Panning 'mud', reasons why and firmware update

commander did you check on the higher ISO theory? The other thread stopped with my last post. A lot of the prosumer video cameras are rated ISO 320 or higher. Did you check to see if getting the ISO higher made any difference on the panning "mud"?
 
An interesting result here.

1/30 F4 ISO 100:
Pretty bad...
iso100.jpg


1/30 F4 ISO 320:
This seemed better...
iso320.jpg


1/30 F4 ISO 800:
Oh dear, then the theory falls down...
iso800.jpg


1/125 F4 ISO 320:
Underexposed but fine once again at ISO 320...
iso320fast.jpg


1/125 F4 ISO 800:
But if it's higher ISO which solves the problem, this tells me otherwise...
iso800fast.jpg


What's interesting here is why the dark one on the fast shutter is so much better than the lighter one shot at ISO 800.

It seems the underexposure has cured the problem. Another test which adds weight to the 'detail' theory.

The more detail in the shot, the more corruption when panning...
 
Last edited:
Was this shot handheld? 1/30 will definitely look muddy handheld with the amount of motion your blur your introducing.

Just curious, can you also shoot this test at 1/60 and only slightly underexposing the footage. And while you're at it try this with MJPEG.
 
Isaac, I think all shots were purposely done this way to help us see the type of mud we get under similar panning situations.
 
Yes, but it actually might be more useful to see the actual shots in motion. Commander can you post the native MTS files? And Ian, we're looking at a downressed jpg file. This was never meant to be viewed as a still frame. In motion this might not even be noticeable.

Is this useful? It's more useful to me to see the original file along with posted settings to make any type of judgment.
 
commanderspike, thanks for posting. May 320 be the treshold? 1/50 is the closest to 1/48. It would be good to test panning there and see what happens. If it works better there we might have an answer. Thanks for the testing again.
 
Hmm... a very interesting test. It sounds like a an ISO threshold test is in order: Do the same pan with the ISO incremented by 100 each time (use same shutter speed for consistent motion blur). This way we could see the potential 'sweet spot' for ISO vs codec mud. I'd love to see a test like that. Nice work!
 
I will test again, especially the issue with ISO 320 being so faultless... interesting.

But it will have to wait until at least Sunday, as away on a trip tomorrow. I do have the GH1 with me though, so will be shooting footage.

If any other lucky GH1 owner would like to step in and do a similar test, be my guest! I reckon we can crack this... or at least confirm 100% to Pana that their camera definitely has this problem.

And yes - the purpose of this test is to try and make the problem occur. The day it doesn't happen is the day we find out why it happens in the first place.

Gladly, the decent footage I meant to look beautiful does look great with the GH1. It remains a fantastically promising video camera.
 
Someone with a camera needs to another test. Besides the break the codec test we have not yet seen a work within the limitations and light a scene test. How good can this camera look when shooting with some care and attention involved? Anyone want to lend me one for a weekend? I'll light a scene and shoot it and see just how good it can look. :)
 
Isaac_Brody said:
Someone with a camera needs to another test. Besides the break the codec test we have not yet seen a work within the limitations and light a scene test. How good can this camera look when shooting with some care and attention involved? Anyone want to lend me one for a weekend? I'll light a scene and shoot it and see just how good it can look. :)
Click to expand...

I was very impressed with Kholi's Heineken imagery but sadly it had to be taken off. With odd lighting and fog I felt it showed it's capabilities quite well in professional hands where care and attention is paramount. Granted there was no chance at frame by frame analysis which I'm sure is what you're after.
 
Nah, I'm not interested in frame by frame. Breaking the codec is easy enough to do, but I want to see what's possible when a little lighting and narrative control is implemented.
 
Yeh, as well as making it look deliberately bad for tests, it would be good for morale and inspiring to see more deliberately good stuff too! Maybe Kholi's prepared piece might be out soon?

I was wondering about the nature of the codec's VBR. Is there some way to trick the camera into putting maximum datarate into a shot. I thought: if there's a static tripod shot, the camera may provide just enough to cover the lack of image changes within the GOP. If the same largely static shot were filmed with the camera on a shoulder mount, I imagine the camera would be forced to provide a higher datarate? Even when you try to hold the camera still the inevitable slight movement means the picture will be changed enough to invoke a greater datarate to accomodate it. Obviously, extra datarate is needed compared to the tripod equivalent (because the image really is changing more), but my question is, would more than enough be forced as a result, providing a possible net gain, keeping the footage closer to the maximum?
 
Isaac_Brody said:
Someone with a camera needs to another test. Besides the break the codec test we have not yet seen a work within the limitations and light a scene test. How good can this camera look when shooting with some care and attention involved? Anyone want to lend me one for a weekend? I'll light a scene and shoot it and see just how good it can look. :)
Click to expand...

Isaac_Brody said:
Nah, I'm not interested in frame by frame. Breaking the codec is easy enough to do, but I want to see what's possible when a little lighting and narrative control is implemented.
Click to expand...

John Caballero said:
We know about the bad lets see now how good.
Click to expand...

Barry_Green said:
.... It's not perfect, there are issues, but there's nothing else within a hundred miles of it, so -- how can we make it work? That's the quest. And I think it's likely perfectly workable. Likely enough that I think we're gonna try soon.
Click to expand...


:lipsrseal

...

:)


_
 
commanderspike said:
An interesting result here.




1/125 F4 ISO 800:
But if it's higher ISO which solves the problem, this tells me otherwise...
iso800fast.jpg


.
Click to expand...

Reminds me of a Van Gough painting!


We could see Dreams II made with the GH1...... The original film by the extraordinaire, Akira Kurosawa. Martin Scorsese could reprise his role as Vincent Van Gogh.......

:-)




.
Michael Pappas
http://www.pbase.com/Arrfilms


Anamorphic DSLR Lens Test Images... links:
http://www.dvxuser.com/V6/showpost.php?p=1638289&postcount=97
http://www.dvxuser.com/V6/showpost.php?p=1637705&postcount=84

Facebook:
http://www.facebook.com/people/Michael-Pappas/573417404
Myspace:
http://www.Myspace.com/PappasArts

Arrfilms@hotmail.com
http://www.PappasArts.com
CONTACT VIA AOL INSTANT MESSENGER
AT { PAPPASARTS2 }

.
 
More ISO tests, but that's exactly what I suggested someone do. Commander, if you've yet to activate the 1/3 incremental ISO, try that then start at the bottom.

Try 160, 320, 640, 1250

If these are the native ISO's and the rest are software driven we might have solved the issue.

Most of my stuff has been at 320 ISO, because I'm using faster glass and ND's as well. Try not to go over 320 in general.

and I know for a fact everything on the Heineken job was 320 ISO.
 
You must log in or register to reply here.
Back
Top