Talk me off the ledge.

dmanthree said:

Anders W said:

dmanthree said:

Anders W said:

dmanthree said:

Based on what I've seen, I like the output from the D810 better than the hi-res mode. It looks sharper with less noise.

Click to expand...

Specifically what is it that you have seen that makes you think the D810 images are sharper and have less noise?

Click to expand...

My opinion is based on samples I've seen.

Click to expand...

Yes you said that much. Please point us to one or more pertinent examples of samples you have seen.

Click to expand...

http://www.dpreview.com/reviews/olympus-om-d-e-m5-ii/9

Scroll to the "artist's easel" and look at the detail on the D810. Then increase the ISO (the Oly is limited here) and see how much more detail the D810 exhibits. Look at raw only.

Click to expand...

I see what you mean with regard to detail. However, what you may regard as an asset of the D810 image, I regard as a liability. The reason is that while the D810 renders more of the detail of the back of that easel, it doesn't render it well. The detail is too fine for the sampling frequency of the D810 sensor and the result is aliasing and moiré.

Below are 100-percent crops from that scene, processed from RAW, that hopefully shows what I have in mind. For the Nikon, the input is the ISO 64 RAW (daylight version) and the output is ACR default, except that I pulled back "exposure" half a stop to make the brightness of the two crops more similar. For the Olympus hi-res file, the input is the ISO 200 RAW (daylight version) and the output is ACR default save for sharpening, which was pulled back to zero. In PS, I then downsampled the Olympus hi-res file to the same height as the D810 file using "bicubic sharper" and then sharpened it with Smart Sharpen (amount 400%, radius 0.5, no NR). The reason for doing the sharpening differently for the two images is of course that the D810 file does not really benefit from more sharpening than ACR default (it just makes the artifacts more visible) whereas the Olympus hi-res file needs, and can take, a significant amount of sharpening.

Please click "original size" below the crops to view them properly.



When I look at the image as a whole, not just this scene, my conclusions are as follows:

1. For luminance detail below a certain spatial frequency (not so fine detail), the two are roughly on a par.

2. For luminance detail above that frequency threshold (really fine detail), the E-M5 II renders less but nicely whereas the D810 renders more but poorly (lots of artifacts). I clearly prefer the former to the latter.

3. For color detail, the E-M5 II is superior on all counts: Sharper color detail and cleaner color detail.

As to your point about looking at RAW only: Although that's what I usually do when checking out the DPR studio scene, I don't think it's very helpful in this particular case. The reason is that the E-M5 II hi-res RAW is not a genuine "unprocessed" RAW but just a way to pack the data after stacking the eight input shots. It therefore requires different PP than ordinary RAWs to look as it should.

As to your point about higher ISO: I didn't look at that since I can't really imagine why anyone would want to shoot the E-M5 II at anything but base ISO in hi-res mode. But I doubt that the E-M5 II hi-res mode would be disadvantaged in such a comparison. Due to it being based on eight shots, it will be ahead for signal-noise performance by about 0.5 EV in the shadows and by about one EV in the midtones up compared to a single shot from the D810 at the same ISO.

dmanthree said:

dmanthree said:

dmanthree said:

Besides, some of the subject matter I'm shooting will be moving, so the hi-res feature of the EM-5II wouldn't work.

Click to expand...

Yes, I noticed that you said that. But while that may make the first point moot for you, it need not do so for others.

Click to expand...

Right. But I was inquiring for me, not someone else. There's no question that the Oly will work in many situations, but not mine. No big deal, really, just pick the tools you need.

Click to expand...

Sure.

Most high-end Nikon camera shooters don't like the cheaper plastic ones. Most I know use m4/3 or Fuji. But maybe the D5500 with a good lens will focus well in dimmer light?

Anders W said:

dmanthree said:

Scroll to the "artist's easel" and look at the detail on the D810. Then increase the ISO (the Oly is limited here) and see how much more detail the D810 exhibits. Look at raw only.

Click to expand...

2. For luminance detail above that frequency threshold (really fine detail), the E-M5 II renders less but nicely whereas the D810 renders more but poorly (lots of artifacts). I clearly prefer the former to the latter.

Click to expand...

