'd have more sharpness throughout.

I've also experienced graininess at odd times with very high SSs. I would say the settings are the issue, not the lens.

 

[Susurs]

Here is f/2.8 ... all the same
(Crop - small windows - center/right)

 

[mollyc]

Well based on the angle of the building it's clear that you were not perfectly square to the building. (No judgment, I shoot like a drunk sailor so much that I moved the Transform panel higher up in LR so I could get to it faster! ) But even at 2.8 there is still a fair margin of error for DOF issues based on where your focal point was.

Is there *any* spot in the image that has true sharpness? Also, how many MP is this camera? If a really high MP, are you new to it? Sometimes you need to view at 50-75% to judge sharpness on a high MP camera. I shoot with a D800 and there is an expectation adjustment based on resolution.

The other thing to consider is that with a new lens/body combo they might be mispaired and you might need to microcalibrate the lens to the body.

 

[Susurs]

As far as I believe the lens per se shoud not generate more noise.

 

[mollyc]

As far as I believe the lens per se shoud not generate more noise.

No, but if it isn't calibrated and is not focusing well to your body then the OOF areas will appear as noise/grain.

 

[Susurs]

42Mpix. I am used to this camera body. And I was used to a combo with SEL55F18.
[doublepost=1553700043][/doublepost]I have tested another images with ISO around 250. And focus on a lit round clock. That was night, however. Same grain in the focus area.

 

[mollyc]

To me it looks like a DOF issue which could be either than you need a smaller aperture or that the lens needs to be calibrated.

Hopefully someone else will chime in.

 

[Susurs]

Thank you!
P.S. As far as I know there are no calibration settings available.

I could try to reset camera or use uncompressed RAW, however compressed RAW shoud not in my opinion cause this.

 

[Clix Pix]

I have the impression that with mirrorless cameras it is not necessary to calibrate lenses the way it is with Nikon and other DSLRs.

 

[kenoh]

As far as I believe the lens per se shoud not generate more noise.

I don’t think it is. I think it is showing the difference between the Zeiss 55mm which is widely regarded as phenomenal and this one which is a more budget friendly option.

I thought it was softness rather than noise.

Stop it down to F5.6 to F8 and see but the lens wouldn’t cause more noise at a given iso.

 

[Susurs]

I don’t think it is. I think it is showing the difference between the Zeiss 55mm which is widely regarded as phenomenal and this one which is a more budget friendly option.

I thought it was softness rather than noise.

Stop it down to F5.6 to F8 and see but the lens wouldn’t cause more noise at a given iso.

It’s vice versa - 55 zeiss is considered more affordable version. This (50mm) lens is a big/heavy but reviews prise it as superior and compare to Zeis Otus for it’s ‘sharpness’...Should have excellent microcontrast. Well, it might be me, but I really can’t see any benefits of it over my previous Zeiss 55/1.8. Except that it (55mm) used to produce some CA in some scenes.

Overall image resolution seems not so high if compared to the 55mm, and it’s softer in corners. Seems softer to me in general... Produces orange fringing at some scenes. Maybe I just do not know how to take pictures with it.

Well ... Here is one from my evening walk. A bit post-processed version. Same building.

 

[mollyc]

I agree with Ken that it looks more like softness vs. noise (which is what I said originally, but nice to have confirmation). But that said, in your original image, the building is in the shadows - the sun is obviously behind the building. And noise typically lives in the shadows. So I'm not inherently surprised that you find it "noisy" in that specific image.

Also, how large do you ever intend to print any of these images? My 36mp D800 can look noisy, but I rarely print anything larger than 12x12 for a wall frame or scrapbook and my images are all perfectly sharp in print. So let's also put things in context.

 

[Susurs]

12’ it normally is as well

I can’t really blame it for being decentred or anyhow deffective. It just seems that lens resolution is not as I would like it to be and is in contrary to what reviewers say.

On the other side some close images are sharp and without any noise (see the attached image) OOB.

