commanderspike
Carbonite Member
That's incorrect I'm afraid. The limit is ISO 1600. The fixed pattern noise (known in the industry as FPN) is an aspect of CMOS technology. The GH1 has it more than most, it seems. But it's still a good sensor.
Low Light Streaks - Is my camera faulty?
- Thread starter nathankw
- Start date
That's incorrect I'm afraid. The limit is ISO 1600. The fixed pattern noise (known in the industry as FPN) is an aspect of CMOS technology. The GH1 has it more than most, it seems. But it's still a good sensor.
When you apply Exposure Correction in Shutter Speed mode, brightness is way more, than on ISO 1600 in M mode. I think it exposed at ISO 3200 in this case.
commanderspike
Carbonite Member
Wow you are right. This is incredible.
When in S mode, set the shutter to 30 and ISO to 1600 or auto. Dial the exposure compensation to maximum (2ev). In full manual (Creative Movie Mode, not the M stills mode on the dial) put the shutter at 30 and ISO at maximum (1600). Shoot both modes at the same F stop. I tried F1.4 with my Leica.
The difference is quite literally night and day. You can see the brightness get amped up, a few moments after recording begins. It seems to adjust ISO and shutter speed on the fly - which is expected in S mode as this is a semi-automatic mode. But it does indeed seem to boost the sensor to something like ISO 3200.
Going out now with a fast lens to see how far it can be pushed. I don't see too much extra noise, but is at least another 2 or 3 stops of brightness.
Great find.
When in S mode, set the shutter to 30 and ISO to 1600 or auto. Dial the exposure compensation to maximum (2ev). In full manual (Creative Movie Mode, not the M stills mode on the dial) put the shutter at 30 and ISO at maximum (1600). Shoot both modes at the same F stop. I tried F1.4 with my Leica.
The difference is quite literally night and day. You can see the brightness get amped up, a few moments after recording begins. It seems to adjust ISO and shutter speed on the fly - which is expected in S mode as this is a semi-automatic mode. But it does indeed seem to boost the sensor to something like ISO 3200.
Going out now with a fast lens to see how far it can be pushed. I don't see too much extra noise, but is at least another 2 or 3 stops of brightness.
Great find.
If you apply exposure compensation when in shutter priority mode, given that this is a fly-by-wire device, it's possible that it's changing ISO, aperture, or even shutter speed (which you'd expect to be fixed to the set value) to attempt to provide what you've asked for. Only some careful tests would reveal what goes on - eg examining DOF to check for aperture change, duplicated frames (which would happen if shutter speed was reduced below frame rate), or neither which would mean the ISO had varied.
Here, in CMPM, shutter priority, shutter=30th, ISO1600, if I increase exposure compensation to the max, a slow pan becomes slightly jerky compared to a slow pan with no exposure compensation. I'd say it's simply slowing down the shutter speed below that set.
Edit - no I'm wrong. I've just tried shooting a record turntable rotating with a small metal bar on top of it in a quite dark room - same settings as above. Upping the exposure compensation to +3 is not having any discernible effect on motion of the rotating bar, and stepping through reveals no duplicated frames. However, some horizontal banding becomes evident (or clearer) which would support the idea that ISO or gain is being increased above the maximum set.
But here's a weird thing - reducing exposure compensation to -3 does not darken the picture as you would expect it would under these conditions. It does if I point the GH1 at the daylight window. It's as if it's deciding for me that the picture is dark enough already when pointing at the turntable. Now come on, who is supposed to be in charge here?!
Edit - no I'm wrong. I've just tried shooting a record turntable rotating with a small metal bar on top of it in a quite dark room - same settings as above. Upping the exposure compensation to +3 is not having any discernible effect on motion of the rotating bar, and stepping through reveals no duplicated frames. However, some horizontal banding becomes evident (or clearer) which would support the idea that ISO or gain is being increased above the maximum set.
But here's a weird thing - reducing exposure compensation to -3 does not darken the picture as you would expect it would under these conditions. It does if I point the GH1 at the daylight window. It's as if it's deciding for me that the picture is dark enough already when pointing at the turntable. Now come on, who is supposed to be in charge here?!
Last edited:
Hi everyone,
I believe I read almost every other thread before I recieved my GH1 (as I´m a bit new in this forum that gave me a lot of much appreciated preparation and mind easing while waiting). And then after shooting the first clips I couldn´t believe what I was seeing: streaks allover and at any ISO!
As the only applicable solution seemed to try the burn-in I of course did it. One night, GH1 pointed at dark area in backyard, ISO 1600, display closed, economy off and good night.
And no I can shoot easily up to 1000 ISO and 1600 doesn´t look too bad either:
http://vimeo.com/6092438
I have to look around for a "before"-clip but I really hated those and most likely deleted them all.