This is a really interesting point. When the HiRes reaches its limit, it fails gracefully. When an AA-less camera like the D810 reaches it limit (single frame), it fails gracelessly sometimes (color moire and maze artifacts) and falsely other times (detail that isn't really there). The picture frame illustrates this difference nicely. Since the source is an engraving with lots of lines to represent shading, the D810's "lines" appear to be real. Most of them actually aren't real detail as you can see by comparing it to the Phase One IQ "benchmark" image below. Instead, they're diagonal moire constructs. The EM5II HiRes rendering on the other hand doesn't create false detail, it simply blurs out as if it were using an AA filter. Sometimes the false detail is acceptable and maybe even good, but more often than not it will be associated with problems like the maze artifacting that's very visible in the D810 frame. The bottom line is that most - if not all - of the additional high frequency detail displayed in the D810 shot isn't real, so it's rather problematic to say that it has "more" detail than the HiRes shot.


Left: Phase One IQ 180 Benchmark; Center: crop from Anders' EM5II HiRes rendering; Right: crop from Anders' D810 rendering

knickerhawk said:

Anders W said:

dmanthree said:

Scroll to the "artist's easel" and look at the detail on the D810. Then increase the ISO (the Oly is limited here) and see how much more detail the D810 exhibits. Look at raw only.

Click to expand...

2. For luminance detail above that frequency threshold (really fine detail), the E-M5 II renders less but nicely whereas the D810 renders more but poorly (lots of artifacts). I clearly prefer the former to the latter.

Click to expand...

This is a really interesting point.

Click to expand...

I think so too. While this is not the first time I think about it, this is the clearest illustration I have come across so far.

knickerhawk said:

When the HiRes reaches its limit, it fails gracefully. When an AA-less camera like the D810 reaches it limit (single frame), it fails gracelessly sometimes (color moire and maze artifacts) and falsely other times (detail that isn't really there). The picture frame illustrates this difference nicely. Since the source is an engraving with lots of lines to represent shading, the D810's "lines" appear to be real. Most of them actually aren't real detail as you can see by comparing it to the Phase One IQ "benchmark" image below. Instead, they're diagonal moire constructs. The EM5II HiRes rendering on the other hand doesn't create false detail, it simply blurs out as if it were using an AA filter. Sometimes the false detail is acceptable and maybe even good, but more often than not it will be associated with problems like the maze artifacting that's very visible in the D810 frame. The bottom line is that most - if not all - of the additional high frequency detail displayed in the D810 shot isn't real, so it's rather problematic to say that it has "more" detail than the HiRes shot.

Click to expand...

Thanks for taking the trouble to add the sample from the PhaseOne IQ 180 to the set. I sometimes use it for this purpose too, i.e., that of getting an idea of what the things we see photographed are really like. I actually looked at it in this particular case too before posting but was too lazy to do what you did.

While the availability of the IQ 180 samples is a nice thing, it would be even better if DPR would provide a set of close-ups for reference, each of which showing only a small portion of the total studio scene, say 1/12 or so. As this example illustrates, we do need to know what things actually look like in order to judge "detail" properly.

One other thought lingering in my mind is whether the current trend toward more and more AA-less cameras is such a good idea. One possibility is that the AA reappears on Olympus bodies via IBIS-shaking of a slightly different kind than that used for the hi-res mode. Perhaps, we might even get the option to regulate its strength that way. According to the rumors I have picked up, Pentax is about to copy the Olympus (and Hasselblad before that) hi-res solution so it would be only right if Olympus copied the Pentax AA-shaking solution. ;-)

Alternatively, we might of course hope that the Olympus hi-res mode quickly develops into something that can be used in hand-held shooting. I must admit that I am not particularly optimistic that this will happen any time soon, but I certainly hope to be proven wrong. ;-)

knickerhawk said:

Left: Phase One IQ 180 Benchmark; Center: crop from Anders' EM5II HiRes rendering; Right: crop from Anders' D810 rendering

Click to expand...

knickerhawk said:

Sigh. I explained in the first posting what I did and how the comparison was to the reference sRGB color settings for the Macbeth chart. The two strips overlaid on the reference colors were from the DP1 and the D800. I've clarified for you that they are also taken from standard/default jpegs produced by the cameras and posted on Imaging Resources. The EXIF data indicates that the Sigma is set to Picture Mode=Standard, Saturation=Normal. The D800 is set to Picture Control=Standard, Saturation=Normal. They custom white balance to the Macbeth gray patch in these studio scene shots. If you need to know anything else, you can go look at the EXIFs yourself, which I recommend you do to ensure that I'm not "weaseling" you further.

Click to expand...

I see now. I should have asked you about those parameters before making comments.

knickerhawk said:

See above. Yes, one can always apply different picture/color modes and other in-camera adjustments that affect color rendering in the jpegs. Maybe the Neutral color mode for the DP1 would do better than the Standard mode, but it purports to reduce saturation and contrast, which is hardly the only or most significant problem I see with the sampled Sigma color patches in my little experiment. Feel free to post your own shot of a macbeth chart in neutral mode that shows better color accuracy across the chart, and then your currently speculative guess will no longer be a guess.

