Reading mode: Light Dark

News Reviews Articles Buying Guides Sample Images Videos Cameras Lenses Phones Printers Showcase Forums Galleries Challenges

All forumsSony Alpha Full Frame E-mount TalkChange forum

What causes stripe noise on a7iii?

Started Mar 1, 2018 | Discussions

Forum

Threaded view

sybersitizen • Forum Pro • Posts: 26,098

Re: was wishful thinking, too bad....

In reply to (unknown member) • Mar 5, 2018

BBQue wrote:

JimKasson wrote:

BBQue wrote:

(sorry the all caps and triple exclamation marks)

Probably not. Happens in the a9, too:

http://blog.kasson.com/a9/pdaf-striping-on-the-a9/

hmmm... that's too bad. But probably not a big deal in "real life".

It (or something very similar) has been happening in a number of E-mount bodies since at least 2014; and no, camera anomalies are not a big deal ... until people decide they are. Then the petitions start.

Reply Reply with quote Reply to thread Complain

Horshack • Veteran Member • Posts: 9,450

Re: A9 sample from Jim using round light source

In reply to JimKasson • Mar 5, 2018

JimKasson wrote:

Horshack wrote:

Jim sent me some great A9 raws he shot using a Godox Flashpoint XPLOR 600PRO as the light source. This creates a perfectly circular light source, which allows us to better delineate the axis of where the G vs G2 banding begins and ends.

Here is an animated GIF I created from the raw (click on image, then on "100% zoom" in the dpreview gallery viewer to animate, panning around if necessary). I darkened out the center of the strobe to make viewing of the stripes around the outer of the circle easier.

Very nice.

I'm embroiled with taxes today, but I'm thinking about writing some code to do heatmaps for this artifact.

Jim

That would be cool Jim. Here's a blend I create to further highlight the transition point between the G/G2 banding. I inverted the G bands so that they're black, while the G2 bands are white. If you follow around the circle you can see where one begins and the other ends, which is almost exactly at 45°. Click on the image to see the full-size.

Reply Reply with quote Reply to thread Complain

Horshack • Veteran Member • Posts: 9,450

Illustration of theory

In reply to Horshack • Mar 6, 2018

Here's an illustration I created to depict how and why the G1 bands are only visible above/below the light source and G2 bands only left/right. In my theory I indicated this might be because of the axis of light relative to the angle of light that the micro-lenses are designed to absorb - in reality I think it may be much simpler. If we assume that light is reflecting off the PDAF pixels radially in all directions (due to the property of the lens optics/aperture), and this is reaching the microlenses at an angle which can be absorbed into the pixels, then the explanation of why the G1/G2 bands taper across their respective axes is just a function of the radial falloff of light (ie, the distance from the center of a circle to its horizontal/vertical edges is shorter than the distance to its diagonal edge).

This also would mean that both G1/G2 are "striping" in the center areas of the light glare - we just can't see it because the luminance values are at/near clipping. We however can see both G1/G2 striping around the circular perimeter of the light glare, where the luminance levels have dropped off enough to reach a tonal area where the stripes would be visible.

I depicted everything described above in the following illustration. Click on the image for the full-sized version.

Reply Reply with quote Reply to thread Complain

bclaff • Forum Pro • Posts: 14,055

Re: I still believe it's related to PDAF pixels.

In reply to bclaff • Mar 6, 2018

bclaff wrote:

JoFa wrote:

Bill, can you post a link of your other analysis showing that same pattern?

Here are a couple of links to older posts:

ILCE-9 (A9) PDAF Row Locations

Re: PDAF

And here's summary of the research I did at that time:

The color code at the bottom indicates sensor with identical PDAF layouts

BTW, this analysis was performed on black frames. Light isn't required to detect the pattern.

So, I looked at the ILCE-7M3 and it has the identical pattern as the ILCE-9 except shifted down 8 rows.

But, there does look like a surprise.

In all my earlier investigations the PDAF pixels were distingished by lower Read Noise.
For example, here's the ILCE-9:

But my initial ILCE-7M3 investigation shows this:

Dramatically (3 stops) higher Read Noise in the PDAF blue pixels.

