Old Sensor vs New Sensor - direct test

N

nathankw

Well-known member
I know that debate has long been raging about whether newer GH1 bodies have less fixed pattern noise in low light. I recently got a second body so decided it was time for a direct test.

I've tried to take an identical shot using the same kit lens. F4, 800 ISO, 1/50th shutter. Both cameras hacked. Deliberately underexposed to get horrible banding.

The old body was bought as soon as the GH1 was available in the UK (SN WE9E...)
The new body was bought a couple of months ago, still hackable - (SN WE0F...)

And the result...
Here are the MTS files - take a look.
If you want to look without knowing which is which, turn the sound off.

http://dl.dropbox.com/u/8356280/00003.MTS
http://dl.dropbox.com/u/8356280/00004.MTS


I'd say there's really very little difference. This surprised me, since I'd always felt I had worse banding than many others - but this seems to indicate not. However, both my bodies were made in China so it's still possible the Japanese models are better. Hopefully someone else can do that comparison.

Would love to know what other think.
 
nathankw said:
I know that debate has long been raging about whether newer GH1 bodies have less fixed pattern noise in low light. I recently got a second body so decided it was time for a direct test.

I've tried to take an identical shot using the same kit lens. F4, 800 ISO, 1/50th shutter. Both cameras hacked. Deliberately underexposed to get horrible banding.

The old body was bought as soon as the GH1 was available in the UK (SN WE9E...)
The new body was bought a couple of months ago, still hackable - (SN WE0F...)

And the result...
Here are the MTS files - take a look.
If you want to look without knowing which is which, turn the sound off.

http://dl.dropbox.com/u/8356280/00003.MTS
http://dl.dropbox.com/u/8356280/00004.MTS


I'd say there's really very little difference. This surprised me, since I'd always felt I had worse banding than many others - but this seems to indicate not. However, both my bodies were made in China so it's still possible the Japanese models are better. Hopefully someone else can do that comparison.

Would love to know what other think.
Click to expand...

I rented a GH1 in Janauary of this year from Lens Rentals. Couldn't find a trace of FPN
 
Psynema said:
I rented a GH1 in Janauary of this year from Lens Rentals. Couldn't find a trace of FPN
Click to expand...

You probably didn't shoot footage that was 800% less exposed than it should have been. Which is what you have to do to get banding (that is before you bring it up with the ISO.)
 
Ben_B said:
You probably didn't shoot footage that was 800% less exposed than it should have been. Which is what you have to do to get banding (that is before you bring it up with the ISO.)
Click to expand...

Shot a wedding with it actually maxed out...some footage with the kit lens.
 
Whilst not in any way questioning your experience, the problem has been we've had lots of people saying "I don't get any banding" and others saying it's terrible but no real way to know if they're doing something different, have different standards, or genuinely different sensors.

I was a believer in the difference because I felt the banding is pretty bad and yet others said, like things like "Couldn't find a trace of FPN" so I figured my early model camera must be particularly bad.
But my new body most definitely does have FPN as well. So either I've been unlucky twice (possible but unlikely) or something else is going on.

Don't suppose you noticed whether the one you rented as a made in japan model?
 
I have also shot with the kit lens and the pancake lens maxed out at like 1250 iso and don't seem to get banding anywhere... while there is a bunch of noise..... it's usable
 
Mine was bought new in February. I get banding on the parts of the screen that are way underexposed at ISO1600. At 800 there is some, but it's not that noticeable.

One thing that might account for the differences people are seeing is white balance setting? On mine at least, I've noticed that when WB is set to tungsten, the banding is a little more prominent, presumably because of the noisier blue channel.

I have to do more testing to confirm, though.
 
I also did a direct comparison, and the new one is Made in Japan (manufactured Dec 09). Old one Made in China (manufactured June 09).

Did a super underexposed torture test. FPN and banding were identical, right down to the worst lines being in the exact same spots.
However, hot pixel noise on long exposure photo shots was significantly better on the new one. Don't know why. Regular sensor noise, and FPN as noted, were the same.

