Seeing any Posterization issues with A7r ll?

[GrahamWelland]

Btw is this specific to the A7RII or all of the cameras using Sony's imaging pipeline, compression and raw format?

 

[mjm6]

> add anything to the web knowledge base
Lloyd presented experimental data and demonstrated several important issues:
- imperfection of default colour management;
- importance of watching for per channel exposure;
- importance of watching for extreme attenuation of the red channel due to the use of polarizing filters (falls somewhat into the per channel exposure);
- complete voids in raw data coming from the camera, stretched to 7-8 missing values after applying white balance (posterization of raw data).

I would say it is quite a lot.

The top three are known troubles associated with digital imaging and have been for years.

The bottom is clearly an issue, but is it the fault of the camera or the top three? I've seen gamut manipulations do some pretty drastic appearing things on the fringes, and at this point I don't believe that this is really anything beyond that unless there has been an effort to prove that the compression is the culprit. It may be, but it is also not inconsistent with how digital files have behaved from other cameras under other circumstances in the past when the gamut it strongly challenged.


---Michael

 

[Amin]

> add anything to the web knowledge base
Lloyd presented experimental data and demonstrated several important issues:
- imperfection of default colour management;
- importance of watching for per channel exposure;
- importance of watching for extreme attenuation of the red channel due to the use of polarizing filters (falls somewhat into the per channel exposure);
- complete voids in raw data coming from the camera, stretched to 7-8 missing values after applying white balance (posterization of raw data).

I would say it is quite a lot.

If Lloyd had written a post presenting the above issues in a level-headed manner, he would have gotten a lot of positive feedback. Instead, he talked about how he was "flabbergasted" by the results of an ISO 100 image taken with a "nearly optimal exposure" and suggested that these results were because of the compression scheme used by "a company that really doesn't understand photography".

 

[tashley]

The top three are known troubles associated with digital imaging and have been for years.

The bottom is clearly an issue, but is it the fault of the camera or the top three? I've seen gamut manipulations do some pretty drastic appearing things on the fringes, and at this point I don't believe that this is really anything beyond that unless there has been an effort to prove that the compression is the culprit. It may be, but it is also not inconsistent with how digital files have behaved from other cameras under other circumstances in the past when the gamut it strongly challenged.


---Michael

Michael, I think the truth is, we'll never know. I think you can discount factor 1) because the RAW file is quite badly posterised before colour management. The exposure was actually quite good, protecting the highlights on the face, but were it to have been bracketed so that the most optimally ETTR file could have been chosen, maybe things would have been different. However at this level of exposure one would not expect posterisation, or at least, I wouldn't. And I think in practical terms most of us don't expect to have to shoot with a magenta filter though I am sure that would have helped. The polariser is the joker in the pack.

One may or may not like Lloyd's style of presentation (Amin clearly does not) but as Iliah says, quite a lot has been demonstrated and I think we have all exercised our brains and learned some new stuff here.

My read of Iliah (he and his colleague Alex are they only people apart from Lloyd and me to have seen the RAW file, but I consider him a very great deal more qualified than me to judge it) is that he thinks the some aspect of the RAW file pipeline is partly to blame, quite possibly regardless of the other factors. I am sure he can answer that directly himself and I would love to hear his answer, even if it's "It isn't possible to say without more information."

 

[rayyan]

The raw pipeline!

Here is a quote from a source that shall remain nameless

"Even though Sony has promised 14 bit raw, it isn’t going to fix the hardware compression that’s occurring before recording – we’re just going to have a larger container (i.e. bigger files) for the same amount of information."

Could there be some merit in the above statement? Hey, I just use raw...I don't know ( I leave the heavy stuff to the experts ) the intricacies of it all. Get .nef files
Or whatever, open in a converter, do some manipulation and like magic it seems to be there.

Kind regards.

 

[spence]

The raw pipeline!
"Even though Sony has promised 14 bit raw, it isn’t going to fix the hardware compression that’s occurring before recording – we’re just going to have a larger container (i.e. bigger files) for the same amount of information."
Kind regards.

I think that's unlikely...they would at least have to ditch the delta compression, or it would be an utterly pointless exercise. Presenting the same lossy data, in an uncompressed format, would solve nothing, and Sony knows that.

They might still cook the RAW (e.g. we might still have orange peel textures to deal with), but losing the RAW artifacts along high contrast borders caused by delta compression would be better than nothing.

 

[mjm6]

The raw pipeline!

Here is a quote from a source that shall remain nameless

"Even though Sony has promised 14 bit raw, it isn’t going to fix the hardware compression that’s occurring before recording – we’re just going to have a larger container (i.e. bigger files) for the same amount of information."

Could there be some merit in the above statement? Hey, I just use raw...I don't know ( I leave the heavy stuff to the experts ) the intricacies of it all. Get .nef files
Or whatever, open in a converter, do some manipulation and like magic it seems to be there.

Kind regards.

Whoa, wait a minute... I think you are conflating lossless compression with lossy compression.

