Quattro & Merrill Moon Shot

[TN Args]

Told you it wasn't magic. ;-)

 

[1P2P3M90]

Hat hier Jemand auch auf das Objektiv gedacht? Dieses Objektiv ist uralt und noch nicht für die höhere Auflösung des Quattro Sensor gemacht. Höchstens für die 15MB der SD Merrill.

Wenn mann ein zeitgenössisches neues Objektiv des ART Reihe benutzt, wird es sicher mit der Quattro Leistung überrascht.

 

[Roland Karlsson]

Hat hier Jemand auch auf das Objektiv gedacht? Dieses Objektiv ist uralt und noch nicht für die höhere Auflösung des Quattro Sensor gemacht. Höchstens für die 15MB der SD Merrill.

Wenn mann ein zeitgenössisches neues Objektiv des ART Reihe benutzt, wird es sicher mit der Quattro Leistung überrascht.

Yes, that is true.

--
/Roland
Kalpanika X3F tools:

GitHub - Kalpanika/x3f: Tools for manipulating X3F files from Sigma cameras

Tools for manipulating X3F files from Sigma cameras - Kalpanika/x3f

github.com

 

[FDecker]

Somewhere along the way to making an image the files of Merrill series cameras use or get processed by some version of the "haze removal" process based on "unsharp mask sharpening," or in the RT system terms, the "Retinex" filter. The result is cutting through haze of all types, even local fog.

See:

https://rawpedia.rawtherapee.com/Retinex#Retinex_at_the_beginning_of_the_processing

Read the whole section.

This is not entirely that great a mystery. It is an excellent demonstration of why the Merrill cameras series cameras do so well at distances, an effect we see again and again, and the Q system does not produce that effect, because it produces an accurate version of the scene.

Yes, well understood that Merrill fabricates ultra-small detail instead of accuracy.

"Fabricates", you say. Methinks you fabricate ultra-small facts. Fake facts, if you will.

Better understood by some than others, it seems. Besides, I am only agreeing with Richard's last sentence, so suggest you redirect your accustions of lying to the source.

How do you think 'haze removal' works? If you had read the article you might be much more agreeable! You take the original (literal) information which is rather hazy, then you create a series of masks by fabricating new information (positive and negative) for each site based on surrounding information, then you overlay the fabrications to the original. The result is you remove some information (image information loss), and you create some new image information (fabrication).

Merrill famously removes haze. Guess what? It wasn't by magic! We know what has happened.

I have no idea how you come to that conclusion. This is simply made up from nothing.

The SD1 sensor has very high local acutance and accuracy. There is aliasing as with all cameras without AA filter.

 

[FDecker]

Hat hier Jemand auch auf das Objektiv gedacht? Dieses Objektiv ist uralt und noch nicht für die höhere Auflösung des Quattro Sensor gemacht. Höchstens für die 15MB der SD Merrill.

Wenn mann ein zeitgenössisches neues Objektiv des ART Reihe benutzt, wird es sicher mit der Quattro Leistung überrascht.

Das Objektiv wurde für beide Kameras verwendet. 4.9um vs. 4.3um Pixelgröße sind kein hinreichend großer Unterschied.

Übrigens hat Google den Text nach „Deutsch“ übersetzt, nicht nach „English“.

 

[TN Args]

Somewhere along the way to making an image the files of Merrill series cameras use or get processed by some version of the "haze removal" process based on "unsharp mask sharpening," or in the RT system terms, the "Retinex" filter. The result is cutting through haze of all types, even local fog.

See:

https://rawpedia.rawtherapee.com/Retinex#Retinex_at_the_beginning_of_the_processing

Read the whole section.

This is not entirely that great a mystery. It is an excellent demonstration of why the Merrill cameras series cameras do so well at distances, an effect we see again and again, and the Q system does not produce that effect, because it produces an accurate version of the scene.

