Potential amp glow on X-E3

Hi,

Just got a new X-E3 this evening and took it for a walk to shoot some snaps at iso 12800 - as you do

I was quite shocked when I came back and wrestled with the RAF files long enough to give up on X Raw Studio and throw them into LR, where I found this purple bar along the bottom of the frame when boosting some shots to 50k-100k equivalent - in this shot you can see that it is 50k with the curve changed to make it more obvious:

(this wasn't taken with the 50 1.0, I didnt change the "no lens" settings )

Please tell me, is this within the acceptable limits? On my pentax DSLR the noise was totally even across the entire frame even up to 512k equivalent. (Obviously unusable, but still even noise patterning). I am coming from shooting mostly b&w film, and some of the b&w shots I was getting at 12k-51k were a lot more usable than some of the prints/scans I end up making from pushed negatives. Luckily this bar is mostly invisible in straight b&w conversions, but it messes with colour tweaking for b&w converting (filter emulation).

This is independent of lens, orientation, electromagnetic environment, light leaks, or anything like that which I could test.

If you're going to abuse your RAW files like that, you should expect them to look like crap. The X-Trans III files color fidelity holds up pretty well to around ISO 25,600 or so if you process carefully. much higher than that and the color will be somewhat unpredictable. My X-T2 and X-T20 (same sensor) will tend to go green in the shadows if you get ridiculous with the brightness gain (correctable), but I don't get any significant purple glow like in your example, so I don't know what's going on there. Was your sensor hot? Did you take lots of images prior to this one? Maybe your camera does have an issue, maybe it doesn't How does it look at ISO 12800 with normal processing?

I shoot in very low light with some regularity and it has to be VERY dark indeed for me to have to go to black and white with my X-Trans III cameras. I have not encountered a better performing APS-C sensor at high ISOs than this one and have always been able to manage a halfway decent looking image in almost any light. I think you may have some unrealistic expectations about what an APS-C sensor can do, If you've got no photons to work with, photography will become problematic with any camera.

Erik Baumgartner wrote:

If you're going to abuse your RAW files like that, you should expect them to look like crap. The X-Trans III files color fidelity holds up pretty well to around ISO 25,600 or so if you process carefully. much higher than that and the color will be somewhat unpredictable. My X-T2 and X-T20 (same sensor) will tend to go green in the shadows if you get ridiculous with the brightness gain (correctable), but I don't get any significant purple glow like in your example, so I don't know what's going on there. Was your sensor hot? Did you take lots of images prior to this one? Maybe your camera does have an issue, maybe it doesn't How does it look at ISO 12800 with normal processing?

I shoot in very low light with some regularity and it has to be VERY dark indeed for me to have to go to black and white with my X-Trans III cameras. I have not encountered a better performing APS-C sensor at high ISOs than this one and have always been able to manage a halfway decent looking image in almost any light. I think you may have some unrealistic expectations about what an APS-C sensor can do, If you've got no photons to work with, photography will become problematic with any camera.

Thanks for the reply,

I of course understand that the sensor isn't meant to do this, and that realistically it is asking a lot - but on other cameras I've owned I've never experienced this kind of... whatever it is. They have had a lot of noise at these ISOs but never as uneven an strange to me as this is. I'm fully expecting noise that makes the image totally unusable by around 50-100k, but the fact is that almost all of the frame is not unusable at those ISOs (depending where you draw the line of usable), except for the purple bar at the bottom. It is interesting that your cameras do not do this. Regularly at 12800 you can't see the purple from what I can tell, but with some slight exposure correction in curves and white balance tweaks its entirely possible that it becomes visible.

If you have a second and wanted to do me a favour, would it be possible for you to take a picture with the body cap on at 12800 and then add a couple of stops in lightroom? This is what I get when I crank it +5 (clearly not desperately unreasonable based on my OP. Note different AF patch has different noise pattern):

Thanks

Monicaa wrote:

Erik Baumgartner wrote:

If you're going to abuse your RAW files like that, you should expect them to look like crap. The X-Trans III files color fidelity holds up pretty well to around ISO 25,600 or so if you process carefully. much higher than that and the color will be somewhat unpredictable. My X-T2 and X-T20 (same sensor) will tend to go green in the shadows if you get ridiculous with the brightness gain (correctable), but I don't get any significant purple glow like in your example, so I don't know what's going on there. Was your sensor hot? Did you take lots of images prior to this one? Maybe your camera does have an issue, maybe it doesn't How does it look at ISO 12800 with normal processing?