Click to expand...

In my experience with Merrills they commonly produce slightly under-saturated reds (opposite to greenish-blue), which gives it this overall unique look. Adding some pure red makes the images look indistinguishable from Nikon's, which in your sampling show as more accurate.

As to standard colour mode, I would not relate to any of it, as I never use it. It may turn to be the most accurate, yet the least appealing most of the time.

knickerhawk said:

knickerhawk said:

Truth be told I would not want to debate Sigma colors against Nikons, as against any cameras,

Click to expand...

And yet I'm the one who's supposedly "weaseling" here...

knickerhawk said:

but I do know that I do like what comes out of Sigmas most of the time, and regardless of the subject. I shoot RAW exclusively, and from what I see I regard those cameras as quite a viable alternative if detail (that you don not need to carry a tripod for) and colors is your primary goal.

Click to expand...

I'm not disputing that you get colors. Of course, you get "colors" from the camera (well, at base ISO at least)! The question is accuracy of the colors. That's what The Suede was addressing in the quote provided by Anders and the reason why I did the little sampling experiment.

Click to expand...

The way I read it his comment is pure bashing (of what he can not have to save his life) - the internet is ablaze of these guys. That's why I asked if I could find any of his images on that other thread where the posters showed their appreciation for the camera and the technology it embraces (or any other to that effect). The question was as loaded as can be, yet the *online expert* appeared not to have a slightest idea of what it was about - are you surprised?

Sergey_Green said:

Sergey_Green said:

I'm not disputing that you get colors. Of course, you get "colors" from the camera (well, at base ISO at least)! The question is accuracy of the colors. That's what The Suede was addressing in the quote provided by Anders and the reason why I did the little sampling experiment.

Click to expand...

The way I read it his comment is pure bashing (of what he can not have to save his life) - the internet is ablaze of these guys.

Click to expand...

Speaking of the The_Suede, what do you think of the photography of bobn2 and Jack Hogan?

Sergey_Green said:

That's why I asked if I could find any of his images on that other thread where the posters showed their appreciation for the camera and the technology it embraces (or any other to that effect). The question was as loaded as can be, yet the *online expert* appeared not to have a slightest idea of what it was about - are you surprised?

Click to expand...

I asked you here on what ground the question was relevant and you failed to come up with an answer.

On top of that, why would a guy who has tested the camera and found it worse than anything else he has encountered, color-wise, first use it and then go on the Internet and tell everyone how much he loves it, thereby contradicting himself?

Anders W said:

Speaking of the The_Suede, what do you think of the photography of bobn2 and Jack Hogan?

Click to expand...

I knew you would ask. If I may answer with question - what do you think the most striking difference on the forums between them and you?

Anders W said:

I asked you here on what ground the question was relevant and you failed to come up with an answer.

On top of that, why would a guy who has tested the camera and found it worse than anything else he has encountered, color-wise, first use it and then go on the Internet and tell everyone how much he loves it, thereby contradicting himself?

Click to expand...

I haven't 'seen' - I've measured actual metameric qualities in the Lab myself. And the Foveons still suck, bigtime. In fluorescent light they're so bad that they're incorrigibly off. In sunshine, they're 'just' 2x worse than anything currently at the market, including cellphones and compact cameras. There is no physical way to make that sensor color-correct.
There is in reality no way to make several objects in front of the lens get the correct color - at the same time, with just one color profile. The definition of metamerism.

Fanboy opinions do not negate facts. And the facts is that Foveon is so badly matched to human vision that it's impossible to color correct it.

In translation

The Monkey and the Spectacles

Sergey_Green said:

Anders W said:

Speaking of the The_Suede, what do you think of the photography of bobn2 and Jack Hogan?

Click to expand...

I knew you would ask.

Click to expand...

That is not an answer to my question. Now go ahead and answer it. What do you think of the photography of bobn2 and Jack Hogan?

Sergey_Green said:

If I may answer with question - what do you think the most striking difference on the forums between them and you?

Click to expand...

Unlike mine, that question is irrelevant to the argument you tried to make.

Sergey_Green said:

Sergey_Green said:

I asked you here on what ground the question was relevant and you failed to come up with an answer.

On top of that, why would a guy who has tested the camera and found it worse than anything else he has encountered, color-wise, first use it and then go on the Internet and tell everyone how much he loves it, thereby contradicting himself?

Click to expand...