Yes, well understood that Merrill fabricates ultra-small detail instead of accuracy.

"Fabricates", you say. Methinks you fabricate ultra-small facts. Fake facts, if you will.

Better understood by some than others, it seems. Besides, I am only agreeing with Richard's last sentence, so suggest you redirect your accustions of lying to the source.

How do you think 'haze removal' works? If you had read the article you might be much more agreeable! You take the original (literal) information which is rather hazy, then you create a series of masks by fabricating new information (positive and negative) for each site based on surrounding information, then you overlay the fabrications to the original. The result is you remove some information (image information loss), and you create some new image information (fabrication).

Merrill famously removes haze. Guess what? It wasn't by magic! We know what has happened.

I have no idea how you come to that conclusion. This is simply made up from nothing.

Nothing: the OP described his local atmosphere as slightly hazy when the images were made.

Nothing: Lin Evans posted quite some time ago on how Merrill gives the appearance of detail above Nyqvist. The Quattro has more detail, but the Merrill gives the impression of (simulates) (fabricates) more detail.

Nothing: Richard Stone said the Merrills incorporate a haze removal process. I quoted him. How could you miss it? If you dispute it, take it up with the originator.

I described the implications of that claim. That is all I did. If the Merrill effectively cuts through light haze, then information is being fabricated on the image.

Why do you think the Merrill is providing more detail of the moon, when the Quattro out-resolves it pure and simple?

The SD1 sensor has very high local acutance and accuracy. There is aliasing as with all cameras without AA filter.

Both cameras alias. It does not explain the difference.

 

[FDecker]

Somewhere along the way to making an image the files of Merrill series cameras use or get processed by some version of the "haze removal" process based on "unsharp mask sharpening," or in the RT system terms, the "Retinex" filter. The result is cutting through haze of all types, even local fog.

See:

https://rawpedia.rawtherapee.com/Retinex#Retinex_at_the_beginning_of_the_processing

Read the whole section.

This is not entirely that great a mystery. It is an excellent demonstration of why the Merrill cameras series cameras do so well at distances, an effect we see again and again, and the Q system does not produce that effect, because it produces an accurate version of the scene.

Yes, well understood that Merrill fabricates ultra-small detail instead of accuracy.

"Fabricates", you say. Methinks you fabricate ultra-small facts. Fake facts, if you will.

Better understood by some than others, it seems. Besides, I am only agreeing with Richard's last sentence, so suggest you redirect your accustions of lying to the source.

How do you think 'haze removal' works? If you had read the article you might be much more agreeable! You take the original (literal) information which is rather hazy, then you create a series of masks by fabricating new information (positive and negative) for each site based on surrounding information, then you overlay the fabrications to the original. The result is you remove some information (image information loss), and you create some new image information (fabrication).

Merrill famously removes haze. Guess what? It wasn't by magic! We know what has happened.

I have no idea how you come to that conclusion. This is simply made up from nothing.

Nothing: the OP described his local atmosphere as slightly hazy when the images were made.

So what? It is a long way from that information plus the fact that the SD1 picture is obviously much better in this specific comparison to the conclusion that SPP uses haze removal for the SD1.

Nothing: Lin Evans posted quite some time ago on how Merrill gives the appearance of detail above Nyqvist. The Quattro has more detail, but the Merrill gives the impression of (simulates) (fabricates) more detail.

The Quattro also produces detail above Nyquist as it also has no AA filter.

Nothing: Richard Stone said the Merrills incorporate a haze removal process. I quoted him. How could you miss it? If you dispute it, take it up with the originator.

How can he know this? By analyzing the code? You agreed to it. „Merrill famously removes haze“.

I described the implications of that claim. That is all I did. If the Merrill effectively cuts through light haze, then information is being fabricated on the image.

It doesn‘t do so. The Quattro picture is simply bad in comparison. May not be in general, but here the SD1 wins.