This lens is also supposed to be a bit different in corner/center sharpness at some wider apertures. 55mm version has better uniformity in center/corner sharpness. Also, light transmission should be lower in comparison to the 55mm (as reviewers say).

 

[Susurs]

Just for a quick update: I did some tests for the same scene with uncompressed/compressed RAW, auto/manual focus w/focus peaking. Everything looks pretty the same.

Tried to compare to the iPhone version. Well, not really useful results as the phone predictably produced notceably worse results (even in .dng)

 

[Susurs]

Not really sure what happened but I got some sharp images out of this thing today. Even some window/building pictures looked better (see the attached file (resized version) and crop). Well, I had DRO switched off for some time, but that probably should not affect RAW anyway.
[doublepost=1553810032][/doublepost]P.S. The crops are bexind pixel level.
[doublepost=1553810742][/doublepost]This looks pretty sharp to me @ f/1.4 (unedited but converted to .jpeg)

https://www.dropbox.com/s/je8jjwf3tw253qs/Photo 28-03-2019, 15 11 25.jpg?dl=0

 

[kallisti]

View attachment 829130 View attachment 829129 Not really sure what happened but I got some sharp images out of this thing today. Even some window/building pictures looked better (see the attached file (resized version) and crop). Well, I had DRO switched off for some time, but that probably should not affect RAW anyway.
[doublepost=1553810032][/doublepost]P.S. The crops are bexind pixel level.
[doublepost=1553810742][/doublepost]This looks pretty sharp to me @ f/1.4 (unedited but converted to .jpeg)

https://www.dropbox.com/s/je8jjwf3tw253qs/Photo 28-03-2019, 15 11 25.jpg?dl=0

I own an A7R3 and both of the lenses in question. I haven't noticed the behavior you are describing.

Several things to note.

@Clix Pix is correct in stating that you do not need to worry about AF fine tuning on the A7R3--one of the advantages of mirrorless is that focus is coming from the sensor and lenses do not need to be calibrated. However it is still possible for lenses to be "off" regarding performance--you can still get a bad sample and Sony is kind of bad in that regard.

Native ISO for the A7R3 is 100. This should be the lowest ISO that you ever use. Several of your examples are at ISO 80 or 50. Shooting at a lower ISO than 100 can introduce artifacts and it should be avoided.

Sony RAW files can have noise even at the base ISO of 100. It's not obvious in every shot, but it can be there for reasons not clear to me. See for example: https://diglloyd.com/prem/prot/ML/SonyFullFrame/SonyA7R_III-noise-SprinterVan.html or https://www.dpreview.com/forums/post/56280397.

Both of these are camera issues however and should not vary when comparing one lens to another.

You don't seem to have performed direct comparisons between the two lenses in a controlled manner. I would suggest this as your next step.

The subject needs to be the same. The ambient light and your exposure settings need to be exactly the same. Since the 55mm lens is a longer focal length than the 50mm lens, the compositions will be challenging to get exactly the same. It probably won't matter though for the purposes of the test. Having the same light on the subject and matching exposure settings *will* matter though.

Ideally you will want both exposures to be "to the right" on the camera histograms, meaning that they will be slightly overexposed without blowing out highlights. This isn't critical however since you will be comparing the two lenses and as long as the light and the exposures are held constant, the comparison should be valid. But as @mollyc said above, noise lives in the shadows so it is better (in general) to avoid underexposure (with any shooting, not just with this test).

The focus point should be the same with each lens--ideally obtained by zooming in with Live View. Or zooming in using the built-in EVF. Focus should be in the center of the frame and this should be the area you crop to in post.

I would perform the test at relatively open apertures (say f/2) and also somewhat stopped down (say f/8). Ideally would do the comparison on a tripod where ISO 100 can be used and the shutter speed becomes irrelevant because you won't see motion blur (because the camera is on a tripod). Alternatively shoot handheld with a shutter speed that is fast enough to mask any camera shake (like no slower than 1/125th sec, just to be safe). Yes, IBIS is awesome but for this test you are trying to compare the two lenses and you want to control for all variables as much as possible.

