Sony Colors

[iiiNelson]

Generally matches my opinion that they produce the most accurate colors I’ve seen straight out of the camera but that doesn’t make them the most aesthetically pleasing for everyone necessarily. Well I guess someone actually did the testing now to essentially co-sign what I’ve long said.

https://www.pdnonline.com/gear/cameras/the-best-cameras-for-color-reproduction-ranked/

 

[Vivek]

I hope that opinion is recent because they do keep changing/tweaking the Bayer dyes on their sensors. It isn’t constant but always evolving as it should be.

 

[RodolfoCC]

A somewhat opposite opinion published a few days ago in Petapixel...

Is Sony’s Color Science Really That Bad?

 

[iiiNelson]

A somewhat opposite opinion published a few days ago in Petapixel...

Is Sony’s Color Science Really That Bad?

I find it ironic that I found most of the Sony images to be more neutral and “accurate” (or realistic for probably a better term) even in the presented comparisons. The Canon looks like the red channel is boosted in nearly every image with a person in there. Some of it may be subjective to what a person likes and also subjective to the skin tone of the person within images. Also almost every picture looks grossly underexposed.

Our eyes/brains tend to do a funny thing and compensate for information that may not actually be seen. There’s plenty of research on this phenomena.

Picture taking ability of the photographer aside, some people actually have yellow/olive undertones (versus red ones) and I find that Sony brings them out (maybe more than some people want) but I agree that there are many variables in comparing cameras from lenses, lens coatings, RAW converter choices and their individual profiles, and whether a person chooses to custom white balance. For portraits - I ALWAYS custom white balance in camera (or by taking a picture with the subject holding a color checker to white balance from) and go from there. So maybe the way I shoot yields results I’m satisfied with more than relying on the camera to decide in AWB.

 

[iiiNelson]

I hope that opinion is recent because they do keep changing/tweaking the Bayer dyes on their sensors. It isn’t constant but always evolving as it should be.

Well I’ve had the opinion that Sony colors were aesthetically pleasing since the A900 days (before I owned anything Sony that wasn’t a TV, PlayStation, or Walkman) when people liked the color to the A77 days as well... but I think somewhere in between the NEX-6 and the A7 Sony went for more accurate color above aesthetically pleasing. I don’t think this was a mistake but I do believe accuracy led to FUD that the color science is bad... I think it’s accurate but it’s like the old FUD of TV adding 10 lbs when in reality people can finally see what they actually look like.

 

[RodolfoCC]

As a colour-blind person myself, accuracy is important to me: at least it gives me a reference point. My main worry is getting images with 'weird' casts, and real colours rarely give them. Nevertheless, the importance of lenses on colour rendering of final images has always been very visible even for me and my handicapped vision. I don't think it can be taken out of the equation.

 

[iiiNelson]

As a colour-blind person myself, accuracy is important to me: at least it gives me a reference point. My main worry is getting images with 'weird' casts, and real colours rarely give them. Nevertheless, the importance of lenses on colour rendering of final images has always been very visible even for me and my handicapped vision. I don't think it can be taken out of the equation.

I largely agree Lens coatings have an effect on the base image but that’s also why I think white balancing is extremely important even when shooting RAW. All too often I hear photographers say it doesn’t matter but when shooting RAW but I’m still of the opinion to get as much right (to desired look) in camera when possible.

 

[Vivek]

It was rumored that gen III had a change of Bayer dyes (“color chemistry”).

Also, with the BSI sensors there more major differences.

 

[iiiNelson]

It was rumored that gen III had a change of Bayer dyes (“color chemistry”).

Also, with the BSI sensors there more major differences.

Possibly but I think there are two distinct competing ideas here. Color reproduction accuracy versus color science which tends to be more rooted in producing a desired look in the end product. One of the “good standards” of color science, Fuji, tends to have boosted greens and blue/cyan bleeding through both the green and blue channels to give more of that filmic cyan look in the skies... they also boost orange into the tones typically associated within skin tones to give that finished filmic look as well. It’s not accurate but it’s definitely pleasing for a wide range of skin tones. Sony files tend to be more accurate and neutral but that requires more work (or custom profiling) to get to a more stylistic look IMO which is where the whole “bad color science” ideas come from. I haven’t seen a great deal of difference in the color reproduction from generation 3 FE bodies but I do know they have wider dynamic ranges - which ironically tend to give a worse perceived “color science” generally speaking but a greater accuracy of color reproduction if you will.

 

[Vivek]

I was saying that there is no constant within Sony products. You can see a clear difference between the RX1and the RX1R II.

 

[iiiNelson]

I was saying that there is no constant within Sony products. You can see a clear difference between the RX1and the RX1R II.

I don’t own either so I can’t say. Apparently there’s differences in all cameras judging from thebreproduction results.

 

[seb]

The comparison is not very sophisticated. It's like looking after the hairstyle of 100-m runners and saying who will win the race. Although the shape may have an influence on the time none of the runners had cut their hair because of that.

