Sony a7 III discussion

[Robert Collins]

9 minutes ago, Yurolov said:

Regardless, you don't understand color science if you don't see that the canon image is far superior.

Well lets start by getting this one out of the way. 'Color science' is an oxymoron. Because science deals with facts and color is a perception. Exactly the same rose will 'look' a different red to anyone against a blue background than a green background. 

And I just feel there is a whole less 'science' and a whole lot more 'subjectivity'than people imagine.

And let me explain something very basic.

I havent been doing video very long (I am a photographer) - about 6 months. But I have been studying hard and trying to learn. I have learnt that pretty much everyone goes for or likes a 'filmic' look. But however you analyse the 'filmic look' (so somewhere near to science) it is 'achieved' by actually 'degrading' in an 'objective' sense the underlying image. Nearly all the characteristics of the filmic look is about making the underlying image crappier. That goes for 'low contrast', 'less sharpness', 'adding grain', 'less accurate colors', even the '180 degree rule'.

The very phrase 'filmic' smacks not of 'science' but 'subjectivity'. When people talk of film cameras, ancient lenses etc and the results they produce they are 'definitely' talking aesthetics and not science. In an artistic sense imperfections are very important - the Mona Lisa wouldnt be the Mona Lisa without the crooked smile. And even the smartphone image we love to hate is largely a case of it being too good rather than too bad.

[sam]

30 minutes ago, Robert Collins said:

Nearly all the characteristics of the filmic look is about making the underlying image crappier.

Anyone else find humor in the irony?

Crappier defined by what metric? 

Some love god. Some love the devil. What's better soft or sharp. What's......get it yet?

[Yurolov]

16 minutes ago, Robert Collins said:

Well lets start by getting this one out of the way. 'Color science' is an oxymoron. Because science deals with facts and color is a perception. Exactly the same rose will 'look' a different red to anyone against a blue background than a green background. 

Look at what you said in your last sentence. That is the science of color. That is what I am talking about. The interplay of colors and their relation to each other. 

Effective use of this science yields results that have the effects that I mentioned previously: 1) subject isolation 2) dimensionality.

Like I said canon also renders all hues similar to skin tones in a similar fashion (e.g a persons eye bags are not so pronounced on a canon because the tones are recognized in camera and brought to uniformity with the rest of the skin) so that skin appears smooth and uniform (not blotchy and patchy like sony) which further reinforces point 1 and 2 above.

This last point is the one that most people don't seem to understand. You can see the blue eye bags with the sony image but that is corrected in the canon. It is a darker, more bluish skintone that is identified and corrected in the canon.    

[Robert Collins]

1 hour ago, sam said:

Anyone else find humor in the irony?

Crappier defined by what metric? 

Some love god. Some love the devil. What's better soft or sharp. What's......get it yet?

Yes, I agree it is a 'bit funny' but I  was trying to make a serious point.

You literally make a video 'better' by one metric by making it 'worse' by another. This is why it is easy to find people saying on this site that think Sigma Art lenses are 'brilliant' because they are super sharp and have little distortion and someone says they like say an FD lens from the 1960s because of the 'character' and 'rendering'.

When Yurolov says above that the Canon skin 'appears smooth and uniform' all he actually means to someone else 'that it lacks detail' or is 'out of focus'. And BTW I have heard people argue that Canon has ' more subtle, detailed skin tones'. So where are we here?

And when Yurolov says 'Effective use of this science yields results that have the effects that I mentioned previously: 1) subject isolation 2) dimensionality.' I would like to know exactly 'what science' he is referring to. Because I believe there has been a fundamental difference that goes back at least 300 years between say Newton and Goethe about the difference between the science and the perception of color. 

'Color science' looks like Gatsby's green light to me.

(BTW bonus point for getting Newton, Goethe and Fitzgerald in one post.)

[sam]

2 hours ago, Robert Collins said:

But I have been studying hard and trying to learn

Do you and Kye share an account? You both have admittedly little experience, yet look at your posts.  Are you the student or the teacher?

[Robert Collins]

14 minutes ago, sam said:

Do you and Kye share an account? You both have admittedly little experience, yet look at your posts.  Are you the student or the teacher?

Well arguably I am both. 

I happen to teach (or give workshops) on composition theory and Photoshop techniques because I am basically a photographer.

I am 'learning' video (or trying to). There is obviously overlap between the two (but less than I imagined).

[sam]

Great. Any posts here from you on either of the subjects of your expertise?  

 

[Robert Collins]

10 minutes ago, sam said:

Great. Any posts here from you on either of the subjects of your expertise?  