Nobody here is talking about lossless compression that I am aware of. This is entirely regarding the lossy compressed RAW files that Sony uses to drop the file size (and possibly some other artifact from the image processing pathways, who knows). As Spence mentions, the Delta compression.

If they implement a lossless compression scheme that reduces file size but does not actually eliminate information that was available to record, that is the question, and it is also what many, many people feel is the right thing to do. Me included. Give me larger files to know that at least the image file has everything that the device was able to collect.


---Michael

 

[mjm6]

Michael, I think the truth is, we'll never know...

Thanks, I agree. I think the combination of a really blue/green lake, high altitude, and a polarizer has resulted in a torture test that possibly many (all?) cameras will fail without taking special precautions.

I guess this was my point. If this were to occur on a scene that didn't have these very unique color gamut kinds of problems, I would be concerned. But this problem he generated isn't really shown to be that yet. I think he would have been better advised to explore the situation more before making the posts that he did. Seems a bit Chicken Little to me.


---Michael

 

[Malina DZ]

Since it is reasonable to assume that the Sony engineers are not idiots, they have chosen the processing pipeline for a reason. I am not clear what it is? Card space saving? Frame rate? ... ? Does anyone else?

How about priorities over video capture speed and resolution? Lossy compression allows Sony to achieve not only 12FPS for 24MPx stills (a77), but also record video files to SD card at a higher frame rate and resolution compared to Canikon DSLRs.

 

[Guy Mancuso]

Thanks, I agree. I think the combination of a really blue/green lake, high altitude, and a polarizer has resulted in a torture test that possibly many (all?) cameras will fail without taking special precautions.

I guess this was my point. If this were to occur on a scene that didn't have these very unique color gamut kinds of problems, I would be concerned. But this problem he generated isn't really shown to be that yet. I think he would have been better advised to explore the situation more before making the posts that he did. Seems a bit Chicken Little to me.


---Michael

This is exactly my point all along.

 

[lambert]

Thanks, I agree. I think the combination of a really blue/green lake, high altitude, and a polarizer has resulted in a torture test that possibly many (all?) cameras will fail without taking special precautions.

I guess this was my point. If this were to occur on a scene that didn't have these very unique color gamut kinds of problems, I would be concerned. But this problem he generated isn't really shown to be that yet. I think he would have been better advised to explore the situation more before making the posts that he did. Seems a bit Chicken Little to me.


---Michael

Diglloyd generates traffic $$$ by making mountains out of mole-hills.

 

[ErikKaffehr]

Hi Tim,

From what I can see in the samples it is not entirely clear what phenomena are attributed to the "orange peel" effect and what is called posterisation.

Much of the area is out of focus so it can be hard to say what is a wave and what is a possible artefact.

Something I noticed is that both Tim Ashley, Lloyd Chambers and myself apply quite a bit intensive sharpening, which is often combined with some masking. I have noticed with my settings that this can cause artefacts. Large amount of low radius sharpening enhances noise. The masking removes sharpening in parts of the image, causing the visible grain pattern to switch on and off. This can often be observed as grain pattern in contours along edges.

In my humble opinion we are sharpening quite a lot at the pixel level which is really seldom optimal. Pictures are not intended to be viewed at the pixel level and prints probably need most sharpening in the low to medium frequencies, say 10-30 lp/mm on 135 film, where the contrast sensivity of human vision is highest. But, he who is without sin may throw the first stone, and that will certainly not be me…

Best regards
Erik

Yes you are right. This is due to molecules scattering higher frequencies of light from the sun (Rayleigh or Tyndall Scattering). However this blue light has to travel through the atmosphere to get to your eye and hence it gets attenuated itself (the blue light trying to get to your sky get attenuated by scattering). Add in dust scatter and humidity reduced viewing distance (haze) and the lower you are the more non-blue light reaches you as well.

High up you get a deep blue because less blue is scattered but there is less (also thinner) atmosphere to pollute the blue.

Add a polariser to this and you can end up with the sky almost black (you effectively have a luminance control of the blue channel with your polariser).

Reflections from waves will alternately reflect the black, polarised sky and possible areas of non-polarised sky (and maybe transparent areas showing the green of the water).

Tim

p.s. Dana Lake is at 11,100 ft

 

[pegelli]

Hi Tim,
Something I noticed is that both Tim Ashley, Lloyd Chambers and myself apply quite a bit intensive sharpening, which is often combined with some masking. I have noticed with my settings that this can cause artefacts. Large amount of low radius sharpening enhances noise. The masking removes sharpening in parts of the image, causing the visible grain pattern to switch on and off. This can often be observed as grain pattern in contours along edges.

