Keep close to authentic look of object or create stunning, award winning and over processed photo?

[vlaiv]

1 minute ago, AstroRookie said:

Hello Vlaiv,

what you're explaining here, isn't that what Siril does with it's "Photometric calibration" functionality?

AstroRookie

 

I'm not familiar with Siril, but I have had discussion about PixInsight Photometric color calibration.

One must be careful to understand what tool does. Sometimes when photometric color calibration term is used - what it implies is "color" in photometric sense, or quantity that is called color index - or B-V color index (there are other color indices depending on photometric system used):

https://en.wikipedia.org/wiki/Color_index

Same principle is used for regular color calibration - photometric measurement of star in different channels - but this time color is represented in sRGB color space, since most devices and whole internet expects sRGB color space.

Two are different (photometric color index and sRGB color triplet) but principle of calibration is similar.

Just make sure that tool that you are using does what you expect it to do and not something else. If you want to display your image on computer screen or publish on internet - use sRGB color space (for print use CMYK color space or in case of digital printer - use supplied color profile. In that case it is worth using extended color space like CIE RGB or Adobe RGB in processing - but then you need to calibrate your RGB channels for that color space prior to processing in Photoshop or Gimp or similar).

(for reference, sRGB colors vs star class is given in following chart):

as you see - these are different than above B-V colors and are proper visual colors (G0V star never has greenish appearance visually like in above chart)

[ollypenrice]

Keep close to authentic look of object or create stunning, award winning and over processed photo?

I'd still like to come back to the top of the thread, above. Could we have some links to images which are 'stunning, award winning and over processes photos' and to others which 'keep close to authentic look of object.' It's very easy to toss out these terms and present them as alternatives. It may not be quite so easy to justify them. Without specific examples of each we are (are we not?) just waffling?

Olly

[Davey-T]

2 minutes ago, ollypenrice said:

Keep close to authentic look of object or create stunning, award winning and over processed photo?

I'd still like to come back to the top of the thread, above. Could we have some links to images which are 'stunning, award winning and over processes photos' and to others which 'keep close to authentic look of object.' It's very easy to toss out these terms and present them as alternatives. It may not be quite so easy to justify them. Without specific examples of each we are (are we not?) just waffling?

Olly

Don't know about award winning but most over processed gaudy coloured images seem to originate from our US cousins especially those blessed with dark Arizona skies, no names no pack drill 🥳

Dave

[raf2020]

1 hour ago, ollypenrice said:

Keep close to authentic look of object or create stunning, award winning and over processed photo?

I'd still like to come back to the top of the thread, above. Could we have some links to images which are 'stunning, award winning and over processes photos' and to others which 'keep close to authentic look of object.' It's very easy to toss out these terms and present them as alternatives. It may not be quite so easy to justify them. Without specific examples of each we are (are we not?) just waffling?

Olly

Hey Olly. My intention wasn't to point with finger on particular photos, especially photographers. I have highlighted in my first post what was my concerns on examples of Rosette or Orion nebulas.

Cheers!

[ollypenrice]

7 hours ago, raf2020 said:

Hey Olly. My intention wasn't to point with finger on particular photos, especially photographers. I have highlighted in my first post what was my concerns on examples of Rosette or Orion nebulas.

Cheers!

That's fair enough. Actually I sometimes hanker for the old days when we didn't go so deep. The great thing is, you can still process your data as you like. The first time I saw the Elephant Trunk Nebula in an image, the only only sign of nebulosity was in the glowing rim of the Trunk. I thought this was stunningly beautiful and eerie. As we all started to go deeper and deeper in H alpha the whole area around the Trunk turned into a rather flat and featureless sea of red. The last time we did it here (this one was with Yves Van den Broek) I decided not to exploit the full strength of the Ha signal available but to emphasize its contrasts and hold it down so it really picked out the rim of the Trunk as in that first picture I saw.

It's also quite common for imagers to post multiple versions. I always do this if I'm posting an Ha version of a target not usually seen with the hydrogen structures visible. (The Double Cluster or the wide field around the Cocoon Nebula, for instance.) I photograph these Ha backgrounds simply because they are there. I don't think they make a better picture,  just a different one. And I do like a faint Rosette seen as a diaphanous and discrete circular object. Deep images connect it in Ha all the way up to the Cone Nebula. We can have both.

Olly

 

[andrew s]

10 hours ago, Davey-T said:

Don't know about award winning but most over processed gaudy coloured images seem to originate from our US cousins especially those blessed with dark Arizona skies, no names no pack drill 🥳

Dave

Have you seen some of the entries in the FLO IKO images processing competition ?