Two things:

OOC-JPGs:
Each camera set a certain style into the JPGs. The reason behind is quite simple. If the jpg would be colour-accurate no one would ever use it and it would be only usable with perfect light conditions. Therefore they have to process it in a certain way that buyers see as pleasing. Currently, it's a contrasty and high vibrancy look with a small green hue range. Brights getting pushed but also compressed on the top. This may change over time. For example the OOC-colour complainings about the II-series (greenish tone) went into the a9/a7III/a7RIII.

Accuracy:
There is a generalised understanding of a dE value greater than 1 the difference is getting visible. Below it's not. If we look at the values on these comparisons it's obvious none of the brands had accuracy in mind when they programmed the algorithms for the OOC-JPGs.

The first question here is what would be possible?
I spent hours and money in the last years to get an answer to that question: Let's assume my setup for profiling my camera (a7RII) was perfect then my best approach to get an "accurate" profile was with an average dE of 0.8 and a max of 3.6. This with a combination of two targets with very different (and many) colour patches. Here is a "similar to the article"-visualisation of one of the used targets (the point in the center is the reference, the rest of the patch shows how my profile render it). It's the accurate profile with a slight push of the darks. I'm too lazy to look after the fully correct one atm.



How practicable is an "accurate" profile?
Well, I never use it. The reason is behind the RAW data. Light (colour, dynamic range, haze, ...), hardware (esp. sensor), software (in camera but also raw-processor) are in a play all together. There are some limitations due to the used hard- and software but the biggest issue here is that the result looks not how my imagination sees it when I press the shutter. And creating my look with the adjustments of the raw-processor is just a lot of work. Well at least with an "accurate" profile. It's the reason why I started to tweak the profile in a certain way. And yes in general, my changes goes to more contrast and pushed brights that are compressed on top. Especially the compressing of the brights is a very important step because differences of dynamic range and saturated colors needs very different adjustments. It's much easier if you just compress them. It's something you can see in the profiles of LR and C1 as well (unless you use a "linear scientific" curve with a "accurate" profile in C1).
Don't ask me about the dE of my tweaked profiles. I don't care because the "accurate" profile may be a good start for creating a profile but the adjustments give the profile it's useability and a certain style. Like it is with the OOC-JPGs...

 

[seb]

The article may be about colours but has nothing to do with science. And instead of blaming Sony they should point at Adobe and X-Rite and their profiling/processing algorithms.
Therefore the title is very very misleading. And the article itself shows clearly that the writer may be a good journalist but not much into colour science (but indirectly he claims to be).

If you are interested in communication science: It's a good example, why framing helps to understand topics and narration helps to get the attention of the recipients but the mediated knowledge is counterproductive.

A somewhat opposite opinion published a few days ago in Petapixel...

Is Sony’s Color Science Really That Bad?

 

[iiiNelson]

The comparison is not very sophisticated. It's like looking after the hairstyle of 100-m runners and saying who will win the race. Although the shape may have an influence on the time none of the runners had cut their hair because of that.

Two things:

OOC-JPGs:
Each camera set a certain style into the JPGs. The reason behind is quite simple. If the jpg would be colour-accurate no one would ever use it and it would be only usable with perfect light conditions. Therefore they have to process it in a certain way that buyers see as pleasing. Currently, it's a contrasty and high vibrancy look with a small green hue range. Brights getting pushed but also compressed on the top. This may change over time. For example the OOC-colour complainings about the II-series (greenish tone) went into the a9/a7III/a7RIII.

Accuracy:
There is a generalised understanding of a dE value greater than 1 the difference is getting visible. Below it's not. If we look at the values on these comparisons it's obvious none of the brands had accuracy in mind when they programmed the algorithms for the OOC-JPGs.

The first question here is what would be possible?
I spent hours and money in the last years to get an answer to that question: Let's assume my setup for profiling my camera (a7RII) was perfect then my best approach to get an "accurate" profile was with an average dE of 0.8 and a max of 3.6. This with a combination of two targets with very different (and many) colour patches. Here is a "similar to the article"-visualisation of one of the used targets (the point in the center is the reference, the rest of the patch shows how my profile render it). It's the accurate profile with a slight push of the darks. I'm too lazy to look after the fully correct one atm.

View attachment 136741

How practicable is an "accurate" profile?
Well, I never use it. The reason is behind the RAW data. Light (colour, dynamic range, haze, ...), hardware (esp. sensor), software (in camera but also raw-processor) are in a play all together. There are some limitations due to the used hard- and software but the biggest issue here is that the result looks not how my imagination sees it when I press the shutter. And creating my look with the adjustments of the raw-processor is just a lot of work. Well at least with an "accurate" profile. It's the reason why I started to tweak the profile in a certain way. And yes in general, my changes goes to more contrast and pushed brights that are compressed on top. Especially the compressing of the brights is a very important step because differences of dynamic range and saturated colors needs very different adjustments. It's much easier if you just compress them. It's something you can see in the profiles of LR and C1 as well (unless you use a "linear scientific" curve with a "accurate" profile in C1).
Don't ask me about the dE of my tweaked profiles. I don't care because the "accurate" profile may be a good start for creating a profile but the adjustments give the profile it's useability and a certain style. Like it is with the OOC-JPGs...