I haven't 'seen' - I've measured actual metameric qualities in the Lab myself. And the Foveons still suck, bigtime. In fluorescent light they're so bad that they're incorrigibly off. In sunshine, they're 'just' 2x worse than anything currently at the market, including cellphones and compact cameras. There is no physical way to make that sensor color-correct.
There is in reality no way to make several objects in front of the lens get the correct color - at the same time, with just one color profile. The definition of metamerism.

Fanboy opinions do not negate facts. And the facts is that Foveon is so badly matched to human vision that it's impossible to color correct it.

In translation

The Monkey and the Spectacles

Click to expand...

Thanks for demonstrating that you have not even understood the question.

Sergey_Green said:

knickerhawk said:

Sigh. I explained in the first posting what I did and how the comparison was to the reference sRGB color settings for the Macbeth chart. The two strips overlaid on the reference colors were from the DP1 and the D800. I've clarified for you that they are also taken from standard/default jpegs produced by the cameras and posted on Imaging Resources. The EXIF data indicates that the Sigma is set to Picture Mode=Standard, Saturation=Normal. The D800 is set to Picture Control=Standard, Saturation=Normal. They custom white balance to the Macbeth gray patch in these studio scene shots. If you need to know anything else, you can go look at the EXIFs yourself, which I recommend you do to ensure that I'm not "weaseling" you further.

Click to expand...

I see now. I should have asked you about those parameters before making comments.

Click to expand...

OK. Maybe next time you'll pause before you jump to conclusions and make ridiculous accusations about me like:

As much as you are trying to weasel through it, you have no idea what you are talking about, and it shows.

Sergey_Green said:

Sergey_Green said:

See above. Yes, one can always apply different picture/color modes and other in-camera adjustments that affect color rendering in the jpegs. Maybe the Neutral color mode for the DP1 would do better than the Standard mode, but it purports to reduce saturation and contrast, which is hardly the only or most significant problem I see with the sampled Sigma color patches in my little experiment. Feel free to post your own shot of a macbeth chart in neutral mode that shows better color accuracy across the chart, and then your currently speculative guess will no longer be a guess.

Click to expand...

In my experience with Merrills they commonly produce slightly under-saturated reds (opposite to greenish-blue), which gives it this overall unique look. Adding some pure red makes the images look indistinguishable from Nikon's, which in your sampling show as more accurate.

Click to expand...

I'm dubious of the simplicity of your solution, but it's neither important enough to me personally or practical for me to test this claim.

Sergey_Green said:

As to standard colour mode, I would not relate to any of it, as I never use it. It may turn to be the most accurate, yet the least appealing most of the time.

Click to expand...

That's fine, but this discussion wasn't about what's most pleasing, it was about what's most accurate. In art repro work (I've done a little), the worst thing you could tell the artist for whom you're producing images is something like: "Well, even though it isn't accurate, I find the color cast in this image makes your painting more pleasing to look at."

Sergey_Green said:

Sergey_Green said:

Sergey_Green said:

Truth be told I would not want to debate Sigma colors against Nikons, as against any cameras,

Click to expand...

And yet I'm the one who's supposedly "weaseling" here...

Sergey_Green said:

but I do know that I do like what comes out of Sigmas most of the time, and regardless of the subject. I shoot RAW exclusively, and from what I see I regard those cameras as quite a viable alternative if detail (that you don not need to carry a tripod for) and colors is your primary goal.

Click to expand...

I'm not disputing that you get colors. Of course, you get "colors" from the camera (well, at base ISO at least)! The question is accuracy of the colors. That's what The Suede was addressing in the quote provided by Anders and the reason why I did the little sampling experiment.

Click to expand...

The way I read it his comment is pure bashing (of what he can not have to save his life) - the internet is ablaze of these guys. That's why I asked if I could find any of his images on that other thread where the posters showed their appreciation for the camera and the technology it embraces (or any other to that effect). The question was as loaded as can be, yet the *online expert* appeared not to have a slightest idea of what it was about - are you surprised?

Click to expand...

You do realize that "the Suede" is a professional color scientist and color printing engineer, right? Frankly, I don't care whether he's all thumbs when it comes to taking pictures. This discussion was NEVER about whether the Sigmas allow photographers to take pleasing pictures. (I'll take your word Sigma owner for that.) It was ALWAYS about whether the Sigmas are good tools for art reproduction (and hence color accuracy). So, your "loaded question" is completely irrelevant. Hence, Ander's perplexity.

knickerhawk said:

Sergey_Green said:

knickerhawk said:

Sigh. I explained in the first posting what I did and how the comparison was to the reference sRGB color settings for the Macbeth chart. The two strips overlaid on the reference colors were from the DP1 and the D800. I've clarified for you that they are also taken from standard/default jpegs produced by the cameras and posted on Imaging Resources. The EXIF data indicates that the Sigma is set to Picture Mode=Standard, Saturation=Normal. The D800 is set to Picture Control=Standard, Saturation=Normal. They custom white balance to the Macbeth gray patch in these studio scene shots. If you need to know anything else, you can go look at the EXIFs yourself, which I recommend you do to ensure that I'm not "weaseling" you further.

