What's the precise name of this phenomenon?

[Gunleik Groven]

Nope. Not until I have checked whether it is faulty...

Trying to get hold of an alternate frame now...

 

[Jay Gannon]

Understandable.

 

[John Marchant]

Misaligned prism optics in a 3 chip camera could cause something like this too... R, G and B data that should align to form the fine luma variance would misalign and create a rainbow effect.

 

[Gunleik Groven]

It's not 3-chip...

My bet is on diffusion on the OLPF, actually.

 

[John Marchant]

You thinking the deliberate diffusion in the OLPF is not acting neutrally across the spectrum? Seems possible, but I think it could be a math problem post-sensor equally, as it seems to only exist at the single pixel / sub-pixel level.

Love to know what the manufacturer of the mystery device have to say

 

[Gunleik Groven]

I am asking.

It is in all my exposures, but all the exposures are from the same cam.

It's not immediately this obvious, though...

But the R1MX and other cam was treated the same on the full/scaled image (after scaling to 1080 on the MX of course...)

 

[Gunleik Groven]

I see it here, too

http://provideocoalition.com/index....sonic_ag-hpx370_1_3_3-mos_p2_hd_camcorder/P2/

Here it's called CA, but that would suggest it wasdue to the lens (right?) which it isn't here... Same lens used on RED and "other"...

 

[Jay Gannon]

In theory, I think defective/cheap micro lenses on the sensor can produce CA.

 

[John Marchant]

That case is definitely the prism optics I was referring to, the cheap newsprint misalignment effect. If your test subject is definitely single chip then it has to be something else.

 

[Gunleik Groven]

As I said: Single chip, same optics, same stop

 

[Curran Giddens]

F3 no good?

 

[John Marchant]

Well, I'm going to place my bet on a 'They all do that, sir' response and a firmware processing/debayer type math deficiency as the source. Dice rolled

 

[Gunleik Groven]

F3 no good?

I'll start digging into that on thursday. So far I've just lifted it...

 

[Shane Betts]

Could the codec cause that? I really don't know, just asking...

 

[Gunleik Groven]

Could the codec cause that? I really don't know, just asking...

Yup. That's one thing I've been wondering. There are definite codec related issues here, too...

You like to see two DPXs?

One RED. one "other"

 

[Graeme Nattress]

Looks like traditional chroma moire from a poorly reconstructed inadequately optically filtered Bayer pattern CFA to me. Email me if you want more help on the diagnosis Gunleik,

Graeme

 

[Gunleik Groven]

Makes more sense, Graeme. Thanks!

G

 

[Graeme Nattress]

I can't see if the chroma moire is just in the horizontal direction or if it's equally vertical too from the images you've posted. If it's just horizontal, it's likely to be from F35 type stripe sensor. If it's vertical only, it's likely from line skipping. If it's equally horizontal and vertical it's likely my first guess.

Graeme

 

[Gunleik Groven]

I'll send you some a DPXs offlist...

Thanks

G

 

[Graeme Nattress]

Thanks Gunleik - happy to help.

Graeme