I also have an idea as to what's going on with the supposed "burn-in" reducing FPN that a lot of people talk about. I think it is software learned correction. If you do this sort of torture test that I did, the same amount of really bad FPN / banding appears on the LCD screen before you start shooting. However, once you start shooting something (especially AVCHD), it magically dissapears.... and it seems to get better and better at this over time (the new Japan one I have doesn't have quite so marked a difference between preview and shooting noise as my older well-used one did, but I think this will change in time). You have to look harder on your computer monitor at larger size to see the FPN on the file than it appeared in the LCD before shooting.
 
Last edited:
I'm sure I'm not an expert on this issue but I only really noticed the banding when I first got the cam, and turned it on in "auto" mode in a dim room. Loads o' banding, inviting instant panic. Then I set it to manual, and most of it was gone, so I imagine on certain settings it is adjusting something, like the ISO or gain.
 
I took mine into Samy's and shot it against a newer one they had, same white box, same ISO, same everything. Not a *huge* difference, but there was enough of one to piss me off.
 
well as can be seen it's all over the place. either everyone has different standards on how bad banding is or the sensors are just all over the place with no real QC
 
Trucci said:
I have yet to see a test that clearly shows the difference between sensors.
Click to expand...

Exactly.
My test shows (to me at least - has anyone else looked at the files) - very little difference between camera bought a year apart. But of couse that doesn't prove that all cameras are the same - you can't prove a negative.

We need someone who can provide a similar test - two shots in exactly the same condition - that shows a clear difference. Until then, the situation will still be unclear.
 
I think a lot of the problem lies with the older cameras being used a lot. The "banding" seems to diminish with time.

FPN is something different. FPN cannot go away, that's part of the sensor design.
 
Here's a question for people claiming to see a difference: Is the difference apparent on the GH1's LCD *before* hitting the record button, or only in the recorded file?

The GH1 really seems to have some magic going on where it will kick out a lot of the FPN upon recording (or even shooting a RAW still image), but where you can really see all the raw sensor noise on the LCD before you start shooting.
This makes me wonder if the difference people are noting is actually in the noise reduction software and not the sensor itself.
The way to tell is just to only look at the LCD before recording, and not the recorded files, and see how bad the FPN is on the LCD.
 
Svart said:
I think a lot of the problem lies with the older cameras being used a lot. The "banding" seems to diminish with time.

FPN is something different. FPN cannot go away, that's part of the sensor design.
Click to expand...

Could explain what you mean by "banding" vs "FPN". I assume the mainly vertical lines on the clips posted above are FPN?

I'm not sure the diminishing with time has ever really been proven either. Again - we need evidence.
 
dasbin said:
Here's a question for people claiming to see a difference: Is the difference apparent on the GH1's LCD *before* hitting the record button, or only in the recorded file?

The GH1 really seems to have some magic going on where it will kick out a lot of the FPN upon recording (or even shooting a RAW still image), but where you can really see all the raw sensor noise on the LCD before you start shooting.
This makes me wonder if the difference people are noting is actually in the noise reduction software and not the sensor itself.
The way to tell is just to only look at the LCD before recording, and not the recorded files, and see how bad the FPN is on the LCD.
Click to expand...

Why? It's worse on the LCD before but the only practical thing to care about are the recorded files. Also, I think it just has something to do with how the preview is being generated when not recording.
 
Also, to the OP, can you post some still images in the original post as well? I'm on an EDGE connection here...but even if I wasn't downloading .mts files is a bit inconvenient, I'd wanna look at the images first. Thanks!
 
dasbin said:
Here's a question for people claiming to see a difference: Is the difference apparent on the GH1's LCD *before* hitting the record button, or only in the recorded file?

The GH1 really seems to have some magic going on where it will kick out a lot of the FPN upon recording (or even shooting a RAW still image), but where you can really see all the raw sensor noise on the LCD before you start shooting.
This makes me wonder if the difference people are noting is actually in the noise reduction software and not the sensor itself.
The way to tell is just to only look at the LCD before recording, and not the recorded files, and see how bad the FPN is on the LCD.
Click to expand...

when shooting stills display is ramped up to aid in better focusing and vision. it's not a true indication unless you are in preview mode. for me the LCD is too small to definitively see banding when recording video , i only see it in post. Only on the rare occasion that i shoot in almost complete darkness do I see a hint of banding on the LCD.
 
You must log in or register to reply here.
Back
Top