Why do you think the Merrill is providing more detail of the moon, when the Quattro out-resolves it pure and simple?

First of all, the 1:1:4 only out-resolves the SD1 in luminance but not in color. I think I can see from my own pictures that the Q has more noise than the SD1. In addition, there can be focus issues.

The SD1 sensor has very high local acutance and accuracy. There is aliasing as with all cameras without AA filter.

Both cameras alias. It does not explain the difference.

Exactly, see above. This statement is true and contradicts what you write above about Merrill making up false detail above Nyquist but the Quattro having „real“ resolution.

 

[TN Args]

Somewhere along the way to making an image the files of Merrill series cameras use or get processed by some version of the "haze removal" process based on "unsharp mask sharpening," or in the RT system terms, the "Retinex" filter. The result is cutting through haze of all types, even local fog.

See:

https://rawpedia.rawtherapee.com/Retinex#Retinex_at_the_beginning_of_the_processing

Read the whole section.

This is not entirely that great a mystery. It is an excellent demonstration of why the Merrill cameras series cameras do so well at distances, an effect we see again and again, and the Q system does not produce that effect, because it produces an accurate version of the scene.

Yes, well understood that Merrill fabricates ultra-small detail instead of accuracy.

"Fabricates", you say. Methinks you fabricate ultra-small facts. Fake facts, if you will.

Better understood by some than others, it seems. Besides, I am only agreeing with Richard's last sentence, so suggest you redirect your accustions of lying to the source.

How do you think 'haze removal' works? If you had read the article you might be much more agreeable! You take the original (literal) information which is rather hazy, then you create a series of masks by fabricating new information (positive and negative) for each site based on surrounding information, then you overlay the fabrications to the original. The result is you remove some information (image information loss), and you create some new image information (fabrication).

Merrill famously removes haze. Guess what? It wasn't by magic! We know what has happened.

I have no idea how you come to that conclusion. This is simply made up from nothing.

Nothing: the OP described his local atmosphere as slightly hazy when the images were made.

So what? It is a long way from that information plus the fact that the SD1 picture is obviously much better in this specific comparison to the conclusion that SPP uses haze removal for the SD1.

Nothing: Lin Evans posted quite some time ago on how Merrill gives the appearance of detail above Nyqvist. The Quattro has more detail, but the Merrill gives the impression of (simulates) (fabricates) more detail.

The Quattro also produces detail above Nyquist as it also has no AA filter.

Lin was contrasting the two and explaining a difference. The point being, and this is easy to see on a resolution test chart, that when the Merrill goes above Nyqvist it produces fake detail in a way that Quattro doesn't.

Point stands.

Nothing: Richard Stone said the Merrills incorporate a haze removal process. I quoted him. How could you miss it? If you dispute it, take it up with the originator.

How can he know this? By analyzing the code? You agreed to it. „Merrill famously removes haze“.

I described the implications of that claim. That is all I did. If the Merrill effectively cuts through light haze, then information is being fabricated on the image.

It doesn‘t do so. The Quattro picture is simply bad in comparison. May not be in general, but here the SD1 wins.

Why do you think the Merrill is providing more detail of the moon, when the Quattro out-resolves it pure and simple?

First of all, the 1:1:4 only out-resolves the SD1 in luminance but not in color. I think I can see from my own pictures that the Q has more noise than the SD1. In addition, there can be focus issues.

The OP has satisfied himself there were no focus issues, with good technique and multiple exposures. You agree the Quattro would out-resolve the SD1 normally, but not in this case. You say the Merrill doesn't exhibit any haze-reduction effects, which would explain the result if it did. A bit of noise doesn't kill detail -- noise reduction does -- so that's not it.
I'm still hoping the raw files might be instructive -- but no guarantees.

 

[Raist3d]

The Quattro shot even has green/magenta zones all over the moon, on what should be a pretty "B&W" looking object. The Merrill has them too, but much less of a problem and hard to see without increasing the contrast.

