PDA

View Full Version : Could the Hack reduce Fixed Pattern Noise?



emory
07-16-2010, 05:22 AM
One of the few issues I have with the GH1 is Fixed Pattern Noise or FPN. To be sure, there are workarounds to make it less obvious, but it remains an annoying fact of life with the GH1.

Using PHOTO mode with high ISOs and in low light, FPN often looks terrible in preview, but the finished shot is quite smooth, thanks to the processing (Venus Engine ?) performed on the file. In MOVIE mode, the video file also appears to be processed (when recording starts, FPN is not as obtrusive as in preview), but the processing is not as extensive or successful as in PHOTO mode.

My question: would it be possible to improve processing via the hack?

alignment1
07-16-2010, 08:55 PM
One of the few issues I have with the GH1 is Fixed Pattern Noise or FPN. To be sure, there are workarounds to make it less obvious, but it remains an annoying fact of life with the GH1.

Using PHOTO mode with high ISOs and in low light, FPN often looks terrible in preview, but the finished shot is quite smooth, thanks to the processing (Venus Engine ?) performed on the file. In MOVIE mode, the video file also appears to be processed (when recording starts, FPN is not as obtrusive as in preview), but the processing is not as extensive or successful as in PHOTO mode.

My question: would it be possible to improve processing via the hack?

If it's any correlation, I feel the FPN has lessened for me since the hack. Still around in certain shots, but seems less prevalent than before.

sammysammy
07-17-2010, 08:27 AM
for me thats the most improvement i see is in reduction of fpn compared to before when using higehr iso 800 and above ,banding to my eyes is also reduced .

emory
07-21-2010, 05:31 AM
I just hacked my GH1, and while I'm quite pleased with the much improved image quality, FPN is still evident.

I'm still wondering if a future hack version could address FPN, which to me as a layman appears to be more of a CODEC issue than an intrinsic problem with the sensor. Here's an example:

Corner of a room with structured white wall paper. The wall hit flatly by the light displays vertical FPN. The wall lit obliquely (displaying fine structure) shows no FPN. It makes no difference if the subject is white or dark: no structure in the subject = FPN; the slightest structure in the subject = NO FPN.

Maybe Vitaliy and his team could look into this? I think a good many of us would appreciate it.