Revisited Milky Way Nightscape + MFT capability considerations

Adrian Harris wrote:

Very very impressive !

Thank you Adrian trying hard out there!

Interceptor121 wrote:

This pattern is not related to the PDAF points, you can get the same behavior from old DSLRs under the right conditions. It's related to diffraction patterns that form within the sensor stack. There's a good discussion about it here.

Image taken with a T2i (Taken by Jerry Lodriguss)

whumber wrote:

Interceptor121 wrote:

This pattern is not related to the PDAF points, you can get the same behavior from old DSLRs under the right conditions. It's related to diffraction patterns that form within the sensor stack. There's a good discussion about it here.

Image taken with a T2i (Taken by Jerry Lodriguss)

We don't know what lens Interceptor used for his image.  I recall previous discussions about differences  between Oly & Panasonic in the AR coatings used on rear lens elements.  If memory serves, the old Pana 7-14 will produce this pattern when used on Oly  cams.

whumber wrote:

Interceptor121 wrote:

This pattern is not related to the PDAF points, you can get the same behavior from old DSLRs under the right conditions. It's related to diffraction patterns that form within the sensor stack. There's a good discussion about it here.

Image taken with a T2i (Taken by Jerry Lodriguss)

The image you posted is not the full frame but a magnification of the image which is of course misleading further more the post goes on talking about reflection in a mirror which in a mirrorless camera is not even there

this is the full image from that post

When you look at that and compare instead with the grouping and spacing of my image which are 9 blocks this is very close to the 11x11 arrays of 16x16 pixels that make the mask which combined with the coating of the sensor create a problem. No idea about the other artifact on the right hand side which I never bothered investigating

On the day that happened I also had my GH5 and managed to take a shot with the same lens from the same position

So maybe they not be the PDAF pixels however the other camera without the PDAF pixels did not produce the same artefacts, maybe it is the coating or perhaps a combination of the PDAF and the coating however as I do take images like that and that kept happening I decided to get rid of the camera EM1MKII

Maybe I should rewrite the post saying for best results, avoid dark frames use Panasonic cameras but I guess that would also attract some feedback?

tradesmith45 wrote:

whumber wrote:

Interceptor121 wrote:

This pattern is not related to the PDAF points, you can get the same behavior from old DSLRs under the right conditions. It's related to diffraction patterns that form within the sensor stack. There's a good discussion about it here.

Image taken with a T2i (Taken by Jerry Lodriguss)

We don't know what lens Interceptor used for his image. I recall previous discussions about differences between Oly & Panasonic in the AR coatings used on rear lens elements. If memory serves, the old Pana 7-14 will produce this pattern when used on Oly cams.

Sigma 56mm and besides the image he posted there is highly misleading which is typical unfortunately have a look at my reply

bclaff wrote:

tjl66 wrote:

bclaff wrote:

Regardless of whether the FPN is horizontal or vertical or otherwise it can definitely affect image quality. I don't consider PDAF pixels that show as FPN to be defective but you may consider that semantics.

Once again thank you for your insights Bill!

Regarding your website dark measurements, can you clarify whether these measurements are based on short (i.e bias) exposures?

The reason I ask is if you look at the example below. which shows 200% enlarged crops from my em1.2 from the same part of the sensor, all at ISO3200 but using 30 second exposures:

Top image: Guided Astrophoto 6 x 30 second stack (uncalibrated)
Middle image: Corresponding dark frame (33x30sec stack) of the same portion of the sensor.
Bottom image: Corresponding dark frame (33x30sec stack) of the same portion of the sensor, greatly stretched

Note: 2 common hotpixels on each image have been marked with a white circle for reference.

Only in the greatly stretched dark frame (bottom image) do I think I can see hints of the row/column structure. What I am trying to illustrate is that in a real astrophoto with lots of stars, only the most prominent hotpixels show and the row/column "defects" are simply not visible.

Yes, the dark measurements (DSNU etc.) and Sensor Heatmap visualizations at PhotonsToPhotos are for very short exposures.

At 30 seconds you are almost certainly seeing the effect of Dark Current Non-Uniformity (DCNU). This is a different class of defect from the typical "hot" pixel.