I shoot in very low light with some regularity and it has to be VERY dark indeed for me to have to go to black and white with my X-Trans III cameras. I have not encountered a better performing APS-C sensor at high ISOs than this one and have always been able to manage a halfway decent looking image in almost any light. I think you may have some unrealistic expectations about what an APS-C sensor can do, If you've got no photons to work with, photography will become problematic with any camera.

Thanks for the reply,

I of course understand that the sensor isn't meant to do this, and that realistically it is asking a lot - but on other cameras I've owned I've never experienced this kind of... whatever it is. They have had a lot of noise at these ISOs but never as uneven an strange to me as this is. I'm fully expecting noise that makes the image totally unusable by around 50-100k, but the fact is that almost all of the frame is not unusable at those ISOs (depending where you draw the line of usable), except for the purple bar at the bottom. It is interesting that your cameras do not do this. Regularly at 12800 you can't see the purple from what I can tell, but with some slight exposure correction in curves and white balance tweaks its entirely possible that it becomes visible.

If you have a second and wanted to do me a favour, would it be possible for you to take a picture with the body cap on at 12800 and then add a couple of stops in lightroom? This is what I get when I crank it +5 (clearly not desperately unreasonable based on my OP. Note different AF patch has different noise pattern):

Thanks

I don't have a camera at hand, but this was shot during a play when all the lights went out, ISO 12800 pushed 5 stops in Lightroom. You can make out the AF patch, but no purple at the bottom. The sensor would have been good and toasty here.

ISO 12800 pushed 5 stops is going to be several bridges too far for any kind of reasonably decent result. This is a more reasonable ISO 12800 pushed 1 stop in post (ISO 25600, Lightroom NR only), and is about as far as I go before B&W starts to look like a great option.

Monicaa wrote:

Just got a new X-E3 this evening and took it for a walk to shoot some snaps at iso 12800 - as you do

wow. Straight to 12,800 ... are you anticipating a lot of shooting in low light using the X-E3 ?

I am coming from shooting mostly b&w film, and some of the b&w shots I was getting at 12k-51k were a lot more usable than some of the prints/scans I end up making from pushed negatives

You're kidding, right? I also did BW in the Film days and even Iso 3200 had more noise than this. You take a shot at Iso 12800, raise two stops, raise contrast to 100 and lift tone curve. This corresponds let's say Iso 100.000. This wouldn't even be possible with film. And last but not least this isn't BW, right? Just convert it to BW and nobody will notice it.

Heinz wrote:

I am coming from shooting mostly b&w film, and some of the b&w shots I was getting at 12k-51k were a lot more usable than some of the prints/scans I end up making from pushed negatives

You're kidding, right? I also did BW in the Film days and even Iso 3200 had more noise than this. You take a shot at Iso 12800, raise two stops, raise contrast to 100 and lift tone curve. This corresponds let's say Iso 100.000. This wouldn't even be possible with film. And last but not least this isn't BW, right? Just convert it to BW and nobody will notice it.

Perhaps you misunderstand, I meant that the shots at 12k-51k on the digi were equivalent to the shots I was getting from some higher speed films (and even some lower speed ones!). I understand that it is significantly less visible with a bw conversion, but if I play with colours to similate colour filters then that can really make it much more visible on the bw image.

I'm really just trying to determine if my sensor is broken or not...

Ok, so this is very very interesting- this issue appears to only occur when using the mechanical shutter. When using the ES the noise is totally clean and even, as it should be. See

Monicaa wrote:

Ok, so this is very very interesting- this issue appears to only occur when using the mechanical shutter. When using the ES the noise is totally clean and even, as it should be. See

Hmm, my example was  likely with the ES. Will post an MS lens cap example later today.

Erik Baumgartner wrote:

Monicaa wrote:

Ok, so this is very very interesting- this issue appears to only occur when using the mechanical shutter. When using the ES the noise is totally clean and even, as it should be. See

Hmm, my example was likely with the ES. Will post an MS lens cap example later today.

Thank you, I appreciate that

Monicaa wrote:

Ok, so this is very very interesting- this issue appears to only occur when using the mechanical shutter. When using the ES the noise is totally clean and even, as it should be. See

Hmm, in both cases there seems to be a green rectangle visible in the middle - may I suppose that is caused by phase detection pixels...?

Here, I`ve boosted contrast to make it more visible:

boogisha wrote:

Monicaa wrote:

Ok, so this is very very interesting- this issue appears to only occur when using the mechanical shutter. When using the ES the noise is totally clean and even, as it should be. See

Hmm, in both cases there seems to be a green rectangle visible in the middle - may I suppose that is caused by phase detection pixels...?

