GH-1 GH-13 Fixed Pattern Noise: Breakthrough?

"However whatever is happening the results speak for themselves."

Exactly. Okay, maybe the camera mixmashes brightness, iso & what not, so? It's happening in the camera, and it's happening while the file is writted to the card. After all.... is there something year 2011 that you *can't* do in post? I mean, don't use it if it doesn't float your boat... to me, these results are fantastic. (And my map test had very black and clean shadows when the lighter was off, that speaks for the iso IMO, if it was just brightness it would look different). Also, if you compare it with my GH2 test, the colors are blown out & contrasty, in GH1 all the gradients & details are there and it looks fantastic.
 
I was soooo skeptical reading this post. Then I tried it, and my jaw literally dropped. Literally, as soon as I hit record, I was amazed. Works with my FD lens. More observations to come, pending further testing (namely, aperture priority)
 
CColton said:
I was soooo skeptical reading this post. Then I tried it, and my jaw literally dropped. Literally, as soon as I hit record, I was amazed. Works with my FD lens. More observations to come, pending further testing (namely, aperture priority)
Click to expand...

Yes, I'm quoting myself. After further testing, it's not quite as miraculous as I thought, but still noticeably helpful and worth the workflow change. In reply to many who asked earlier in the thread, this works just fine with both Panasonic and adapted lenses; no workflow change for my Panny lenses versus my FD prime.
 
Hmmmmmmm.... having a much harder time now after switching the cam to NTSC... this needs more testing... much more normal FPN banding on NTSC... suggestions on tests? Charts exluded... and no don't have any hot chicks to lit up with candles, sorry ;)

edit: ignore this for now, likely no difference between NTSC/PAL, i was thinking first that FPN was less on the cameras "native" format but if so still likely not a big deal
 
Last edited:
The difference I'm seeing is that noise is reduced somewhat, but the horizontal lines of noise dissipate, making the noise that resists seem more "natural."
 
I must say it seems to work... Thanks a lot for the tip!

Metering system in video mode seems to be bogus when one tries this, when in video mode it under-exposes severely and when recording video on 'photo modes' it exposes quite nicely (as FPN appears more easily in under-exposed shots this seems credible), AE lock seems to help as well ... but it could be that something like shutter speed is not kept, only way to tell is to make an frame-by-frame analysis to a spinning wheel...

 
Last edited:
zebarnabe, can you do me a favor post the average original data rate of each of the clips you shot above. Thanks.

R
 
The posted video was edited on Nero Vision, it might have lost some of its quality, but at 100% view it is barely noticeable (except perhaps during panning).

This is what one of my media player reports for the original MTS files:
S Mode, shutter at 1/50s, ISO set at 3200, overall bit rate: 21.9Mbps, Maximum bit rate: 34.0Mbps
S Mode, shutter at 1/50s, ISO set at 1600, overall bit rate: 23.4Mbps, Maximum bit rate: 34.0Mbps
S Mode, shutter at 1/50s, ISO set to Auto, overall bit rate: 26.2Mbps, Maximum bit rate: 34.0Mbps
Movie Mode, S setting, shutter at 1/50s, ISO set at 1600, EV at +3, overall bit rate: 27.1Mbps, Maximum bit rate: 34.0Mbps
Movie Mode, S setting, shutter at 1/50s, ISO set to Auto, EV at +3, overall bit rate: 26.8Mbps, Maximum bit rate: 34.0Mbps

I had to set EV to +3 or else it would be too dark. The higher bit rate on movie mode could be related with noise being recorded, thus increasing the amount of 'detail' perceived by the codec.

There is a short 'intro' to each movie, where I put a piece of paper in front of the lens, with the settings it is running. This affected the overall bit rate as an average on all video, the differences for the shown clips aren't so big, using VLC to check the real amount of video bit rate for the 'same' (or close) footage of each movie, shown that the ones in Movie Mode had around 1Mbps more (26Mbps vs 27Mbps). The bit rate when I put the paper with the settings in front of the lens drops to 11~15Mbps.

Later it occurred me that it would have been good to did it in Movie Mode with a shutter of 1/30s - sadly assuring the same light conditions might prove difficult. I could, however, repeat all the shots with artificial light on a windowless room.