You may find Dark Current Non-Uniformity interesting reading.

The presence of non imaging pixels together with hot pixels is likely to further aggravate the issue of phase difference pixels not to alleviate it.

Besides the problem is not when there is a dark scene but where there is a highlight in the scene like a bright star this is the one that generates the issue not the dark one which in fact may be cleaned up by a noise reduction process

There was a post from an OIympus user on a special group on facebook and when I asked him why he was using an EM10 MKII said I do not want to have patterns he then went to show me a set of images taken with that and with the EM1MKII which helped me to conclude I could not use the camera for that purpose

I will see if I can dig it out although as I sold the last Olympus camera I left the group

Interceptor121 wrote:

whumber wrote:

Interceptor121 wrote:

This pattern is not related to the PDAF points, you can get the same behavior from old DSLRs under the right conditions. It's related to diffraction patterns that form within the sensor stack. There's a good discussion about it here.

Image taken with a T2i (Taken by Jerry Lodriguss)

The image you posted is not the full frame but a magnification of the image which is of course misleading further more the post goes on talking about reflection in a mirror which in a mirrorless camera is not even there

this is the full image from that post

When you look at that and compare instead with the grouping and spacing of my image which are 9 blocks this is very close to the 11x11 arrays of 16x16 pixels that make the mask which combined with the coating of the sensor create a problem. No idea about the other artifact on the right hand side which I never bothered investigating

On the day that happened I also had my GH5 and managed to take a shot with the same lens from the same position

So maybe they not be the PDAF pixels however the other camera without the PDAF pixels did not produce the same artefacts, maybe it is the coating or perhaps a combination of the PDAF and the coating however as I do take images like that and that kept happening I decided to get rid of the camera EM1MKII

Maybe I should rewrite the post saying for best results, avoid dark frames use Panasonic cameras but I guess that would also attract some feedback?

Fuji has a related problem w/ backlighting that I've never seen.  It's called the purple grid flare artifact.

tradesmith45 wrote:

Interceptor121 wrote:

whumber wrote:

Interceptor121 wrote:

This pattern is not related to the PDAF points, you can get the same behavior from old DSLRs under the right conditions. It's related to diffraction patterns that form within the sensor stack. There's a good discussion about it here.

Image taken with a T2i (Taken by Jerry Lodriguss)

The image you posted is not the full frame but a magnification of the image which is of course misleading further more the post goes on talking about reflection in a mirror which in a mirrorless camera is not even there

this is the full image from that post

When you look at that and compare instead with the grouping and spacing of my image which are 9 blocks this is very close to the 11x11 arrays of 16x16 pixels that make the mask which combined with the coating of the sensor create a problem. No idea about the other artifact on the right hand side which I never bothered investigating

On the day that happened I also had my GH5 and managed to take a shot with the same lens from the same position

So maybe they not be the PDAF pixels however the other camera without the PDAF pixels did not produce the same artefacts, maybe it is the coating or perhaps a combination of the PDAF and the coating however as I do take images like that and that kept happening I decided to get rid of the camera EM1MKII

Maybe I should rewrite the post saying for best results, avoid dark frames use Panasonic cameras but I guess that would also attract some feedback?

Fuji has a related problem w/ backlighting that I've never seen. It's called the purple grid flare artifact.

I see a common trait and is a mask on the sensor. Perhaps this mask somehow reflects or reflects differently.

Of course if there is zero reflection from the sensor this problem won't exist

This post suggests the issue is PDP phase difference pixels

https://petapixel.com/2017/02/23/fuji-cameras-produce-strange-purple-flaregrid-artifact/

I also suspect a lens with better coating may also be at play here however I did not intend to abandon a lens that has given me many successful shots because camera A was not working with it while B was working fine. So I decided to get rid of camera A which in turn stopped my investigations

My objective is to take good images not to spend time debugging issues. When I find problems that I cannot resolve I avoid them

Interceptor121 wrote:

whumber wrote:

Interceptor121 wrote:

This pattern is not related to the PDAF points, you can get the same behavior from old DSLRs under the right conditions. It's related to diffraction patterns that form within the sensor stack. There's a good discussion about it here.