Certainly no posts here.

This is a video site. I dont think anyone here is interested in the advantages of using luminosity masks in Photoshop or why you should use curves instead of levels. And in terms of composition - the geometry in dynamic symmetry and why it matters.(And I dont think these subjects adapt well to the internet in any case.)

 

[TheRenaissanceMan]

https://indiefilmhustle.com/color-science-let-pixar-teach/

http://www.dvinfo.net/article/optical-science/a-short-history-of-camera-color.html

Smarter, more experienced people in this industry seem to disagree with your dismissal of color science as a concept. Perhaps, as a beginner, you should spend more time listening and learning, rather than espousing ideas you don't understand?

All photosites only capture luminance data. That data is assigned an R, G, or B value, depending on the filter that covers it. Those filters have to be carefully designed to let in their color and a bit of the others: if made too pure, the sensor will not be able to accurately reproduce secondary colors, like yellow and cyan; not pure enough, and colors blend together into mud from lack of separation.

These color values must be interpolated from nearby photosites, as each one only capture data for R, G, or B. How that data is cobbled together has a profound result on the resulting detail and color. Then this information feeds into a camera's color matrix, which carefully subtracts channels from each other to create clear and distinct colors. All of which are made to hit specific mathematical targets within the manufacturer's carefully designed color space. These are tuned partially for accuracy, but also to create certain effects on the viewer based on the psychological effect of various colors. 

So sure, subjectivity plays a part. That's why every manufacturer does their color slightly differently, and has their own "look." But to purport that no math or science goes into the process is misinformed and asinine. 

[Robert Collins]

25 minutes ago, TheRenaissanceMan said:

Smarter, more experienced people in this industry seem to disagree with your dismissal of color science as a concept. Perhaps, as a beginner, you should spend more time listening and learning, rather than espousing ideas you don't understand?

You have misunderstood me here. I stated I am 'new to video' and that I have largely ignored 'out of camera' colors because I shot raw. I didnt state I have no knowledge of color theory. In fact I have a lot. In fact I can clearly state that 'anyone' who believes that 'color' is inherently a science simply doesnt know what they are talking about. Goethe's work is 300 years old and absolutely nobody has disproved it. People keep on throwing around phrases like 'science' without understanding what science is.

All photosites only capture luminance data. That data is assigned an R, G, or B value, depending on the filter that covers it. Those filters have to be carefully designed to let in their color and a bit of the others: if made too pure, the sensor will not be able to accurately reproduce secondary colors, like yellow and cyan; not pure enough, and colors blend together into mud from lack of separation.

These color values must be interpolated from nearby photosites, as each one only capture data for R, G, or B. How that data is cobbled together has a profound result on the resulting detail and color. Then this information feeds into a camera's color matrix, which carefully subtracts channels from each other to create clear and distinct colors. All of which are made to hit specific mathematical targets within the manufacturer's carefully designed color space. These are tuned partially for accuracy, but also to create certain effects on the viewer based on the psychological effect of various colors. 

So sure, subjectivity plays a part. That's why every manufacturer does their color slightly differently, and has their own "look." But to purport that no math or science goes into the process is misinformed and asinine. 

Everyone knows exactly how this works. Sensors collect data in an RGB array. But inherently you are looking at non-linear equations here on the basis that a red pixel will be 'perceived' differently by a person depending on the next pixel and the next pixel after that etc... Mathematically you are simply trying to solve unsolvable equations which is not science but art. 

 

[Aussie Ash]

2 hours ago, Robert Collins said:

Yes, I agree it is a 'bit funny' but I  was trying to make a serious point.

You literally make a video 'better' by one metric by making it 'worse' by another. This is why it is easy to find people saying on this site that think Sigma Art lenses are 'brilliant' because they are super sharp and have little distortion and someone says they like say an FD lens from the 1960s because of the 'character' and 'rendering'.

 

Canon FD was introduced in March 1971 so FD lenses from the 1960s would be collectors items !!!

[sam]

@Robert CollinsColor science is an industry standard term. Are you trying to dispute that? Or  others interpretation of the term? If it's the latter, thats understandable. 

 However, if its the former, I don't feel an internet forum lends itself to such a vast topic. (Like your areas of expertise)

 If you truly want to learn about video, why not a reputable book/publication. Anything by Charles Poynton seems like it might be a good fit.  (For the mathematical and technical side of video at least)

 

 

[hansel]

6 hours ago, Robert Collins said:

Well lets start by getting this one out of the way. 'Color science' is an oxymoron. Because science deals with facts and color is a perception. Exactly the same rose will 'look' a different red to anyone against a blue background than a green background. 