People who doubt this, please go ahead and check the shot with a color reader/dropper and see which values you get in different regions for both shots.

Also the Merrill looks decidedly more detailed.

--
Raist3d/Ricardo (Photographer, software dev.)- I photograph black cats in coal mines at night...
“The further a society drifts from truth the more it will hate those who speak it.” - George Orwell

 

[richard stone]

The Quattro shot even has green/magenta zones all over the moon, on what should be a pretty "B&W" looking object. The Merrill has them too, but much less of a problem and hard to see without increasing the contrast.

People who doubt this, please go ahead and check the shot with a color reader/dropper and see which values you get in different regions for both shots.

Also the Merrill looks decidedly more detailed.

 

[TN Args]

The Quattro shot even has green/magenta zones all over the moon, on what should be a pretty "B&W" looking object. The Merrill has them too, but much less of a problem and hard to see without increasing the contrast.

Exactly. I saw it right away. Now why do you think that is? You would "normally" only see those colour blotches in severely underexposed shadow areas.

Interested in those raw files.

People who doubt this, please go ahead and check the shot with a color reader/dropper and see which values you get in different regions for both shots.

Easy to see with the naked eye. And not normal. So, why?

I know at least 6 forum posters whose trembling fingers are dying to write "it's the fault of that evil unclean 1:1:4 architecture!!" But they hesitate because they know they will be asked why all well-exposed near-white areas on a Quattro sensor don't have easily visible colour blotches. And they don't know what to say to that.

Interested in those raw files.

Also the Merrill looks decidedly more detailed.

I agree. And yet it is a less detailed sensor. So is it fake? Or what?

 

[Baumgartner]

In 80-90% of similar comparisons, the Merrill "wins", and in 80-90% of the polls, the Merrill wins. Hope Mr. Yamaki frequently visits this forum.
I'm not saying that the Quattros can't produce satisfying results, but for some reason people have less problems to achieve it with the Merrills.

cheers
B.

 

[Roland Karlsson]

The Quattro shot even has green/magenta zones all over the moon, on what should be a pretty "B&W" looking object. The Merrill has them too, but much less of a problem and hard to see without increasing the contrast.

Exactly. I saw it right away. Now why do you think that is? You would "normally" only see those colour blotches in severely underexposed shadow areas.

The blotches are lurking under the image everywhere and will be stronger and more visible if the actual image is kind of low saturation and not very well exposed.

Those moons are not well exposed. They could easily be exposed 3-4 stops more. They are also, by nature, kind of low saturation.

Also the Merrill looks decidedly more detailed.

I agree. And yet it is a less detailed sensor. So is it fake? Or what?

Both images lacks details.

Now, the full moon is not easy to take a good photo off. It is by way too low contrast, and you need to enhance the contrast to see anything. Now, you have shot noise. Lets say the moon is only 1000 photons per pixel. Then, the shot noise is 3%. And, if the contrast is only 10%, then you have obvious problems. So, exposure to the right is a must. The brightest parts shall be almost over exposed.

And to make things worse, the Quattro is one stop less exposed, increasing the shot noise half a stop.

 

[Roland Karlsson]

In 80-90% of similar comparisons, the Merrill "wins", and in 80-90% of the polls, the Merrill wins. Hope Mr. Yamaki frequently visits this forum.
I'm not saying that the Quattros can't produce satisfying results, but for some reason people have less problems to achieve it with the Merrills.

cheers
B.

Merrill seems to have an upper hand in low contrast comparisons.

 

[Raist3d]

The Quattro shot even has green/magenta zones all over the moon, on what should be a pretty "B&W" looking object. The Merrill has them too, but much less of a problem and hard to see without increasing the contrast.

Exactly. I saw it right away. Now why do you think that is? You would "normally" only see those colour blotches in severely underexposed shadow areas.