As they say ... interesting !

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

Reply Reply with quote Reply to thread Complain

Horshack • Veteran Member • Posts: 9,450

Re: I still believe it's related to PDAF pixels.

In reply to bclaff • Mar 6, 2018

bclaff wrote:

bclaff wrote:

JoFa wrote:

Bill, can you post a link of your other analysis showing that same pattern?

Here are a couple of links to older posts:

ILCE-9 (A9) PDAF Row Locations

Re: PDAF

And here's summary of the research I did at that time:

The color code at the bottom indicates sensor with identical PDAF layouts

BTW, this analysis was performed on black frames. Light isn't required to detect the pattern.

So, I looked at the ILCE-7M3 and it has the identical pattern as the ILCE-9 except shifted down 8 rows.

But, there does look like a surprise.

In all my earlier investigations the PDAF pixels were distingished by lower Read Noise.
For example, here's the ILCE-9:

But my initial ILCE-7M3 investigation shows this:

Dramatically (3 stops) higher Read Noise in the PDAF blue pixels.

As they say ... interesting !

That is interesting, thanks. I wonder what the distribution of pixel values for those A7III PDAF rows look like.

Reply Reply with quote Reply to thread Complain

KKnipser • Regular Member • Posts: 233

Re: Illustration of theory

In reply to Horshack • Mar 6, 2018

Does that apply to all models with on sensor PDAF or is it A7III specific?

Reply Reply with quote Reply to thread Complain

Entropy512 • Veteran Member • Posts: 6,197

Re: A7rII observation - PDAF bands on G rather than G2

In reply to Horshack • Mar 6, 2018

Horshack wrote:

The banding on the G2/B channel of the A9/A7III is somewhat logical considering they're on the same Bayer row (even though we don't yet understand why the G2 pixels would be affected since PDAF uses the blue pixels).

At least for this particular item - it was my understanding that the PDAF sensels "stole" a blue pixel to create a dual-pixel site under a single microlens - with the other pixel being one of the green pixels.

As a result while the green pixel might still have data, that data is going to be impacted by the different optical arrangement of the microlens for that sensel - to the point where it would be very difficult to do anything other than simply interpolate THAT pixel out too.

-- hide signature --

Context is key. If I have quoted someone else's post when replying, please do not reply to something I say without reading text that I have quoted, and understanding the reason the quote function exists.

Entropy512's gear list:Entropy512's gear list

Sony a6000 Pentax K-5 Pentax K-01 Sony a6300 Canon EF 85mm F1.8 USM +5 more

Reply Reply with quote Reply to thread Complain

JimKasson • Forum Pro • Posts: 44,299

Re: I still believe it's related to PDAF pixels.

In reply to bclaff • Mar 6, 2018

bclaff wrote:

bclaff wrote:

JoFa wrote:

Bill, can you post a link of your other analysis showing that same pattern?

Here are a couple of links to older posts:

ILCE-9 (A9) PDAF Row Locations

Re: PDAF

And here's summary of the research I did at that time:

The color code at the bottom indicates sensor with identical PDAF layouts

BTW, this analysis was performed on black frames. Light isn't required to detect the pattern.

So, I looked at the ILCE-7M3 and it has the identical pattern as the ILCE-9 except shifted down 8 rows.

But, there does look like a surprise.

In all my earlier investigations the PDAF pixels were distingished by lower Read Noise.
For example, here's the ILCE-9:

But my initial ILCE-7M3 investigation shows this:

Dramatically (3 stops) higher Read Noise in the PDAF blue pixels.

As they say ... interesting !

Good work, Bill! Could this mean that the blue pixels are split and aren't interpolated over? Do those particular blue pixels taken as a group have lower FWC?

Jim

-- hide signature --

http://blog.kasson.com

JimKasson's gear list:JimKasson's gear list

Leica Q2 Monochrom Nikon Z7 Nikon Z9 Hasselblad X2D 100c Fujifilm GFX 100 II +1 more