And I just feel there is a whole less 'science' and a whole lot more 'subjectivity'than people imagine.

And let me explain something very basic.

I havent been doing video very long (I am a photographer) - about 6 months. But I have been studying hard and trying to learn. I have learnt that pretty much everyone goes for or likes a 'filmic' look. But however you analyse the 'filmic look' (so somewhere near to science) it is 'achieved' by actually 'degrading' in an 'objective' sense the underlying image. Nearly all the characteristics of the filmic look is about making the underlying image crappier. That goes for 'low contrast', 'less sharpness', 'adding grain', 'less accurate colors', even the '180 degree rule'.

The very phrase 'filmic' smacks not of 'science' but 'subjectivity'. When people talk of film cameras, ancient lenses etc and the results they produce they are 'definitely' talking aesthetics and not science. In an artistic sense imperfections are very important - the Mona Lisa wouldnt be the Mona Lisa without the crooked smile. And even the smartphone image we love to hate is largely a case of it being too good rather than too bad.

I think you are getting two different things mixed up here. "Colour science" data acquisition and translation of the sensor pipeline on to the memory.  The other thing is "filmic look". As a photographer you should know, there is no wrong or right, there are trends and is changing and evolving depending how you look at it. What film? What format? Lenses? Push, pull, resulting grain......put in endless factors here......

Canon compared to Sony simply has a fairly well controlled translation of what the sensor is capturing in regards to the human eye.

1 hour ago, sam said:

@Robert CollinsColor science is an industry standard term. Are you trying to dispute that? Or  others interpretation of the term? If it's the latter, thats understandable. 

 However, if its the former, I don't feel an internet forum lends itself to such a vast topic. (Like your areas of expertise)

 If you truly want to learn about video, why not a reputable book? Anything by Charles Poynton seems like it might be a good fit.  

 

 

That.

[Django]

I know I'm repeating myself but in the Canon vs Sony video, once again all the tests are done in daylight conditions (as indicated by their WB settings). Sony works fine in daylight imo. It's in more complicated lighting situations (mixed/artificial) that Sony cams start to struggle.

The best video showcasing & explaining the dreaded Sony WB/color issues imo is Andrew's own EOSHD Pro Color samples vid:

 

[anonim]

7 hours ago, Yurolov said:

Like I said canon also renders all hues similar to skin tones in a similar fashion (e.g a persons eye bags are not so pronounced on a canon because the tones are recognized in camera and brought to uniformity with the rest of the skin) so that skin appears smooth and uniform (not blotchy and patchy like sony) which further reinforces point 1 and 2 above.

This last point is the one that most people don't seem to understand. You can see the blue eye bags with the sony image but that is corrected in the canon. It is a darker, more bluish skintone that is identified and corrected in the canon.    

Out of discussion, I'm always so glad when anybody actually point out to what (concretely) we should look at discussing our opinions... I think it is the best way to test ourselves, progress in creatively utilized perception and learn about nuances and differences... So, thumbs up for noting eye bags!

[hmcindie]

5 hours ago, Robert Collins said:

You literally make a video 'better' by one metric by making it 'worse' by another. 

If you think about it a bit, that's how films are shot. You take a scene, then block and eliminate things that you don't want the audience to see. Is the table in the background necessary, do I want to focus on a certain characters face on a certain moment? By taking AWAY that table and focusing on a character instead, you might be making the film better. But the table is gone! Now it's worse if the metric is "where is the table?".

It takes taste & skill to know when that table should be in the shot and when not. That's what filmmakers do, point the focus of the story on certain elements and away from others. If that includes softening the image from a certain part of the shot (be it using dof or masking it out in grading) then that's what the film needs. It's not about "making it worse". It is about "making it better". What is better is subjective but film grain is usually a pretty good addition if done well, even though on a certain metric it might be bad. But good taste is what a director is there for.

[Robert Collins]

3 hours ago, sam said:

@Robert CollinsColor science is an industry standard term. Are you trying to dispute that? Or  others interpretation of the term? If it's the latter, thats understandable. 

 However, if its the former, I don't feel an internet forum lends itself to such a vast topic. (Like your areas of expertise)

 If you truly want to learn about video, why not a reputable book/publication. Anything by Charles Poynton seems like it might be a good fit.  (For the mathematical and technical side of video at least

 

I am certainly not talking semantics here and I know ‘color science’ is a standard industry term. But it’s a bit like another industry term - ethical banking - I just don’t see much evidence that exists.