Well to be clear: I have seen this a lot. So much that it stopped me from ordering a refurbished Quattro. Not saying I have seen it on every Quattro shot. And on Merrill. Seems like a very narrow band of operation to get the good look. I also after looking over and over, overall, think under good light the Merrill has the best overall IQ. The Quattro has more reasonable IQ at higher ISO.

The DR of both is similar- you can shift the exposure up or down to match with Merrill more in highlights, and Quattro more in shadows.

Interested in those raw files.

People who doubt this, please go ahead and check the shot with a color reader/dropper and see which values you get in different regions for both shots.

Easy to see with the naked eye. And not normal. So, why?

I know at least 6 forum posters whose trembling fingers are dying to write "it's the fault of that evil unclean 1:1:4 architecture!!" But they hesitate because they know they will be asked why all well-exposed near-white areas on a Quattro sensor don't have easily visible colour blotches. And they don't know what to say to that.

Interested in those raw files.

Also the Merrill looks decidedly more detailed.

I agree. And yet it is a less detailed sensor. So is it fake? Or what?

The Merrill when it comes to color, under good light, it is the most detailed sensor But this is certainly a different case. I would have expected the Quattro to pull ahead here in this subject.

 

[Raist3d]

The Quattro shot even has green/magenta zones all over the moon, on what should be a pretty "B&W" looking object. The Merrill has them too, but much less of a problem and hard to see without increasing the contrast.

People who doubt this, please go ahead and check the shot with a color reader/dropper and see which values you get in different regions for both shots.

Also the Merrill looks decidedly more detailed.

 

[xpatUSA]

Lets say the moon is only 1000 photons per pixel.

Why would we say that, Roland, when the camera captures perhaps 50,000 - 77,000 photons, i.e. electrons, when saturated - depending on model.

Were you exaggerating to make a point?

 

[TN Args]

Lets say the moon is only 1000 photons per pixel.

Why would we say that, Roland, when the camera captures perhaps 50,000 - 77,000 photons, i.e. electrons, when saturated - depending on model.

Were you exaggerating to make a point?

 

[TN Args]

The Quattro shot even has green/magenta zones all over the moon, on what should be a pretty "B&W" looking object. The Merrill has them too, but much less of a problem and hard to see without increasing the contrast.

Exactly. I saw it right away. Now why do you think that is? You would "normally" only see those colour blotches in severely underexposed shadow areas.

Well to be clear: I have seen this a lot. So much that it stopped me from ordering a refurbished Quattro. Not saying I have seen it on every Quattro shot. And on Merrill. Seems like a very narrow band of operation to get the good look. I also after looking over and over, overall, think under good light the Merrill has the best overall IQ. The Quattro has more reasonable IQ at higher ISO.

The DR of both is similar- you can shift the exposure up or down to match with Merrill more in highlights, and Quattro more in shadows.

Interested in those raw files.

People who doubt this, please go ahead and check the shot with a color reader/dropper and see which values you get in different regions for both shots.

Easy to see with the naked eye. And not normal. So, why?

I know at least 6 forum posters whose trembling fingers are dying to write "it's the fault of that evil unclean 1:1:4 architecture!!" But they hesitate because they know they will be asked why all well-exposed near-white areas on a Quattro sensor don't have easily visible colour blotches. And they don't know what to say to that.

Interested in those raw files.

Also the Merrill looks decidedly more detailed.

I agree. And yet it is a less detailed sensor. So is it fake? Or what?

The Merrill when it comes to color, under good light, it is the most detailed sensor

So much wishful thinking! Tell you what: you say it often enough, and who knows? It might come true!

But this is certainly a different case. I would have expected the Quattro to pull ahead here in this subject.

So what happened?

 

[raylob]

So much wishful thinking! Tell you what: you say it often enough, and who knows? It might come true!

This was similar to Tony Blair's strategy when he was in power as a labour Prime Minister from 1997 to 2007. Say something often enough and the electorate will believe you. That failed!!