My major trouble is to upload the video, it took my 2 hours for the one you've seen previously (actually 3 hours, in the first hour upload failed a couple of times)...

Hope it helps taking conclusions :/
 
CColton,

I shot that with the 20mm 1.7 pancake lens. I honestly can't tell you the ISO since it was set to auto. The F-Stop was at 1.7 for sure though. When you press the shutter button, it shows aperture at 1.7 and ISO at 400, but there's no way the ISO could be accurate. Also, it's shooting at 1/30 shutter also, if the display is right.

All in all, for a cheap camera, that so far is not hackable I'm pretty pleased with the results. I've tried it with my 24mm 2.8 FD lens as well, and it STILL looks pretty good.

CColton said:
What ISO were these shot at? Or since the ISO is auto, with what lens (and f-stop) and what do you think the ISO would've been?
Click to expand...
 
CColton,

I shot that video with the 20mm 1.7 pancake using the settings in the first post. Aperture showed at 1.7 and shutter (when video is capturing) is at 1/30. The ISO is on AUTO, but when you half press the shutter before initiating video it shows as 400... There's no way that it's accurate though. It was pretty dark in there.

Can't really say what the ISO would have been, but pretty high to say the least. I'm very pleased with the performance of the stock gh1 using these settings and it makes it a keeper as far as I'm concerned. Hopefully it can be hacked in the future though!

Tony
 
Some snapshots from another videos I've done...

Movie Mode, S 1/30s ISO 1600:
https://picasaweb.google.com/lh/photo/KX3QIULh1ILdfaDRCJUstFULI05CfiF65SjGudwuyr4?feat=directlink
Custom 1, Shutter priority, S 1/30s ISO Auto:
https://picasaweb.google.com/lh/photo/dQIpUITFbGyony_LgjeATVULI05CfiF65SjGudwuyr4?feat=directlink

The biggest point is the underexposure that exists in movie mode (I did not use EV compensation), that alone allows for a more noticeable FPN... As the 1/30s at ISO 1600 in movie mode for a given aperture defines the lowest light level you can use in movie mode, the camera must be boosting sensitivity in some way: Using ISO 3200 when needed would be a valid guess.
 
Thanks for those bitrate numbers zabarnabe, the results were different than mine. I was hoping to see significantly higher bitrates on the brighter clips like with my results.

Oh well, doesn't matter, it's still a very usable setting with the GH1 and if Vitaliy can enable 3200 thru the menu in Movie mode, even better.
 
I just used this setting extensively for shooting a banquet where there was practically no light (shot entirely at ISO 1600 using f1.7 lenses). It performed very well. I'd recommend using Standard film mode with contrast all the way up; the noise is still present but is less noticeable than any other 1600 shooting I've done with my GH1.
 
I just wonder what the camera does when messing with EV compensation when in movie mode... Only when you go from +2 to +3 in low light situation that it boosts the exposure... It doesn't look identical to what happens when using a custom mode to record video (where you don't even touch the EV comp. to get it a lot brighter), but in custom mode you can't quite boost it if the limit has been reached (meaning that EV or EV Lock does nothing if it's not possible).

The trick with the settings for the settings doesn't completely remove noise, however it might help quite a lot in certain situations.

I've been messing with the camera setting for a while ... I found one that was slightly underexposed, yet with almost no visible noise... sadly, since it didn't look that good on camera screen, I didn't keep the settings >_<'' I played with contrast, saturation, NR and sharpness aside of the standard ISO and EV compensation for both movie mode and custom one.

Overall, I found this trick very strange, most likely a glitch in camera firmware or a conscious design option... either way, finding what is best is always good :]

PS: If I find what settings did I got for the cleanest footage, I shall post again ;]
 
zebarnabe said:
PS: If I find what settings did I got for the cleanest footage, I shall post again ;]
Click to expand...

So far I've had the best results (which actually reduced the noise to tolerable levels) was the exact settings the Emery posted, with exposure compensation down 1/3-step, in Standard film mode with Contrast all the way up, saturation and sharpness at 0, and noise reduction all the way up.
 
Only downside i've found is that the bokeh breaks up similar like the GH2 does, like the bitrate drops or something... gradients gets squashed a bit. With EC.
 
You must log in or register to reply here.
Back
Top