I know it has been narrow band so far but my eyeballs are still recovering.  🤩🤩

On a more serious note, as a non imager, it's about the craft of rendering what the processor (artist) wants to draw from the data and its relationship to the intended impact on the viewer.

Regards Andrew 

 

[saac]

11 hours ago, andrew s said:

On a more serious note, as a non imager, it's about the craft of rendering what the processor (artist) wants to draw from the data and its relationship to the intended impact on the viewer.

Regards Andrew 

 

Agreed and in this respect there is no "authentic" because it is very much an artistic judgement.

Jim  

[vlaiv]

1 hour ago, saac said:

Agreed and in this respect there is no "authentic" because it is very much an artistic judgement.

Jim  

Would you characterize following scenario as "authentic" or "artistic":

You take DSLR camera, set it to faithful color reproduction and take an image of a scene and transfer it to computer without any further processing.

 

[saac]

42 minutes ago, vlaiv said:

Would you characterize following scenario as "authentic" or "artistic":

You take DSLR camera, set it to faithful color reproduction and take an image of a scene and transfer it to computer without any further processing.

 

The "faithful" colour reproduction setting of the camera will be an algorithm chosen/designed by the camera's engineers - it will no doubt have incorporated some image processing.  The final output conforming to what the designers define as "faithful " will therefore produce an image influenced by the internal algorithm and instrument response.  When transferred to a computer and rendered on the display, the colour reproduction will again be subject to further interpretation based on the monitor's response.  You could argue then the image has been "authentically" reproduced  - but you would qualify authentic in this case based on the response characteristics of the instrument train.   At the same time you could equally argue that the image has been presented artistically; here you have chosen how the image is to be presented and viewed. Therefore, you have made (consciously or not) an artistic value judgment of your work.  So artistic and authentic in superposition   

 

Now ask me what I think of Tracey Emin's bed  or the formaldehyde shark!

https://www.tate.org.uk/art/artworks/emin-my-bed-l03662

 

Jim 

Edited December 31, 2020 by saac

[vlaiv]

32 minutes ago, saac said:

The "faithful" colour reproduction setting of the camera will be an algorithm chosen/designed by the camera's engineers - it will no doubt have incorporated some image processing.  The final output conforming to what the designers define as "faithful " will therefore produce an image influenced by the internal algorithm and instrument response.  When transferred to a computer and rendered on the display, the colour reproduction will again be subject to further interpretation based on the monitor's response.  You could argue then the image has been "authentically" reproduced  - but you would qualify authentic in this case based on the response characteristics of the instrument train.

Actually no, or at least not in sense that algorithm used is not left to designer discretion but is rather well defined.

There is standard developed to assign certain values based on frequency response of recorded light. That is CieXYZ value of each color and corresponding spectrum. Since whole spectrum is compressed into three values this is by definition lossy transform, but it is closely modeled on human vision and human sensory response will be the same for all spectra that give the same XYZ values. Contrary to popular belief - each color does not have unique spectrum but rather infinite set of different spectra can and do produce same color in human vision.

Knowing these three values and having light sources that produce each of these three spectra:

you can produce equivalent spectrum that will induce same psychological response in human being as original spectrum. Although you don't capture exact spectrum - you do capture important information to recreate perception of original spectrum. There is no freedom in this part so far. Above matching functions are well known and modeled on human vision.

First step in process of creation of faithful image is recording some values from camera response (what ever camera QE and filter responses are) and doing linear transform (since light response is element of vector space) to above well defined CieXYZ. This is what color correction in cameras really does (color balance does something different - but in exact same way).

Once you have CieXYZ tristiumuls value you can then use that to convert to color space of your choice. In particular - computer and phone screens and internet devices all expect sRGB color space - That color space again is well defined.

Well calibrated computer screen will take sRGB tristimulus value, decode it and produce again spectrum (by mixing three light sources of certain spectra that we see as red, green and blue) that will create the same psychological response in human observer.

In another words - if you have red torch that has particular red cast that is within gamut of both camera and sRGB monitor and you take that camera and shoot the light of that torch at faithful setting and then save that as jpeg format (again using sRGB as standard) and display that on properly calibrated computer screen and ask people - is that the same color as color of torch light - you will get the answer, yes that is the same color.

It is important to understand that R, G and B are not arbitrary red, green and blue hues and they depend on actual color space used. If you have something encoded in "RGB" model but you don't have well defined hues of these three colors - you don't have color information.

In sRGB color space - we have well defined R, G and B components - we know their XYZ values - for all three of them and that is basis for our linear transform between spaces.

There is also CieRGB space - that is also RGB space, but R, G and B have different red green and blue color to sRGB color space. Same for AdobeRGB and other RGB spaces. Here is an overview of RGB like color spaces and their primaries:

If you follow above rules and do proper color management - then there is really not much artistic or left to your designer choice in creating a documentary image. Everything is well defined with purpose of producing perceptually the same color / vision response in human subject.

