Am I correct in calling this Chromatic Aberration?

Started by timbytheriver, December 14, 2013, 02:10:22 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

timbytheriver

Please look at the colour artefacts on the water ripples below.

https://dl.dropboxusercontent.com/u/21590714/raw2Cdng.jpg

https://dl.dropboxusercontent.com/u/21590714/raw2dng.jpg

Is it CA or is it something different? Does anyone know how to fix it?

I see it more heavily in C-dng, and Cineform-raw [through raw2gpcf] – but almost non-existently in 'normal' dng processed from .raw through ACR.

If anyone can help I'd really appreciate some [more] expert eyes on this! ;)

Many thanks.

5DMkII - These shots 1880 x 1058 23.976, taken on a.d. build 4.10.13

5D3 1.1.3
5D2 2.1.2

g3gg0

that looks like some debayering artifacts.
CA itself is a result of an inavoidable lens error and will cause green/blue/magenta color bleeding on edges of bright areas.
if CA *correction* is *falsely* applied to areas that dont have any CA, it may result in weird looking borders of bright areas.

so it *could* be a result of a CA correction algorithm or some debayering artifacts
Help us with datasheets - Help us with register dumps
magic lantern: 1Magic9991E1eWbGvrsx186GovYCXFbppY, server expenses: [email protected]
ONLY donate for things we have done, not for things you expect!

timbytheriver

@g3gg0 Thanks for responding.

But if I've not misunderstood the whole process, the debayering is performed by software which produces a visual raster (in this case ACR). So why are these files [both in this example being processed identically by ACR] showing very different levels of debayering artifacts? Or is it instead down to the .raw > dng converter method, into which the debayering is already 'pre-baked' to the dng/Cdng?

I've confused myself! :o



5D3 1.1.3
5D2 2.1.2

a1ex

Looks like aliasing to me, but without seeing the DNGs I'm not able to diagnose it.


timbytheriver

@a1ex Thanks. If I post them would you [or someone else] be able to have a look please? Bear in mind that they are two dngs created from one identical .raw file!

https://dl.dropboxusercontent.com/u/21590714/raw2dng.dng

https://dl.dropboxusercontent.com/u/21590714/raw2Cdng.dng

Here's hoping!
5D3 1.1.3
5D2 2.1.2

mageye

Looks like moire artefacts to me. It's pretty typical with water where there's lots of fine detail. I don't really see any evidence of CA. I inspected the edge of the frame where it should be more obvious. For some reason the 'raw2dng.dng' file is hugely overexposed (with lots of clipping). I am not sure why that is?
5DMKII | 500D | KOMPUTERBAY 32GB Professional 1000x |Canon EF 50mm f/1.8 II | Samyang 35mm f/1.4 ED AS UMC | Canon EF 75-300mm f/4-5.6 III | Zoom H2 (4CH. audio recorder) | Mac OS X 10.9.2 | Photoshop CC | After Effects CC | Final Cut Pro 7

timbytheriver

Quote from: mageye on December 14, 2013, 05:09:58 PM
Looks like moire artefacts to me.
Thanks. But if that were so, wouldn't it be present on *both* dngs? These are two types of dng from the *same* raw file – both processed through ACR!

It doesn't make sense to me... :o

**This shot was a test shot – deliberately overexposed by +1.5 stops. It pulls back fine in ACR.**
5D3 1.1.3
5D2 2.1.2

AnotherDave

It is present on both DNGs.  It is more noticeable on the CDNG version, but it is certainly there on both files.

The color fringing you're seeing IS being initially caused by the moire pattern of the camera. Is RAW2CDNG  still converting the footage to 16bit so it can be recognized by Resolve?  That is no longer necessary. 

timbytheriver

Thanks. Though I've worked with it, I don't use Resolve in my workflow at the moment.

The raw2Cdng frame here is the 16b flavour – but it [the moire, ca, artifacts] is also present in a 12b file, to the same degree.

Why should the raw2dng version allow ACR to process the moire artefact so much better?

5D3 1.1.3
5D2 2.1.2

chmee

@AnotherDave my approach is to get dng 100% working in Adobe-universe. I'm not really interested in Resolve anymore..
[size=2]phreekz * blog * twitter[/size]

timbytheriver

5D3 1.1.3
5D2 2.1.2