Shoot with both lenses, again making everything as constant as you can between the shots with the two lenses. Shutter speed must stay the same. Aperture must stay the same. ISO must stay the same. Focus point must stay the same. The ambient light must stay the same (so either shoot indoors with the same light or outdoors only on a day without clouds or with uniform overcast cloud cover).

In post, look at both images at 100%. Do *not* look at higher magnifications (as you seem to have done in your crops in this thread) as they are not accurate. Because the focal length isn't the same between the two lenses, it will be tempting to crop to 100% on the 55mm lens and then crop to >100% on the 50mm lens so the crops look equal. The 50mm lens will look worse (and should look worse) because you are cropping it beyond the sensor resolution.

The results should be that the two lenses are fairly close to each other. They are both very good lenses. I prefer the 50mm in general because it is 2/3 stops faster and while the difference between 50mm and 55mm is small, my "ideal" walk about or general purpose focal length is closer to 40mm and 55mm is just too long for me (which is one reason among many why my current favorite general purpose prime is the Sigma 40mm f/1.4--but that's a topic for another day!).

If under controlled testing, your Sony 50mm f/1.4 is significantly worse in IQ compared to your Zeiss 55mm f/1.8 then the problem may very well be that you got a bad copy of the lens. Sony quality control regarding lenses isn't ideal, so this is a very real possibility. I'd consider returning the lens and seeing if a different copy performs better.

 

[tizeye]

What @kallisti said.

I might add, most of my experience has been with the A7RII as I just upgraded to the A7RIII a couple weeks ago (and still have both bodies as I traded in a Nikon mount Tamron crop sensor lens for the $300 bonus)

- While I have enjoyed the 42mp sensor, there is a limit to cropability. At the level of major crop of just the window, the mp reduction would place it around a 5 mp photo...if you are lucky. I find that with all lens but the 42mp sensor does make the 70-200/f4 (my longest lens) viable with wildlife - quick reaction catching birds in flight, then crop. The more I crop, the blurrier, so finding the happy medium. Excessive cropping to a headshot - forget about it. That is the equivalent of what you are doing with the windows (except you could slow down and compose and presumably use a tripod vs my monopod). You may have hinted at it in the post of the table lamp and the windows noting not evident in near objects. Translation - you don't have a radical excessive crop.

- To compare, need to take as close to identical exposure of the same subject. While obviously there will be a slight different angles of coverage between the two lens if mounted at the same place, that is minor. What you can control is ISO, Shutter and aperture (keeping compensation levels constant) to the level that they overlap - or possible mathematical equivalence when dropping to f1.4 by adjusting the shutter speed with ISO constant.

- reiterating the need for identical photos from both lens on same camera body is the earlier example of the phone booth and the building. While no crop review on the phone booth, it was exposed for dark to retain the highlights. By contrast, with all that blue sky, the building was exposed for light. The windows on the shadow side of the building are effectively overexposed, losing their shadow detail that becomes apparent when cropped. It has little to nothing to do with the capability of the lens. In real estate photography, I do this to the extreme on interior photos (and to a limited extent on exterior when forced to shoot into the sun). Essentially, I expose (use manual settings - not "A") for the window view grossly underexposing the interior features, and use fill flash to bring the interior lighting up and possibly an ambient to bring back shadows at a 20% opacity. Exterior may set up fill flash if the front of the house - like, but impossible to do with your large building - is in shadows from the rear lighting.

 

[Susurs]

I own an A7R3 and both of the lenses in question. I haven't noticed the behavior you are describing.

Several things to note.

@Clix Pix is correct in stating that you do not need to worry about AF fine tuning on the A7R3--one of the advantages of mirrorless is that focus is coming from the sensor and lenses do not need to be calibrated. However it is still possible for lenses to be "off" regarding performance--you can still get a bad sample and Sony is kind of bad in that regard.

Native ISO for the A7R3 is 100. This should be the lowest ISO that you ever use. Several of your examples are at ISO 80 or 50. Shooting at a lower ISO than 100 can introduce artifacts and it should be avoided.

