Artifacts while shooting against the light of the sun

Skeeterbytes wrote:

whumber wrote:

TomzeLhotky wrote:

Hello dear. I'm new to the Olympus system. Own E-M1 mk III + 100-400 lens. In the photos I take at sunset and I have this light in the shot, I see a "grid of rectangles", or whatever I call it. Best if I could send you a photo. My question is whether this is normal behavior or something is wrong. Thank you, Tomas.

I enclose a 1: 1 cut-out

You're seeing the PDAF pixels.

My first thought as well but isn't that rather too many squares for the 121-point version of the PDAF sensor, or is the pattern applied to the entire surface and not just the PD sensors themselves?

¯\_(ツ)_/¯

Rick

121 selectable AF points doesn't mean there are only 121 PD points on the sensor. The selectable PD points are a firmware construct that aggregate data from a large number of individual PDAF pixels.

Interceptor121 wrote:

Pedagydusz wrote:

I have been taking pictures since the awn of Digital Photography, used Canon, Sony Olympus and Panasonic cameras, even a Sony camcorder and a variety of cell phones, and shoot regularly against the light (setting sun). I have never seen an image like that! Not even similar.

I have instead and not a coincidence a Sigma lens

That would explain why I sometimes get those artifacts on the Sigma 56mm (on the G9, no PDAF there).

ahaslett wrote:

https://blog.kasson.com/a7riv/pdaf-striping-in-the-sony-a7riv/

Horshack has also done work on this as well. Jim and Horshack might respond if you post on the PST forum.

Andrew

That is an excellent test report. Thanks for the link.

-- hide signature --

Infinite are the arguments of mages. Truth is a jewel with many facets. Ursula K LeGuin
Please feel free to edit any images that I post

olrett wrote:

Interceptor121 wrote:

Pedagydusz wrote:

I have been taking pictures since the awn of Digital Photography, used Canon, Sony Olympus and Panasonic cameras, even a Sony camcorder and a variety of cell phones, and shoot regularly against the light (setting sun). I have never seen an image like that! Not even similar.

I have instead and not a coincidence a Sigma lens

That would explain why I sometimes get those artifacts on the Sigma 56mm (on the G9).

The G9 does not have PDAF pixels so I have never had issues. I use the 56mm mostly for long focal landscape with plenty of backlighting and I have not seen a problem yet with that combo.

It is of course still possible to get a full sensor reflection though

whumber wrote:

Skeeterbytes wrote:

whumber wrote:

TomzeLhotky wrote:

Hello dear. I'm new to the Olympus system. Own E-M1 mk III + 100-400 lens. In the photos I take at sunset and I have this light in the shot, I see a "grid of rectangles", or whatever I call it. Best if I could send you a photo. My question is whether this is normal behavior or something is wrong. Thank you, Tomas.

I enclose a 1: 1 cut-out

You're seeing the PDAF pixels.

My first thought as well but isn't that rather too many squares for the 121-point version of the PDAF sensor, or is the pattern applied to the entire surface and not just the PD sensors themselves?

¯\_(ツ)_/¯

Rick

121 selectable AF points doesn't mean there are only 121 PD points on the sensor. The selectable PD points are a firmware construct that aggregate data from a large number of individual PDAF pixels.

In order to detect a phase difference you need at least two pixels even if it is called single point in reality it is likely to have 4 to cover horizontal vertical as minimum

In addition it is possible that the PDAF mask is an entirely separate circuit which then generates a grid.

Time ago Bill Claff posted the EM1MKII results where you could see the fixed pattern and it was not 121 dots

Interceptor121 wrote:

whumber wrote:

Skeeterbytes wrote:

whumber wrote:

TomzeLhotky wrote:

Hello dear. I'm new to the Olympus system. Own E-M1 mk III + 100-400 lens. In the photos I take at sunset and I have this light in the shot, I see a "grid of rectangles", or whatever I call it. Best if I could send you a photo. My question is whether this is normal behavior or something is wrong. Thank you, Tomas.

I enclose a 1: 1 cut-out

You're seeing the PDAF pixels.

My first thought as well but isn't that rather too many squares for the 121-point version of the PDAF sensor, or is the pattern applied to the entire surface and not just the PD sensors themselves?

¯\_(ツ)_/¯

Rick

121 selectable AF points doesn't mean there are only 121 PD points on the sensor. The selectable PD points are a firmware construct that aggregate data from a large number of individual PDAF pixels.

In order to detect a phase difference you need at least two pixels even if it is called single point in reality it is likely to have 4 to cover horizontal vertical as minimum

In addition it is possible that the PDAF mask is an entirely separate circuit which then generates a grid.

Time ago Bill Claff posted the EM1MKII results where you could see the fixed pattern and it was not 121 dots

You're probably thinking of this thread. For the OP, here's another user that ran into the same issue. It's the same thing people made a stir about back when either the A7Rii or A7iii were released, it just looks different because of the crosstype pattern Olympus uses. I haven't been able to induce anything similar with the OM-1 which is expected since it no longer needs to use PD masks.

Adrian Harris wrote:

ahaslett wrote:

https://blog.kasson.com/a7riv/pdaf-striping-in-the-sony-a7riv/

Horshack has also done work on this as well. Jim and Horshack might respond if you post on the PST forum.

Andrew

That is an excellent test report. Thanks for the link.

Jim is an outstanding contributor.

Andrew

If you look at the jpeg I made from the RAW that OP supplied, you can see hundreds of clearly defined squares in a square grid across the whole image.

Andrew

Sometimes that happens shooting into the sun when you get the internal reflections as in the image first image below (one of my EM1s - I don't remember which one).  Of course, sometimes you don't get it even with the sun in the image (this happens when you follow a BIF who flies by the sun in S mode - second image)

whumber wrote:

Interceptor121 wrote:

whumber wrote:

Skeeterbytes wrote:

whumber wrote:

TomzeLhotky wrote:

Hello dear. I'm new to the Olympus system. Own E-M1 mk III + 100-400 lens. In the photos I take at sunset and I have this light in the shot, I see a "grid of rectangles", or whatever I call it. Best if I could send you a photo. My question is whether this is normal behavior or something is wrong. Thank you, Tomas.

I enclose a 1: 1 cut-out

You're seeing the PDAF pixels.E

My first thought as well but isn't that rather too many squares for the 121-point version of the PDAF sensor, or is the pattern applied to the entire surface and not just the PD sensors themselves?

¯\_(ツ)_/¯

Rick

121 selectable AF points doesn't mean there are only 121 PD points on the sensor. The selectable PD points are a firmware construct that aggregate data from a large number of individual PDAF pixels.

In order to detect a phase difference you need at least two pixels even if it is called single point in reality it is likely to have 4 to cover horizontal vertical as minimum

In addition it is possible that the PDAF mask is an entirely separate circuit which then generates a grid.

Time ago Bill Claff posted the EM1MKII results where you could see the fixed pattern and it was not 121 dots

You're probably thinking of this thread. For the OP, here's another user that ran into the same issue. It's the same thing people made a stir about back when either the A7Rii or A7iii were released, it just looks different because of the crosstype pattern Olympus uses. I haven't been able to induce anything similar with the OM-1 which is expected since it no longer needs to use PD masks.

Yes exactly that. The OM-1 sensor is back illuminated so the circuitry is not on the side exposed to the lens and in addition the cells are all the same. Yet I am not sure if the allocation to a cell to behave as PD or imaging change when the sensor is readout

Either way with a lens with AR coating on the back you are extremely unlikely to see anything