In my humble opinion we are sharpening quite a lot at the pixel level which is really seldom optimal. Pictures are not intended to be viewed at the pixel level and prints probably need most sharpening in the low to medium frequencies, say 10-30 lp/mm on 135 film, where the contrast sensivity of human vision is highest. But, he who is without sin may throw the first stone, and that will certainly not be me…

Best regards
Erik

Hi Erik, I think no sins are committed so no stones need to be thrown. But you raise an interesting point. I've learned most of my sharpening from reading a lot of material by Bruce Frazer and Jeff Schewe who tell us to do as little "capture sharpening" as possible and then do your creative and final sharpening specifically for the output size and medium you're preparing the file for (print or screen). Since the sharpening settings in Lightroom (and I think C1 as well) are "capture sharpening" I do those very mildly and not far from the default settings. Just enough to judge if a file is sharp enough to further process or not. This way I hardly ever have problems with exaggerating noise and/or orange peel effect. After export from LR to prepare the file for printing or web showing I do a final round of noise reduction (only when required) followed by output sharpening specific for the output size and medium. I like the results this workflow gives me and wonder what advantages you (and may be the others you mention) see in doing a very aggressive capture sharpening, since that tends to exaggerate noise and other artifacts that are present in the file. What do you think I am missing by doing only a very mild capture sharpening?

 

[timparkin]

Hi Tim,

From what I can see in the samples it is not entirely clear what phenomena are attributed to the "orange peel" effect and what is called posterisation.

In search of this 'orange peel' phenomenon I downloaded the test raw files for D810 and A7R from DPReview's studio comparison and then processed them in rawdigger using 2x2 (pixel binning i.e. without raw demosaic'ing) and I can't find any difference between them. I used a test patch (third grey from the left) on the colourchecker target at the top of the image.

The Sony's red/blue channels appear a little darker possibly but even accounting for that there is no difference.

Anybody suggest I should look somewhere else for this orange peel?

I'm also writing some software to analyse raw files to see if posterisation is causing visible differences (using delta E 2000 taken from a sample processed raw to compare with). Hopefully it will create a contour map of areas of posterisation (multiple contiguous pixels with identical numbers adjacent to another patch of multiple contiguous identical numbers). this can't guarantee picking up posterisation but it will point to possible areas to look at. It may make sense to correlate this with the delta compression values from rawdigger.

This seems more common sense than guessing if something is a visible problem or not.

Thoughts?

 

[Guy Mancuso]

Tim orange peel seems to be related to ACR and LR as folks using C1 are not getting that effect. Not sure what is causing it myself but it seems related to sharpening in those programs.

 

[Amin]

Like Guy, I've never seen orange peel effects in files processed by C1. I have seen them in LR. However, Tim Ashley has seen orange peel from Sony files that were processed to TIFF without sharpening in C1 and then subsequently sharpened in LR, whereas he doesn't see it under any circumstances with D810 files. This is very interesting. I'd like to know what causes it.

Once I get a break, I'm gonna do a lot of D810 - A7RII comparing. For me, all this stuff doesn't matter a whit for my photography. Artifact peeping is a separate hobby unto itself, but it's one I enjoy.

 

[timparkin]

Like Guy, I've never seen orange peel effects in files processed by C1. I have seen them in LR. However, Tim Ashley has seen orange peel from Sony files that were processed to TIFF without sharpening in C1 and then subsequently sharpened in LR, whereas he doesn't see it under any circumstances with D810 files. This is very interesting. I'd like to know what causes it.

Once I get a break, I'm gonna do a lot of D810 - A7RII comparing. For me, all this stuff doesn't matter a whit for my photography. Artifact peeping is a separate hobby unto itself, but it's one I enjoy.

Hi Guy/Amin,

I'll have a look at both files post processed in L1 the same way and compare them with the actual raw data..

Tim

 

[Iliah Borg]

This is exactly my point all along.

The file rendered into ProPhoto RGB is acceptably clean, needs only minor retouch on small spots, done in less than 4 minutes. The conversion to Adobe RGB, however, is the whole different story. Concentrating on the camera and shooting conditions shifts the accent into the wrong direction here.

 

[Guy Mancuso]

Hi Guy/Amin,

I'll have a look at both files post processed in L1 the same way and compare them with the actual raw data..

Tim

Thanks Tim. It is really weird to be honest as I am not seeing that effect at all in C1. If I push sharpness really heavy for instance I may see more crosshair look but you really have to push it.

 

[Guy Mancuso]

The file rendered into ProPhoto RGB is acceptably clean, needs only minor retouch on small spots, done in less than 4 minutes. The conversion to Adobe RGB, however, is the whole different story. Concentrating on the camera and shooting conditions shifts the accent into the wrong direction here.

Point taken. I guess the whole thing that throws this all in the air are the polarizer and the altitude than with color management issues, maybe some compression issues this file was just destined to fail. Some may want to blame one thing but I really just think its a combination of circumstances. Or maybe better said I could not take this file to Sony and say your compression scheme alone completely ruined it and that was kind of what was said. So I could not confront Sony based on what everyone has found. Its just not a single issue.

Other files where compression was the issue. Absolutely present that to Sony this one Im sorry it would be wrong.

Those are my thoughts. Right or wrong it would be uncomfortable for me at least given what we know now to blame one thing.