 

[saac]

17 minutes ago, vlaiv said:

Actually no, or at least not in sense that algorithm used is not left to designer discretion but is rather well defined.

There is standard developed to assign certain values based on frequency response of recorded light. That is CieXYZ value of each color and corresponding spectrum. Since whole spectrum is compressed into three values this is by definition lossy transform, but it is closely modeled on human vision and human sensory response will be the same for all spectra that give the same XYZ values. Contrary to popular belief - each color does not have unique spectrum but rather infinite set of different spectra can and do produce same color in human vision.

Knowing these three values and having light sources that produce each of these three spectra:

 

I think we will differ on this one vlaiv.  The algorithm has been "designed" therefore it responds to a particular prescription, the image it produces will also be subject to the particular instrument response. No two sensor and certainly not between different manufacturers will produce identical images. As for artistic classification then I consider that is the prerogative of the person presenting the image - hence my reference to the modern art efforts of Tracey Emin. I see a messy unmade bed of no artistic value but the artist considers otherwise. The value judgement of what is artistic is not easily defined if it is at all possible to define, it is afterall something that provokes a personal response in the viewer. The mere act of presenting an image is considered by some as artistic. 

Jim 

[andrew s]

I feel @vlaiv proposal is as authentic as it can get but the artist still chooses the framing and exposure etc. However, I like @saac I see astro images as acts of creation albeit from  "nature" .

Regards Andrew 

[saac]

2 hours ago, andrew s said:

I feel @vlaiv proposal is as authentic as it can get but the artist still chooses the framing and exposure etc. However, I like @saac I see astro images as acts of creation albeit from  "nature" .

Regards Andrew 

Isn't this the problem with the term artistic I guess, it is so loosely defined and without any common agreement.  I accept also that CCD/CMOS and displays can be matched to the broad spectral response of the human eye. I still have difficulty accepting that our wonderfully richly saturated astro photographs can be considered representative of "authentic" - or at least we need to define what we mean by authentic for this to hold true. Are we defining  authentic in terms of how closely the image replicates our own visual response or does authentic refer to replication of what is actually there.  I think the later is more in the realm of science/cosmology than astrophotography as it will look for a more quantitative outcome covering the whole of the EM spectrum. Looking at the former definition  I'm sure a past thread somewhere contains a discussion on how we would visually perceive a nebula were we closer or indeed inside.  Our cameras would continue to register the deep saturated colours but would our eyes, with shorter integration time,  remain unable to discern the feint color?

Jim

[saac]

2 hours ago, andrew s said:

I feel @vlaiv proposal is as authentic as it can get but the artist still chooses the framing and exposure etc. However, I like @saac I see astro images as acts of creation albeit from  "nature" .

Regards Andrew 

Actually Andrew I think you are right; was it not yourself that described the discussion as a false dichotomy.  We are being presented with too narrow an option field to properly explore.   Anyway, be our images artistically presented or records of the authentic I'm just happy that we have the means to capture and enjoy them.  I am also grateful that some clever people are really skilled at the capture and processing as my efforts could never be mistaken for either artistic nor authentic ! 

Jim 

[vlaiv]

2 minutes ago, saac said:

Isn't this the problem with the term artistic I guess, it is so loosely defined and without any common agreement.  I accept also that CCD/CMOS and displays can be matched to the broad spectral response of the human eye. I still have difficulty accepting that our wonderfully richly saturated astro photographs can be considered representative of "authentic" - or at least we need to define what we mean by authentic for this to hold true. Are we defining  authentic in terms of how closely the image replicates our own visual response or does authentic refer to replication of what is actually there.  I think the later is more in the realm of science/cosmology than astrophotography as it will look for a more quantitative outcome covering the whole of the EM spectrum. Looking at the former definition  I'm sure a past thread somewhere contains a discussion on how we would visually perceive a nebula were we closer or indeed inside.  Our cameras would continue to register the deep saturated colours but would our eyes, with shorter integration time,  remain unable to discern the feint color?

Jim

I think that there are more levels of authenticity.

Authenticity of color for example is one - which would be defined as - image produces the same color as would be seen by eyes if light was strong enough

Authenticity of shape - where spatial distribution of perceived intensity of light matches that of object with certain projection

Authenticity of ratios - where ratios of light intensity displayed matches ratios of light intensity of object

Authenticity of instrument response - when image contains differences to real life due to instrument used

Authenticity of information - where information is captured that is beyond human senses yet is displayed in such way that it conveys information of the object (like narrowband images and images created with information in other parts of EM spectrum)

Etc.