Sony RAW files can have noise even at the base ISO of 100. It's not obvious in every shot, but it can be there for reasons not clear to me. See for example: https://diglloyd.com/prem/prot/ML/SonyFullFrame/SonyA7R_III-noise-SprinterVan.html or https://www.dpreview.com/forums/post/56280397.

Both of these are camera issues however and should not vary when comparing one lens to another.

You don't seem to have performed direct comparisons between the two lenses in a controlled manner. I would suggest this as your next step.

The subject needs to be the same. The ambient light and your exposure settings need to be exactly the same. Since the 55mm lens is a longer focal length than the 50mm lens, the compositions will be challenging to get exactly the same. It probably won't matter though for the purposes of the test. Having the same light on the subject and matching exposure settings *will* matter though.

Ideally you will want both exposures to be "to the right" on the camera histograms, meaning that they will be slightly overexposed without blowing out highlights. This isn't critical however since you will be comparing the two lenses and as long as the light and the exposures are held constant, the comparison should be valid. But as @mollyc said above, noise lives in the shadows so it is better (in general) to avoid underexposure (with any shooting, not just with this test).

The focus point should be the same with each lens--ideally obtained by zooming in with Live View. Or zooming in using the built-in EVF. Focus should be in the center of the frame and this should be the area you crop to in post.

I would perform the test at relatively open apertures (say f/2) and also somewhat stopped down (say f/8). Ideally would do the comparison on a tripod where ISO 100 can be used and the shutter speed becomes irrelevant because you won't see motion blur (because the camera is on a tripod). Alternatively shoot handheld with a shutter speed that is fast enough to mask any camera shake (like no slower than 1/125th sec, just to be safe). Yes, IBIS is awesome but for this test you are trying to compare the two lenses and you want to control for all variables as much as possible.

Shoot with both lenses, again making everything as constant as you can between the shots with the two lenses. Shutter speed must stay the same. Aperture must stay the same. ISO must stay the same. Focus point must stay the same. The ambient light must stay the same (so either shoot indoors with the same light or outdoors only on a day without clouds or with uniform overcast cloud cover).

In post, look at both images at 100%. Do *not* look at higher magnifications (as you seem to have done in your crops in this thread) as they are not accurate. Because the focal length isn't the same between the two lenses, it will be tempting to crop to 100% on the 55mm lens and then crop to >100% on the 50mm lens so the crops look equal. The 50mm lens will look worse (and should look worse) because you are cropping it beyond the sensor resolution.

The results should be that the two lenses are fairly close to each other. They are both very good lenses. I prefer the 50mm in general because it is 2/3 stops faster and while the difference between 50mm and 55mm is small, my "ideal" walk about or general purpose focal length is closer to 40mm and 55mm is just too long for me (which is one reason among many why my current favorite general purpose prime is the Sigma 40mm f/1.4--but that's a topic for another day!).

If under controlled testing, your Sony 50mm f/1.4 is significantly worse in IQ compared to your Zeiss 55mm f/1.8 then the problem may very well be that you got a bad copy of the lens. Sony quality control regarding lenses isn't ideal, so this is a very real possibility. I'd consider returning the lens and seeing if a different copy performs better.

Thank’s a lot for excellent answers everybody! I really appreciate it!

Probably I’ll get used to this lens as well. Just some ‘adoption’ time is needed.

I don’t think there is something wrong with the lens itself at the moment. At least now I like some of the images it has produced a lot..and I hope that with practice there will be more and more of such images.

Unfortunately, I had an option to keep only one lens, so there is no 55mm to compare at the moment... but I’ll see what I can do there.

@kallisti,

Could I ask you in terms of pure sharpnes - center/periphery - do you find both lenses equal?

Do you have DRO on/auto? But that probably does not matter for RAW if I am not mistaken?

Do you prefer Adobe or sRGB?

As far as I am concerned 55mm produced a bit cooler color images. 50mm is a bit warmer.