I don´t know if the explanation of sensor calibration is a good one, but the burn in worked just fine for me. So thank you for that!
But I am pretty sure, that having a stronger wider lens helps too.
Regards!
Patrick
I believe I read almost every other thread before I recieved my GH1 (as I´m a bit new in this forum that gave me a lot of much appreciated preparation and mind easing while waiting). And then after shooting the first clips I couldn´t believe what I was seeing: streaks allover and at any ISO!
As the only applicable solution seemed to try the burn-in I of course did it. One night, GH1 pointed at dark area in backyard, ISO 1600, display closed, economy off and good night.
And no I can shoot easily up to 1000 ISO and 1600 doesn´t look too bad either:
http://vimeo.com/6092438
I have to look around for a "before"-clip but I really hated those and most likely deleted them all.
I don´t know if the explanation of sensor calibration is a good one, but the burn in worked just fine for me. So thank you for that!
But I am pretty sure, that having a stronger wider lens helps too.
Regards!
Patrick
Hi Patrick,
Good to hear that you cleared up your sensor. I had to send my back to the store because I tried the "burn in" method and didn't get nothing better.
They don't have others in stock so I don't know what to do. If I wait untill they get it or if I jump to Canon 5D Mark 2. Do you had a chance to see the Canon? I would like to know if it has the same FPN issue.
Best,
Ricardo
I haven't shot anything low-light again since then. I'll try another test and let you know. I did show what I had shot at that low-light to a director that wanted to see what the camera could do and he didn't seem to notice or be bothered by it. That's a good sign. Actually he was extremely impressed. We're using it for one his projects tomorrow in fact.
That is a cruel story! I never had my hands on a Canon or even first hand footage. Sorry!
sammysammy
Well-known member
Jack ..did you have a lens attached when you did the burn in over night or with out any lens?
BhambuNath
Well-known member
I think burn in has nothing to do with the streaks.
I got a brand new GH1 and the first thing I tried was if the streaks are a burn in issue or simply due to low available light.
What I've found out is that it might be because people were underexposing. When the light was low I could see the streaks on LCD, but as soon as I put in more light the streaks disappeared. It was with a brand new GH1 in Creative Movie Mode...
I got a brand new GH1 and the first thing I tried was if the streaks are a burn in issue or simply due to low available light.
What I've found out is that it might be because people were underexposing. When the light was low I could see the streaks on LCD, but as soon as I put in more light the streaks disappeared. It was with a brand new GH1 in Creative Movie Mode...
Barry_Green
Moderator
Streaks happen if you're 2.75 stops underexposed. Period. Happens in any ISO, any shutter speed, any lighting conditions, as long as the area you're looking at is 2.75 stops underexposed you'll get the streaks.
BhambuNath
Well-known member
I haven't seen the footage out of camera yet, but when only viewing on LCD I think the streaks are lot less when underexposed at Lower ISO.
Any idea how to deliberately underexpose for artistic reasons and avoid the streaks?
Barry_Green
Moderator
The streaking appears to me to be consistent regardless of ISO. If you're 2.75 stops underexposed, you'll get streaks. Now, footage is generally uglier the higher the ISO you use, but it doesn't appear to affect the streaks.
If you want darker footage without streaks, your best bet is to expose properly and then crush it down in post.
If you want darker footage without streaks, your best bet is to expose properly and then crush it down in post.
Martti Ekstrand
Veteran
Barry - I know you say the streaks are constant but... I didn't exactly do a methodic test of burning in and collecting sample images but I have from time to time pointed the GH1 at my latte coloured bedroom wall in fairly faint light and the streaks are clearly less visible now than when I got the camera. A guy over at the GH1 group in Vimeo was more determined and left the camera on for a whole night on external power and claims it helped.
http://vimeo.com/groups/gh1/forumthread:7666
http://vimeo.com/groups/gh1/forumthread:7666
Barry_Green
Moderator
Well, I'll tell you what -- I have a brand-new GH1 sitting in the box that just arrived.
I'll get to the bottom of the burn-in thing once and for all. I'll set up a controlled lighting circumstance that shows streaks, then leave it on for 24 hours, then shoot the same thing again, and post comparison photos. :thumbsup:
Now, would someone be so kind as to elucidate the exact procedure for "burn-in" that people are asserting works?
I'll get to the bottom of the burn-in thing once and for all. I'll set up a controlled lighting circumstance that shows streaks, then leave it on for 24 hours, then shoot the same thing again, and post comparison photos. :thumbsup:
Now, would someone be so kind as to elucidate the exact procedure for "burn-in" that people are asserting works?