AG-HPX371E jello issue....

I know of one way to fix this:) Stop demanding ultra high resolution, ultra sharp, noise free cameras with small sensors! Putting low dynamic range aside, this results in heavy preprocessing which has an impact on texture. Low noise is great but the way to judge noise is by texture quality in a dark shade. If it's flat and it looks like plastic, where is the advantage? In a dB noise metric? Nobody shoots lego bricks for a living and nobody sets his/her rates by the S/N performance of the camera. Companies are competing too heavily in the noise reduction department already.
 
Yes! I think that the more luminance have increased the soft detail in the picture, and the
SD format compress the frame so you can't notice the issue at a first look. But... you don't
know how happy i'm today with my new test footage!!! I've to say THANK YOU to Henry Epstein! Now i can see how the 1/50 shutter on 1080 25pn does his job very well!!! But i've tried more... 1/500 with gain -3db ,ND 4 and iris full open at 19.00PM... pure beauty! And all with detail level of 0! Great! That's what i need in my shoots! By the way.. where can i buy this new hair product? I need it! More noise is much,much,much better then a tree that flies around without reason in the frame!
 
Roberto Tafuro said:
But... you don't know how happy i'm today with my new test footage!!! I've to say THANK YOU to Henry Epstein! Now i can see how the 1/50 shutter on 1080 25pn does his job very well!!! But i've tried more... 1/500 with gain -3db ,ND 4 and iris full open at 19.00PM... pure beauty! And all with detail level of 0! Great! That's what i need in my shoots! By the way.. where can i buy this new hair product? I need it! More noise is much,much,much better then a tree that flies around without reason in the frame!
Click to expand...

Roberto,
Are your new tests on the 371 or 301?
I can't be happier with my new 370!
I've been trying harder to find those some weird artifacts at the correct shutter speed- not at the camera default shutter speed settings, and so far this new camera rocks!
Glad to hear your comments. Great for you, man!
Best,
Henry
 
Hi Henry! Thank you so much! Well... with my new 301! I've changed the 371 with the 301 yesterday. I have to say that i like it very much! Your settings and tips make me very happy today! VERY HAPPY! It's the most important thing in this work: be happy at work! So... i'm happy that your new 371 rocks!
Roberto
 
Roberto,
So happy that you found the way to go with your new 301... The NTSC 300 is rock solid too!
The 370 works wonderful... super clean at 0db and with no evident noise even at +3db...
Regarding sensitivity, have to admit is almost 1.25 stops faster that my 300.
A quick low-end test showed up 640 ASA @ 30pn and almost near 800 ASA @ 24pn... unbelievable!
Once again, happy future and the best for you!
H.
 
Otis Grapsas said:
This is not a CMOS issue. An horizontal pixel line is exposed at the same time. It's impossible that rolling shutter can produce this effect.

My guess is that it's a bug in the preprocessing, it could be in some form of motion adaptive noise reduction. This is a common algorithm, the pixel values of different frames are compared and the algorithm uses current and past values in order to discard what it considers to be noise. Small changes are considered noise and large changes are considered true motion. It can freeze motion in some parts of the frame for a while if the change is detected as noise. The error is usualy only noticable in full resolution. On a scaled format like SD it can be masked because it's less than a pixel after downsampling.
Click to expand...

That is an interesting theory and could certainly explain it. I think I might have mis-spoken on what settings we had when we saw it. It was with the latest firmware, but DVCProHD100 1080/24PA. I remember now that we could not confirm with our post house that they could handle the AVC-I formats yet. We noticed the shifting backgrounds in-camera and in the post suite on their calibrated monitor. It was enough that the editor asked what the hell it was. We'd forgotten about it, but all watched it several times over on more than one monitor.

How can pre-processing be minimized? Again, the HPX300/370 is billed as a [the only?] 10-bit native 1920 x 1080 under $XX,XXX dollars. A truly remarkable bit of kit for the money. Still, the questions remain about just how well the platform was implemented.
 
This kind of aliasing pops up all over the place with different kinds of cameras. Check out the irish musical movie "ONCE". On the opening scene, the graffiti wall behind Hansard dances as the camera moves. Same stuff as Roberto's moon.

According to the director John Carney: "We shot it on three of those little Sony HDVs. I think they’re called 325’s or something"


Jason
 
If Sony disabled noise reduction on the EX1 and used similar processing to lets say a Z1, you would discover that its sensitivity and signal per noise is 2 stops inferior to the Z1. They can't afford to allow the user to disable the entire preprocessing. The user can use a camera at its lowest gain setting and with all noise reduction controls set to off and hope for the best.

In the HPX case, it must be a bug. It looks a little too aggresive. A firmware update will fix it.

Noise reduction and compression is typically engineered on static shots that contain flat color areas and parts of resolution test charts. If you ask most video engineers they do not even care about texture, some are not even aware of the term. It's all about detail. Detail sells camcorders. But detail is high contrast changes and texture is low contrast changes. Under this perspective, the processing (and compression) will look great when there is hard detail but it will quickly deteriorate texture or even artifact, especially at low exposure levels, at Zone IV, III, II, I.

On online clips, the online compression could be the problem.
 
You must log in or register to reply here.
Back
Top