55mm produced some CA in some cases, 50mm on the other hand, produces some minimal red fringing in some images.

 

[kallisti]

Thank’s a lot for excellent answers everybody! I really appreciate it!

Probably I’ll get used to this lens as well. Just some ‘adoption’ time is needed.

I don’t think there is something wrong with the lens itself at the moment. At least now I like some of the images it has produced a lot..and I hope that with practice there will be more and more of such images.

Unfortunately, I had an option to keep only one lens, so there is no 55mm to compare at the moment... but I’ll see what I can do there.

@kallisti,

Could I ask you in terms of pure sharpnes - center/periphery - do you find both lenses equal?

Do you have DRO on/auto? But that probably does not matter for RAW if I am not mistaken?

Do you prefer Adobe or sRGB?

As far as I am concerned 55mm produced a bit cooler color images. 50mm is a bit warmer.

55mm produced some CA in some cases, 50mm on the other hand, produces some minimal red fringing in some images.

I took some pics of a color calibration chart with a Sony A7RIII and 55mm f/1.8 lens and 50mm f/1.4 lens.

For both: 1/125th sec, f/5.6, ISO 100. Uncompressed RAW. All light came from a strobe and the strobe output was the same between lenses. Taken on a tripod and I did my best to not move the tripod when changing lenses. Used a wired remote to trip the shutter in both cases.

Zeiss 55mm f/1.8 lens

Sony 50mm f/1.4 lens

Of note as it relates to your original post, there isn't excessive noise with either lens and I can't see a difference between them.

For this particular series, the 50mm lens was sharper (the "Spydercheckr" in the upper left for example). However I wouldn't hold this up as proof one way or the other.

To answer your other questions:

Adobe is a better color space than sRBG, but if you are shooting RAW it doesn't matter for your files. However it can influence the histogram you see on the camera when reviewing pics, even if you are shooting only in RAW (sRBG erroneously showing blown out highlights that may not reflect details that are actually retained in the RAW file).

I agree that the 50mm renders a bit warmer than the 55mm. Easy enough to change in post if you are shooting RAW and have a preference one way or the other.

 

[Susurs]

[doublepost=1554156037][/doublepost]

[doublepost=1554156486][/doublepost]

Full images RAW/JPEG: https://www.dropbox.com/sh/eqz1g9i8z66imay/AAAET92p5lsL4rMI7nHpio3Qa?dl=0

Well... Not sure what to think... On windows there is slightly a bit more noise with 50mm. Those are crops behind 100%, however. Link to original files included as well (Dropbox). Other images look pretty similar. What do you think? I can’t blame 50mm regarding sharpness. On windows - the focus is where the crops are - third window from the lower side/blue building.

The only idea that comes to my mind is that 50mm has lower light transmission according to some reviews...

Thank’s to my friends for helping me arrange some quick test with 55mm.

 

[Susurs]

Looking a bit more at the whole series of images I took yesterday - seems like there are overall a bit more details/sharpness in 50mm (even when cropped to the same level/that means 50mm is more cropped). That is good.

Tiny bit more noise on 50mm on same settings when objects are far @ ISO 50/100 (at least to my eye and ‘more’ visible on some screens than other). @640 ISO both images look the same outside.

I do not see any other noticeable differences.

 

[kallisti]

View attachment 829910 View attachment 829911 View attachment 829906 View attachment 829905
[doublepost=1554156037][/doublepost] View attachment 829909 View attachment 829908
[doublepost=1554156486][/doublepost]

Full images RAW/JPEG: https://www.dropbox.com/sh/eqz1g9i8z66imay/AAAET92p5lsL4rMI7nHpio3Qa?dl=0

Well... Not sure what to think... On windows there is slightly a bit more noise with 50mm. Those are crops behind 100%, however. Link to original files included as well (Dropbox). Other images look pretty similar. What do you think? I can’t blame 50mm regarding sharpness. On windows - the focus is where the crops are - third window from the lower side/blue building.

The only idea that comes to my mind is that 50mm has lower light transmission according to some reviews...

