I’ve been experimenting a bit trying to reducing the effect of Green/Cyan CA with the FX9. I have discovered a couple of things that can help reduce interactions between the cameras processing and areas of high contrast that may be exhibiting bright green/cyan fringes.
First thing to note is that a lens with less or no CA will not have the same issue. But as no lens is totally CA free the below settings can help.
These changes are for Custom Mode Only.
1: Turn OFF the aperture correction in the paint menu. Turning off the aperture correction noticeably reduces the cameras tendency to create black halos in areas of extreme contrast and it also reduces the enhancement of areas of strong CA. This has a softening/smoothing effect on bright CA making it much less noticeable. There is very, very little loss of sharpness in the rest of the image and I actually prefer the way the camera looks with this turned off.
2: Use the Multi-Matrix to reduce the brightness of Green/Cyan CA. The most common examples of CA causing an issue are with background out of focus high contrast areas. In this case the CA is normally Green/Cyan. It’s possible to tune the cameras multimatrix to reduce the brightness of these green/cyan edges. If you turn ON the mutli-matrix and then select CY+ and set this to -30 you will see a very useful reduction in the intensity of the CA. For a stronger reduction in addition select CY and set this to -15.
Changing these setting will have an impact on the reproduction of other cyan objects in your shots, but you should see this in the VF and testing various scenes these changes are typically not noticable. In some cases I am finding I actually like this slightly modified look!
Use both of the above together for the strongest impact. But if you are only going to use one, turn off the aperture correction.
14 thoughts on “Reducing CA Artefacts In The Sony FX9.”
What lense did you use to get that green outline in high contrast backlight area.
I seen some funky CA and purple in water on a video but never seen this green until some1 mentioned it.
Does this happen on Sonys GM glass? Or just older glass?
Curious to know what lens is being used?
This was with a Sony 24-70mm GM. But it’s worth noting that I had to find exactly the right combination of exposure (very over exposed) aperture and focus to get this artefact and the example images are 400% crops. It is not something that I normally see and in 7 months of shooting with the FX9 I have not had an issue. But some people do.
It can happen with any lens. All lenses exhibit some CA, generally you will see less with more expensive lenses and more with zoom lenses.
Thanks Alister, Very useful post. I’ve read comments claiming that the FX9 produces Green/Magenta fringing on highlights but have been confused by that as I always assumed that its a common function of lens CA rather than a sensor issue. Very interesting to see how the aperture setting can increase how noticeable that is. BTW – Does the FX9 itself have any CA issues as I’ve read?
Love these continuing blog posts.
Alister, sorry a question for another topic, but can you explain with hypergammas what is meant by 800% or 8009? 100% IRE is 0.714 volts. It doesn’t seem the meaning can be 800% of pk-pk voltage. 109% of IRE gets you 800% of what?
Have I figured out the meaning, 800% would be 3 stops of gamma?
To be honest I am not completely sure how the number is derived but it is a % measure of the dynamic range relative to a typical Rec-709 camera. 460% is just over 10 stops, 800% equates to around 11 stops of total captured DR.
The last bit after the 800 which is either a 0 or a 9 is the recording range. 0 = 100% (broadcast safe) and 109 = 109% (Full range with super whites).
Ah, then the answer was right in front and you figured it out yourself. Thanks! If 8009 is full range 800%, since each stop is a doubling of light and 8009 = 11 stops, then 8009 represents the percentage recording range of Rec-709 camera equal to 3 stops over (2e+3=800%), the baseline being 8 stops, 8 plus 3 = 11.
We can extend this nomenclature out:
1009 = 109% Rec-709 full range 8-stop
4609 = 460% / 10.2-stop
6009 = 600% / 10.6-stop
8009 = 800% / 11.0-stop
Except it doesn’t work out like that as Sony rate 14 stop S-Log3 at 1300%. I’ve never been able to figure it out.
I believe the answer is the overexposure range for Rec-709 and S-Log3 is measured from from 90% white. So for S-Log3, 61% is 2.3 stops above middle gray. 1300% overexposure is 3.7 stops more, so 3.7+2.3=6 stops above middle gray, with 8 stops below middle gray, there’s your 14 stops for S-Log3.
No, it’s not that either because it’s a measure of the total DR, not just the over exposure. For example the 460% hypergammas go around +4.5 stops over middle grey. Further to that S-Log3 and the FX9 is rated at 1600%. but also goes to +6 stops, but down over 9 stops.
It is somehow related to the nominal total range of 709, but 100% is not 1 stop and it doesn’t appear to be a log scale. There is also a lot of variation between manufactures as to how it is applied, for example Canon rate 1600% as 14 stops, while Sony use 1300%.
In simple terms, it’s ratio of the input % that gives 100% output. It follows that STD gamma 100% input gives 100% output. HG4600 gives 100% output at 460% input. The input is linear, the output non-linear, so the ratio only holds for the 100% output level.
The problem is it is an old video terminology that originally was very specifically related to stops above middle grey, relative to rec-709.
But because log gamma behaves very differently to power law gamma it doesn’t “fit” log gamma.
S-Log which is a 12 stop gamma goes to around 5.2 stops above middle grey and this is regarded as 800%. But then we go to S-Log3 which goes to +6 stops, suddenly this is a 1300% gamma.
And then it is further confused when you have a power law gamma like the Hypergammas that go to around +5.5 stops but are only 460%.
And how do you equate a 10 stop gamma as 460%, at 12 stop gamma as 800% and a 14 stop gamma as 1300%? This clearly isn’t a linear relationship or a log relationship.
There is no logic to the use of % anymore because originally it was pegged against stops above middle grey and didn’t consider anything below middle grey. But now we have log gammas that go far beyond the -3 or -4 stops of Rec-709 the numbers don’t truly make sense as the reference is still middle grey but the range is now dramatically increased downwards. Different manufacturers give different % ratings for the same input ranges. After the release of S-Log3 Sony stopped including % values in any of their dynamic range or gamma white papers and now refer to dynamic range in stops.