Here, I`ve boosted contrast to make it more visible:

Yes, the green rectangle is due to the different AF sensors in the center of the frame, I realise that there is no way that this will disappear, but in real world shots I can't make the center square visible at all. The discolouration at the bottom when using the mechanical shutter is plainly visible in real world shots when pushing the raw files well within their actual limits.

Monicaa wrote:

Erik Baumgartner wrote:

Monicaa wrote:

Ok, so this is very very interesting- this issue appears to only occur when using the mechanical shutter. When using the ES the noise is totally clean and even, as it should be. See

Hmm, my example was likely with the ES. Will post an MS lens cap example later today.

Thank you, I appreciate that

These are both with the X-T20 at ISO 12800 pushed five stops AND with the Blacks at +100 in Lightroom. I also get a purple glow along the bottom only with the mechanical shutter (really not sure why). I can say though, having used this sensor for a number of years in many very low light situations, that it has never been a problem in any real world image making.

Mechanical Shutter, ISO 12,800 pushed +5 stops (ISO 409,600) AND with the Blacks at +100

Electronic Shutter, ISO 12,800 pushed +5 stops (ISO 409,600) AND with the Blacks at +100

...and with a couple of graduated filter tweaks your ISO 409,600 shots can still look awesome. Apparently the MS doesn't quite manage perfectly uniform exposure

The top MS shot with some added correction

Erik Baumgartner wrote:

Monicaa wrote:

Erik Baumgartner wrote:

Monicaa wrote:

Ok, so this is very very interesting- this issue appears to only occur when using the mechanical shutter. When using the ES the noise is totally clean and even, as it should be. See

Hmm, my example was likely with the ES. Will post an MS lens cap example later today.

Thank you, I appreciate that

These are both with the X-T20 at ISO 12800 pushed five stops AND with the Blacks at +100 in Lightroom. I also get a purple glow along the bottom only with the mechanical shutter (really not sure why). I can say though, having used this sensor for a number of years in many very low light situations, that it has never been a problem in any real world image making.

Mechanical Shutter, ISO 12,800 pushed +5 stops (ISO 409,600) AND with the Blacks at +100

Electronic Shutter, ISO 12,800 pushed +5 stops (ISO 409,600) AND with the Blacks at +100

...and with a couple of graduated filter tweaks your ISO 409,600 shots can still look awesome. Apparently the MS doesn't quite manage perfectly uniform exposure

The top MS shot with some added correction

Very interesting thank you, I am glad that my camera is seemingly not defective, but its a shame I can't return it and get one that doesnt do this

I appreciate you taking the time to do the test for me!

I wonder what it is about the mechanical shutter that makes this happen, and if it happens on newer cameras? I will ask my friend who shoots an XT4 if they can test this too.

Monicaa wrote:

Erik Baumgartner wrote:

Monicaa wrote:

Erik Baumgartner wrote:

Monicaa wrote:

Ok, so this is very very interesting- this issue appears to only occur when using the mechanical shutter. When using the ES the noise is totally clean and even, as it should be. See

Hmm, my example was likely with the ES. Will post an MS lens cap example later today.

Thank you, I appreciate that

These are both with the X-T20 at ISO 12800 pushed five stops AND with the Blacks at +100 in Lightroom. I also get a purple glow along the bottom only with the mechanical shutter (really not sure why). I can say though, having used this sensor for a number of years in many very low light situations, that it has never been a problem in any real world image making.

Mechanical Shutter, ISO 12,800 pushed +5 stops (ISO 409,600) AND with the Blacks at +100

Electronic Shutter, ISO 12,800 pushed +5 stops (ISO 409,600) AND with the Blacks at +100

...and with a couple of graduated filter tweaks your ISO 409,600 shots can still look awesome. Apparently the MS doesn't quite manage perfectly uniform exposure

The top MS shot with some added correction

Very interesting thank you, I am glad that my camera is seemingly not defective, but its a shame I can't return it and get one that doesnt do this

I appreciate you taking the time to do the test for me!

I wonder what it is about the mechanical shutter that makes this happen, and if it happens on newer cameras? I will ask my friend who shoots an XT4 if they can test this too.

These are from the X100V, same settings as before. Noisier, but definitely more uniform...

X100V, Mechanical Shutter

X100V, Electronic Shutter

I'd be curious to see the same test from the X-T3, Although it uses the same sensor, the X100V seems to me to be the slightly better high ISO performer (I owned an X-T3 briefly). Despite the more uniform noise here (at ISO 409,600 equiv. + the black slider at 100)), with less noise and fewer artifacts in real world images, I still give the older X-Trans III sensor the edge for "normal" high ISO shooting