Thank’s to my friends for helping me arrange some quick test with 55mm.

I downloaded the RAW files from you dropbox.

They both look fine. There is no significant noise that I could see in either image at 100%.

Your crops in the post I quoted are at >100%. They are thus not helpful in an evaluative sense. 100% is 100%. Anything > 100% means you are trying to see detail that is beyond what the sensor captured (even in a theoretical sense). Of course there is going to be "noise" in these crops--you are hoping to see detail that is beyond what the sensor captured. It isn't there and can't be there. Because it was beyond the limits of physics (based on the abilities of the sensor and focal length/optical characteristics of the lens used) to capture it. CSI and Law & Order notwithstanding, the highest IQ you can ever hope for in the real world has a hard limit at a 100% crop.

For real world use, both of your RAW images with either of the lenses are perfectly usable. There is no noise in either of them that would be apparent with any reasonable output for the images.

You are pixel peeping. The "flaws" you are seeing are not ones that will be evident with output appropriate to the focal length of the lenses you are using. You have an awesome camera. You have an awesome lens (either one).

I think that some of the flaws you are seeing aren't lens-dependent but are instead related to light and/or technique. Digital sensors have a problem with underexposed areas. It's a problem related to signal-to-noise ratio. The lower the light in an important area of an image, the more noise there is there compared to actual signal. Exposing to the right on the camera histogram for the important areas increases the ratio of "important" or "significant" data that is captured compared to the unimportant "noise".

Sometimes you can adjust settings related to exposure at the time of capture to fix this (more open aperture, longer exposure (sometimes requiring a tripod), higher ISO (accepting the loss in image quality which can result)). Sometimes you just need better/more light (perhaps using a speedlight or a strobe). You actually noticed this in one of your earlier posts--the images looked better with the 50mm lens in better/stronger light.

There *is* a difference in light transmittance between the two lenses with the 50mm transmitting less light at a given aperture than the 55mm. But I don't think it's significant enough to explain the differences you were reporting.

 

[Susurs]

I downloaded the RAW files from you dropbox.

They both look fine. There is no significant noise that I could see in either image at 100%.

Your crops in the post I quoted are at >100%. They are thus not helpful in an evaluative sense. 100% is 100%. Anything > 100% means you are trying to see detail that is beyond what the sensor captured (even in a theoretical sense). Of course there is going to be "noise" in these crops--you are hoping to see detail that is beyond what the sensor captured. It isn't there and can't be there. Because it was beyond the limits of physics (based on the abilities of the sensor and focal length/optical characteristics of the lens used) to capture it. CSI and Law & Order notwithstanding, the highest IQ you can ever hope for in the real world has a hard limit at a 100% crop.

For real world use, both of your RAW images with either of the lenses are perfectly usable. There is no noise in either of them that would be apparent with any reasonable output for the images.

You are pixel peeping. The "flaws" you are seeing are not ones that will be evident with output appropriate to the focal length of the lenses you are using. You have an awesome camera. You have an awesome lens (either one).

I think that some of the flaws you are seeing aren't lens-dependent but are instead related to light and/or technique. Digital sensors have a problem with underexposed areas. It's a problem related to signal-to-noise ratio. The lower the light in an important area of an image, the more noise there is there compared to actual signal. Exposing to the right on the camera histogram for the important areas increases the ratio of "important" or "significant" data that is captured compared to the unimportant "noise".

Sometimes you can adjust settings related to exposure at the time of capture to fix this (more open aperture, longer exposure (sometimes requiring a tripod), higher ISO (accepting the loss in image quality which can result)). Sometimes you just need better/more light (perhaps using a speedlight or a strobe). You actually noticed this in one of your earlier posts--the images looked better with the 50mm lens in better/stronger light.

There *is* a difference in light transmittance between the two lenses with the 50mm transmitting less light at a given aperture than the 55mm. But I don't think it's significant enough to explain the differences you were reporting.

Thank’s again!

You’re right! - ‘Case closed’