I believe we are actually saying the same things. Accuracy doesn’t equal aesthetically pleasing but rather just accuracy. Yes it’s a factor and it’ll make more of a difference with those that are concerned with a baseline color accuracy before applying their own look and profiles.

 

[seb]

I believe we are actually saying the same things. Accuracy doesn’t equal aesthetically pleasing but rather just accuracy. Yes it’s a factor and it’ll make more of a difference with those that are concerned with a baseline color accuracy before applying their own look and profiles.

Exactly. It was your post that inspired me to add my findings to your conclusion.

 

[iiiNelson]

Exactly. It was your post that inspired me to add my findings to your conclusion.

I have a GIS Analysis background that I did for the better part of my adult life until moving more into the management side of things. As such we had to keep color accurate workstations that we used for certain types of multispectral and htperspectral analysis where color accuracy was paramount. As such I’ve always been super sensitive to color accuracy for photo work but you’d be surprise how many people don’t calibrate their monitors regularly and/or white balance their cameras. In my experience this solves a lot of issues before they’re issues.

 

[seb]

My background and daily business is communication science. A few years ago I was responsible for a colour managed workflow of a company for a while but colour science or colour accuracy is a completely different thing. It's at least my learning since I'm more into it with photography. I'm still in the very basics and I would love to have a background like you in that case.

I have a GIS Analysis background that I did for the better part of my adult life until moving more into the management side of things. As such we had to keep color accurate workstations that we used for certain types of multispectral and hyperspectral analysis where color accuracy was paramount. As such I’ve always been super sensitive to color accuracy for photo work but you’d be surprise how many people don’t calibrate their monitors regularly and/or white balance their cameras. In my experience this solves a lot of issues before they’re issues.

 

[iiiNelson]

My background and daily business is communication science. A few years ago I was responsible for a colour managed workflow of a company for a while but colour science or colour accuracy is a completely different thing. It's at least my learning since I'm more into it with photography. I'm still in the very basics and I would love to have a background like you in that case.

I agree and also learning how it’s best to shoot a camera is another factor. I know a lot of Canon people live and die by ETTR where I found shooting with the M9 (and subsequently the Sony’s) that underexposing a bit to protect highlights and recovering shadows works best within reasons (for instance not having to recover more that 2 or 3 stops though stress testing says recovering 4-5 can be doable at times). Outside that I tend to use OCF for subjects if I need to underexpose to illuminate a subject for say a portrait.

Cool. I’m back in school now working to get my MBA and to move deeper into the executive management realm from the mid-management space. I’ll always love the GIS world (especially when helping with crisis management relief efforts like I did with the Haiti Earthquake in 2010, the Fukushima Disaster in 2011, and various other contingencies around the world) but my newest imminent/long term goal is to get more into the executive marketing and strategic vision spaces.

Photography is primarily a hobby that is providing a little extra side income to pay for my gear purchases and a few bills. At times I consider just going all innon it but money tends to complicate passions IMO. I’m trying to keep it as “pure” as possible while allowing the occasional side job to feed the passion to simultaneously pay for itself.

 

[archivue]

No leaf, no phase one, no Fuji GFX...

Generally matches my opinion that they produce the most accurate colors I’ve seen straight out of the camera but that doesn’t make them the most aesthetically pleasing for everyone necessarily. Well I guess someone actually did the testing now to essentially co-sign what I’ve long said.

https://www.pdnonline.com/gear/cameras/the-best-cameras-for-color-reproduction-ranked/

 

[iiiNelson]

No leaf, no phase one, no Fuji GFX...

Well to answer the question I was speaking more in reference to small format systems.

I’d say this. I love the Fuji colors, especially those from the GFX, but I wouldn’t call them super accurate. They’re stylistic and aesthetically pleasing. I love the classic Leaf color and hands down it produces the best out of camera skin tones (no matter the skin tone of a person)I’ve seen from any camera that I’ve demoed but I believe they call it a “Leaf look” for a reason. To me Leaf Cameras were the medium format equivalent of “the look” we got from the Leica M9 (which I still think produces the best portraits from any small format camera I ever used). Maybe that’s a matter of the look being derived from trying to recreate the look of Kodak Color Science in an attempt to give a “Kodachrome” like look. Phase One colors are easily the most accurate from all the cameras you just mentioned and probably the most neutral of the three you mentioned but I think they have the same “issue” (if you will) as Sony when compared to other medium format systems. They’re extremely accurate colors but they’re not the most aesthetically pleasing before tweaking or pushing through a Capture One to add the secret sauce - which is why I believe Phase One spends so much time to profile the cameras through Capture One. Capture One adds a Leaf profile still for a reason IMO. Capture One and Phase Backs play hand in hand (and I’m not saying this is a bad thing at all because Capture One does it’s magic with Sony files as well compared to Adobe) and people should think of a camera/RAW converter as a system.

...and I’d add that I find Hasselblad to be somewhere between Leaf and Phase One which is to say it may have the best balance but I don’t like the odd shaped bokeh of the 5 bladed leaf shutters... it’s a little unsettling for me personally but the colors are very pleasing without a doubt.