And let’s look at this logically - when Yurolov says ‘that you don’t understand color science if you can’t recognize Canon colors are superior’ - it implies he understands color science and I do not - which is obviously pretty reasonable. But, as a corollary it also implies that he understands ‘color science’ but Sony does not. Now essentially that has to be heroically optimistic. I mean Sony seems to have a good grasp of BSI tech, stacked sensors, on sensor adc, that back of the sensor ram stuff - gee they can even build a sensor for security companies that they can sell to video camera manufacturers as the greatest thing since sliced bread - but still apparently they can’t get a grip of color science.

 

[Kisaha]

I have just accepted the fact that @Robert Collins is a master of everything, and my life has been better since.

6 hours ago, sam said:

Do you and Kye share an account? You both have admittedly little experience, yet look at your posts.  Are you the student or the teacher?

At least @kye accepts the opinion of others, and trying to learn something, he is too optimistic about things sometimes (and physics, and science as a whole), but I believe he is learning the basics, while Robert is trying to teach us everything on a short course.

5 minutes ago, Robert Collins said:

I am certainly not talking semantics here and I know ‘color science’ is a standard industry term. But it’s a bit like another industry term - ethical banking - I just don’t see much evidence that exists.

And let’s look at this logically - when Yurolov says ‘that you don’t understand color science if you can’t recognize Canon colors are superior’ - it implies he understands color science and I do not - which is obviously pretty reasonable. But, as a corollary it also implies that he understands ‘color science’ but Sony does not. Now essentially that has to be heroically optimistic. I mean Sony seems to have a good grasp of BSI tech, stacked sensors, on sensor adc, that back of the sensor ram stuff - gee they can even build a sensor for security companies that they can sell to video camera manufacturers as the greatest thing since sliced bread - but still apparently they can’t get a grip of color science.

 

Yes.

[Yurolov]

19 minutes ago, Robert Collins said:

I am certainly not talking semantics here and I know ‘color science’ is a standard industry term. But it’s a bit like another industry term - ethical banking - I just don’t see much evidence that exists.

And let’s look at this logically - when Yurolov says ‘that you don’t understand color science if you can’t recognize Canon colors are superior’ - it implies he understands color science and I do not - which is obviously pretty reasonable. But, as a corollary it also implies that he understands ‘color science’ but Sony does not. Now essentially that has to be heroically optimistic. I mean Sony seems to have a good grasp of BSI tech, stacked sensors, on sensor adc, that back of the sensor ram stuff - gee they can even build a sensor for security companies that they can sell to video camera manufacturers as the greatest thing since sliced bread - but still apparently they can’t get a grip of color science.

 

Sony can but it hasn't been utilised in the aloha series which is why I was excited about the fact that the purported as7iii will be getting Venice color science. 

People in video do not use the term color science in the same way as someone like Goethe did (I imagine). Simply put, it means how a camera reproduces color and also why it does it this way (which is color theory e.g. skin tone preference). There may be other definitions too. But we all know what we mean.  

Your arguments are semantic. Mine are substantive. 

I have explained to you why there are differences and what these mean to the final image. None of which you have responded to. 

1. Subject isolation.

2. Dimensionality.

3. Skin tones and uniformity (see above).

Now you can tell me why you think the Sony is better. 

[Kisaha]

5 minutes ago, Yurolov said:

Sony can but it hasn't been utilised in the aloha series which is why I was excited about the fact that the purported as7iii will be getting Venice color science. 

People in video do not use the term color science in the same way as someone like Goethe did (I imagine). Simply put, it means how a camera reproduces color and also why it does it this way (which is color theory e.g. skin tone preference). There may be other definitions too. But we all know what we mean.  

Your arguments are semantic. Mine are substantive. 

I have explained to you why there are differences and what these mean to the final image. None of which you have responded to. 

1. Subject isolation.

2. Dimensionality.

3. Skin tones and uniformity (see above).

Now you can tell me why you think the Sony is better. 

A big part of Canon, Nikon and Fuji sales are based on their "color science", I am not even sure how this is not a thing anymore!

..and to continue with the Sony argument, they made a whole mkII camera in their "cinema" series, and the "only" difference is the improvement of their color science, so it must be a thing for Sony as well...

and to take that further, GH5S is a lot improved in their color science department, and we discussed that in the appropriate thread. I am not sure how it is not an issue for Collins, Robert but every camera manufacturer make it such a big issue.

Fuji? Eterna and the such.

Black Magic? Just released their color improvement 4.0, if I am not mistaken, they did it a big deal too, so it must be a thing..