Click to expand...

I see now. I should have asked you about those parameters before making comments.

Click to expand...

OK. Maybe next time you'll pause before you jump to conclusions and make ridiculous accusations about me like:

As much as you are trying to weasel through it, you have no idea what you are talking about, and it shows.

Click to expand...

In this exchange I was wrong.

knickerhawk said:

knickerhawk said:

knickerhawk said:

See above. Yes, one can always apply different picture/color modes and other in-camera adjustments that affect color rendering in the jpegs. Maybe the Neutral color mode for the DP1 would do better than the Standard mode, but it purports to reduce saturation and contrast, which is hardly the only or most significant problem I see with the sampled Sigma color patches in my little experiment. Feel free to post your own shot of a macbeth chart in neutral mode that shows better color accuracy across the chart, and then your currently speculative guess will no longer be a guess.

Click to expand...

In my experience with Merrills they commonly produce slightly under-saturated reds (opposite to greenish-blue), which gives it this overall unique look. Adding some pure red makes the images look indistinguishable from Nikon's, which in your sampling show as more accurate.

Click to expand...

I'm dubious of the simplicity of your solution, but it's neither important enough to me personally or practical for me to test this claim.

Click to expand...

From the files I regularly open it appears to be exactly so. It has been observed by the reviewers as well.

knickerhawk said:

knickerhawk said:

As to standard colour mode, I would not relate to any of it, as I never use it. It may turn to be the most accurate, yet the least appealing most of the time.

Click to expand...

That's fine, but this discussion wasn't about what's most pleasing, it was about what's most accurate. In art repro work (I've done a little), the worst thing you could tell the artist for whom you're producing images is something like: "Well, even though it isn't accurate, I find the color cast in this image makes your painting more pleasing to look at."

Click to expand...

Well, kind of. Glossy colorful magazines (about the art) do often look just as pleasing (and very attractive) as what they show, even though they are seldom accurate. For a good reason.

knickerhawk said:

knickerhawk said:

knickerhawk said:

knickerhawk said:

Truth be told I would not want to debate Sigma colors against Nikons, as against any cameras,

Click to expand...

And yet I'm the one who's supposedly "weaseling" here...

knickerhawk said:

but I do know that I do like what comes out of Sigmas most of the time, and regardless of the subject. I shoot RAW exclusively, and from what I see I regard those cameras as quite a viable alternative if detail (that you don not need to carry a tripod for) and colors is your primary goal.

Click to expand...

I'm not disputing that you get colors. Of course, you get "colors" from the camera (well, at base ISO at least)! The question is accuracy of the colors. That's what The Suede was addressing in the quote provided by Anders and the reason why I did the little sampling experiment.

Click to expand...

The way I read it his comment is pure bashing (of what he can not have to save his life) - the internet is ablaze of these guys. That's why I asked if I could find any of his images on that other thread where the posters showed their appreciation for the camera and the technology it embraces (or any other to that effect). The question was as loaded as can be, yet the *online expert* appeared not to have a slightest idea of what it was about - are you surprised?

Click to expand...

You do realize that "the Suede" is a professional color scientist and color printing engineer, right?

Click to expand...

I would assume Sigma engineers are no less professional in that regard.

knickerhawk said:

Frankly, I don't care whether he's all thumbs when it comes to taking pictures.

Click to expand...

I don't think he can be, he has neither need nor appreciation for it. Why would a man with bad hearing be posting on the forums how much a certain piano sucks, just think about it.

knickerhawk said:

This discussion was NEVER about whether the Sigmas allow photographers to take pleasing pictures. (I'll take your word Sigma owner for that.) It was ALWAYS about whether the Sigmas are good tools for art reproduction (and hence color accuracy).

Click to expand...

I tried it on that dollar bill (similar to the one you were posting for comparison). It is not just close, it is better. Same colors, and at similar framing. Why would you think it is lesser of a tool?

knickerhawk said:

So, your "loaded question" is completely irrelevant. Hence, Ander's perplexity.

Click to expand...

Would you say his perplexity was genuine?