Banding on Z9

MCLV wrote:

JimKasson wrote:

vtpeters wrote:

JimKasson wrote:

MCLV wrote:

As I described here https://www.dpreview.com/forums/post/65887363 , using mechanical shutter will blur the banding. For f/2.8 and assuming shutter location 4 mm in front of the sensor plane we get a diameter of light cone of 4 mm / 2.8 = approximately 1.4 mm. Therefore, blurring caused by mechanical shutter would be roughly 1.4 mm / 24 mm = 5.8% of image height and 1.4 mm / 24 mm * 2000 = 117 pixels of the original image. Hence, I simulated how the image would look if a mechanical shutter was used by applying a linear blur of 117 pixels in vertical direction. I then applied layer mask around Yngwie, so he is not completely blurry. You can see results below.

Original - Electronic shutter

Simulated mechanical shutter at f/2.8

Good demo. You did this in the spatial domain, but you could have also done it in the frequency domain. A vertical 1xn kernel or its frequency-domain equivalent lowpass filter is also an effective mitigation tactic in post.

Could you explain this further for me.

I'm not that well versed in using Photoshop or Lightroom.

I don’t think there’s any way in Ps to get a custom kernel bigger than 5x5. AFAIK, Lr doesn’t support custom kernels at all. That’s why I suggested FFT earlier.

If you’d like. I can post code for doing it in Matlab.

You can use motion blur function instead of 1xn kernel. However, I didn't check if it averages the light intensity correctly or if it operates on RGB values directly.

I'm not familiar with 5x5 or 1xn kernel in Photoshop. I'll have to Google that and learn more about it.