Image taken with a T2i (Taken by Jerry Lodriguss)

The image you posted is not the full frame but a magnification of the image which is of course misleading

this is the full image from that post

That's a different image from a different poster taken with a 6D, the image I posted is the full image from another individual. I'm not sure how you could think that the image I posted is a crop of the one you've posted. There are plenty of examples of this phenomenon in that thread.

When you look at that and compare instead with the grouping and spacing of my image which are 9 blocks this is very close to the 11x11 arrays of 16x16 pixels that make the mask which combined with the coating of the sensor create a problem. No idea about the other artifact on the right hand side which I never bothered investigating

A combination of sensor coatings, stack thickness, AA filter design and microlens spacing (as well as the size of the source within the frame) are what seem to primarily control the diffraction patterns.

On the day that happened I also had my GH5 and managed to take a shot with the same lens from the same position

So maybe they not be the PDAF pixels however the other camera without the PDAF pixels did not produce the same artefacts, maybe it is the coating or perhaps a combination of the PDAF and the coating however as I do take images like that and that kept happening I decided to get rid of the camera EM1MKII

Sensor coating differences are the more likely cause but couldn't say for sure.

tradesmith45 wrote:

whumber wrote:

Interceptor121 wrote:

This pattern is not related to the PDAF points, you can get the same behavior from old DSLRs under the right conditions. It's related to diffraction patterns that form within the sensor stack. There's a good discussion about it here.

Image taken with a T2i (Taken by Jerry Lodriguss)

We don't know what lens Interceptor used for his image. I recall previous discussions about differences between Oly & Panasonic in the AR coatings used on rear lens elements. If memory serves, the old Pana 7-14 will produce this pattern when used on Oly cams.

The 7-14 f/4 produces flaring but it's not in diffraction grid pattern like this, rather it comes off as a purple blob. That said, coating differences on the sensor are a very likely explanation.

whumber wrote:

Interceptor121 wrote:

whumber wrote:

Interceptor121 wrote:

This pattern is not related to the PDAF points, you can get the same behavior from old DSLRs under the right conditions. It's related to diffraction patterns that form within the sensor stack. There's a good discussion about it here.

Image taken with a T2i (Taken by Jerry Lodriguss)

The image you posted is not the full frame but a magnification of the image which is of course misleading

this is the full image from that post

That's a different image from a different poster taken with a 6D, the image I posted is the full image from another individual. I'm not sure how you could think that the image I posted is a crop of the one you've posted. There are plenty of examples of this phenomenon in that thread.

When you look at that and compare instead with the grouping and spacing of my image which are 9 blocks this is very close to the 11x11 arrays of 16x16 pixels that make the mask which combined with the coating of the sensor create a problem. No idea about the other artifact on the right hand side which I never bothered investigating

A combination of sensor coatings, stack thickness, AA filter design and microlens spacing (as well as the size of the source within the frame) are what seem to primarily control the diffraction patterns.

On the day that happened I also had my GH5 and managed to take a shot with the same lens from the same position

So maybe they not be the PDAF pixels however the other camera without the PDAF pixels did not produce the same artefacts, maybe it is the coating or perhaps a combination of the PDAF and the coating however as I do take images like that and that kept happening I decided to get rid of the camera EM1MKII

Sensor coating differences are the more likely cause but couldn't say for sure.

https://petapixel.com/2017/02/23/fuji-cameras-produce-strange-purple-flaregrid-artifact/

A clear example of how PDP pixels create flare in the fuji system

I guess simply nobody has bothered looking into olympus cameras

Interceptor121 wrote:

whumber wrote:

Interceptor121 wrote:

whumber wrote:

Interceptor121 wrote:

This pattern is not related to the PDAF points, you can get the same behavior from old DSLRs under the right conditions. It's related to diffraction patterns that form within the sensor stack. There's a good discussion about it here.

Image taken with a T2i (Taken by Jerry Lodriguss)

The image you posted is not the full frame but a magnification of the image which is of course misleading

this is the full image from that post

That's a different image from a different poster taken with a 6D, the image I posted is the full image from another individual. I'm not sure how you could think that the image I posted is a crop of the one you've posted. There are plenty of examples of this phenomenon in that thread.

When you look at that and compare instead with the grouping and spacing of my image which are 9 blocks this is very close to the 11x11 arrays of 16x16 pixels that make the mask which combined with the coating of the sensor create a problem. No idea about the other artifact on the right hand side which I never bothered investigating

A combination of sensor coatings, stack thickness, AA filter design and microlens spacing (as well as the size of the source within the frame) are what seem to primarily control the diffraction patterns.

On the day that happened I also had my GH5 and managed to take a shot with the same lens from the same position

So maybe they not be the PDAF pixels however the other camera without the PDAF pixels did not produce the same artefacts, maybe it is the coating or perhaps a combination of the PDAF and the coating however as I do take images like that and that kept happening I decided to get rid of the camera EM1MKII

Sensor coating differences are the more likely cause but couldn't say for sure.

https://petapixel.com/2017/02/23/fuji-cameras-produce-strange-purple-flaregrid-artifact/

A clear example of how PDP pixels create flare in the fuji system

I guess simply nobody has bothered looking into olympus cameras

There are multiple threads on those types of actual PDAF artifacts regarding the E-M1ii in this very forum. That's an entirely different phenomenon from what you've posted though.

idk why are you keep focusing on the minor technical details that matter the least instead of working on your PP more, especially your colour grading for low light shots.

whumber wrote:

Interceptor121 wrote:

whumber wrote:

Interceptor121 wrote:

whumber wrote:

Interceptor121 wrote:

This pattern is not related to the PDAF points, you can get the same behavior from old DSLRs under the right conditions. It's related to diffraction patterns that form within the sensor stack. There's a good discussion about it here.

Image taken with a T2i (Taken by Jerry Lodriguss)

The image you posted is not the full frame but a magnification of the image which is of course misleading

this is the full image from that post

That's a different image from a different poster taken with a 6D, the image I posted is the full image from another individual. I'm not sure how you could think that the image I posted is a crop of the one you've posted. There are plenty of examples of this phenomenon in that thread.

When you look at that and compare instead with the grouping and spacing of my image which are 9 blocks this is very close to the 11x11 arrays of 16x16 pixels that make the mask which combined with the coating of the sensor create a problem. No idea about the other artifact on the right hand side which I never bothered investigating

A combination of sensor coatings, stack thickness, AA filter design and microlens spacing (as well as the size of the source within the frame) are what seem to primarily control the diffraction patterns.

On the day that happened I also had my GH5 and managed to take a shot with the same lens from the same position

So maybe they not be the PDAF pixels however the other camera without the PDAF pixels did not produce the same artefacts, maybe it is the coating or perhaps a combination of the PDAF and the coating however as I do take images like that and that kept happening I decided to get rid of the camera EM1MKII

Sensor coating differences are the more likely cause but couldn't say for sure.

https://petapixel.com/2017/02/23/fuji-cameras-produce-strange-purple-flaregrid-artifact/

A clear example of how PDP pixels create flare in the fuji system

I guess simply nobody has bothered looking into olympus cameras

There are multiple threads on those types of actual PDAF artifacts regarding the E-M1ii in this very forum. That's an entirely different phenomenon from what you've posted though.

I am not sure how you are so confident this isn't while others are. Or perhaps nobody has looked at backlit scenes.

Either way I would not stop taking some images because my camera has a problem with the scene but would change the camera so the conclusion is the same

There was a very competent German guy who showed me some nice patterning examples of EM1MKII when I asked him why he was using an EM10 on facebook but as I got rid of the camera very shortly after I did not bother keeping the links

I will see if I can find it but my advice at this point would change from use a contrast detect camera to use a panasonic camera if the issue is also with the coating

Interceptor121 wrote:

I am not sure how you are so confident this isn't while others are. Or perhaps nobody has looked at backlit scenes.

Either way I would not stop taking some images because my camera has a problem with the scene but would change the camera so the conclusion is the same

It's absolutely a weakness of the E-M1ii/iii/X, it's just that it's not coming from the PD pixels but other design considerations.

There was a very competent German guy who showed me some nice patterning examples of EM1MKII when I asked him why he was using an EM10 on facebook but as I got rid of the camera very shortly after I did not bother keeping the links

I will see if I can find it but my advice at this point would change from use a contrast detect camera to use a panasonic camera if the issue is also with the coating

If you create a master dark, for example, from the E-M1ii you'll see a very distinct signature of the PDAF pattern on the sensor if you stretch it to an extreme, no question about that.

400% view of E-M1iii master dark pushed +8 stops.

I suspect that this is the pattern that you saw from the German individual. This is the same pattern you'll see in certain glaring flare conditions except there you won't need to do any kind of extreme push to induce it. These are all very different from the diffraction grating pattern you showed though which shows up on plenty of non-PDAF patterns.

jimboyvr wrote:

Interceptor121 wrote:

Once home I have reprocessed the shots already shared here on my desktop and realised I had omitted some steps in my process so here are some updated files

Shots with GH5M2 PL12mm

This is a previous shot with the GH6 + PL15mm now that DxO is available

The shots are processed in Siril an astrophotography program on mac using standard techniques for deep space.

I then take the results in photoshop for blending as I do not use a one stop shop program like sequator as this only runs on PC

While deep space photography can certainly be done with small sensor and there a number of specialised cameras with MFT sensors nightscapes are actually more challenging

Frankly shooting a single exposure on MFT is very challenging and with all the noise reduction software you want you will go very high in ISO the DR will be low, the stars will clip and your image will look monochrome. You can try editing adding colors, dehaze etc but at the end it will not look as good as blending the shots.

In this field are the various fisheye lenses, the laowa 7.5 and any ultrawide lens

Then you move into stacking. I have maintained a table since a few years of the best lenses for Milky Way photography (the best for single shot being the PL12mm and other lenses 15-17mm good for mosaics or composites with more detail of the Milky Way)

Here MFT has two key issues.

1. When the sensor gets hot you get many pixel defects (hot pixels, glow etc)

2. The lens physical aperture is tiny due to the crop factor

The two issues above are of particular relevance as the 2 is mitigated by having longer exposure times while the first is impacted by exposure times.

In particular thermal noise goes linear with exposure however the benefit of stacking increases with the number of shots and the total exposure

What matters the most is that thermal noise depends more strongly on temperature than exposure time so if you can take your picture below 10C there will not be many pixel defects

I do not use dark frames as those are taken care by a raw editor however if you worked with raw files or dng you would need to take them and those will cure your hot and defective pixels as long as you take them in the same conditions otherwise they normally add noise.

A special mention are PDAF pixels, those show up with the process I use when the image is stressed so if you ever considered using an Olympus camera with PDAF pre OM-1 (there should not be a problem there) you need to take a whole set of flat, bias, dark frames in addition to your exposures otherwise you may end up with posterization issues and other nasty problem. I recommend not using a camera with PDAF and using one with nothing on your sensor. Please note the PDAF issue show up in any situation and are aggravated by thermal effects.

Finally when you do operate in a hot climate and you are taking your Milky Way shots in Thailand or similar there will be a disadvantage and you just need to make sure you don't end up with sensor glow that is hard to fix even with dark frames and destroys your image.

So other than the fact I need to spend more time imaging on the field to get the same result of a full frame camera (in effect 4 as much from 5' to 20') the results are pretty much the same for the Milky Way

Once you work with blending you know that you can take your foreground at twilight (my favourite is the nautical twilight) and this is not actually an issue for the format as the exposures will be short and if you want you can use LENR there are no visible issue once you crunch your exposure with a noise reduction program

So in short if you are prepare to do the work the challenges can be overcome but if you are expecting to snap away and get it looking great with a single shot you are going to face challenges. Obviously quality standards are personal and some people may say what I get is good enough

Finally my consideration is that the GH5M2 is the best camera for this job at present as it retains colors at high ISO this is important if you don't have a tracker an shoot at ISO 2000/2500. The OM-1 looks promising but I have not put my hands on one yet

If you really care about night sky shots buy a full frame like a used canon 6d. Literally night and day difference from even the best m43 at 1/5th the cost.

Not really the 6D is completely outdated today and dslr for night use are terrible with their backscreek live view

the progression would be A7III or Panasonic S5

I had the S5 but there were no lenses so I could not do anything with it and sold it

the 24mm and 35mm would be interesting