Astrophotography that we are discussing is 99% of the time in "breach" of any number of above.

Arbitrary color balance and color saturation certainly distort authenticity of color and the way we stretch images distort ratios. But this is either decision or lack of knowledge - if one chooses to do so - they can minimize distortion of both.

Most of the time authenticity of shape is preserved - except in wide field images where it is impossible to do so as it is mathematically impossible to map sphere onto a plane without distortion.

Sometimes authenticity of instrument response is also distorted - people add star spikes / remove reflections and such

Some are mutually exclusive - for example authenticity of color and instrument response - we do color correction and change data to match human vision and loose information on camera+filter QE response in the process.

By the way - two cameras will match recorded colors as long as colors are within their gamuts and therefore will create identical images consisting out of those colors.

[saac]

That is a good summation vlaiv.  So here is a different but related question.  I think we would agree that professional astronomers/cosmologist are more concerned with the integrity of the data in their images and less with the aesthetic.  I'm guessing that for the likes of their spectroscopy analysis then instrument response, spectral range, resolution etc are the important concerns. What else is important , what are the main data streams that they are looking for - position, velocity, rotation ?  How do they ensure that what they are recording is "authentic" ?  I'm guessing careful and constant calibration of instruments  and use of clever mathematics to reduce uncertainty errors.  Are aesthetically pleasing photographs ever of scientific merit beyond say publicity and outreach?

Jim

[sn2006gy]

I'm curious about this topic as well, mostly because what we enjoy/share and publish is almost entirely inconsistent

We buy all this gear, calibrate for perfection and we strive for dark skies, long integrations and perfect data. Those who push the limits of capturing based on trying to capture data gently, seem to build a smaller and smaller audience of only those who understand the difficulty of such, but those who saturate, blow out colors,  over stretch contrast and make things look vibrant get a huge audience because you don't need to understand any of the complexity to be impressed by the sheer magnitude of what was imaged.

Obviously, an audience doesn't matter but when it comes to creativity, art and imaging - without sharing it with people and without feedback, you'll probably burn out and move along or your interest wasn't in the imaging to begin with - you like building, making, tweaking, experimenting.

Take reddit for example, some very impressive images of hundreds of hours of work get a few upvotes and a few comments, but generally don't rank up. BUT.. if you had a dslr and a telescope that more people know and you got an image that is bright and stretched you can get 1000s of likes.... i look at those images and i'm like "wow, look at those gradients, look at those blown out stars, look at the loss of color, dust motes/donus, bad colors, can't make anything out"... but... this is more relative to the general population then my expensive dark sky setup so most people see "good" where i see "problems" and their scale of good is much different than mine because its more relative to their experience.

So in the end, I guess do what you want to do, find your audience and realize not everyone's context is equal.

[vlaiv]

I'm starting to feel that we are entering philosophical domain here and we are really asking - how can we tell if measurement is authentic?

Let's say that we agree that measurement is authentic - as a postulate. Taking image is actually a measurement.

Both professional and amateur image start in the same way - there is not much difference there - it starts as measurement. What happens next makes the distinction. Scientists do know it is a measurement and they see it as data that needs to be handled carefully. For them image is just visualization of the data and you need to preserve data integrity first.

I guess amateur astrophotographers see that data as basis for creating the image - like clay that will end up being a vase. I actually like to see myself as belonging to scientists camp but in reality what happens is that I switch to the other side - just because I want to make presentable image out of the data.

When called for - I can perfectly preserve data integrity and still make image and if you follow certain rules - you can take data and make simulation of what can be seen with eyes (thus preserving authenticity on several levels like we discussed above). This is because rules for that are well defined.

Maybe good distinction would be - if you can take image and still do for example photometric measurement on it - then it is authentic on that level - it preservers intensities. If you can make relatively accurate astrometric measurement - then it preserves shapes.

If you can determine stellar class by examining thricromatic value of it - then it is authentic in color on one level. If it is displayed on some display device and produces the same color as actual star would produce to human eye - then it is authentic in color on another level.

Way you handle your data can ensure that authenticity is either preserved or destroyed.

For example - I can do photometry on stretched image (non linear) - if I know exact parameters used to stretch image. Even pretty image can be used for this as long as rule is written down with image (as metadata or whatever). We could argue that such image is authentic in that regard?

But if you open photoshop and just fiddle around with curves until you like result - well, unless you save that action - even yourself can't really repeat that 100% and let alone do inverse of it.

 

[andrew s]

1 hour ago, saac said:

That is a good summation vlaiv.  So here is a different but related question.  I think we would agree that professional astronomers/cosmologist are more concerned with the integrity of the data in their images and less with the aesthetic.  I'm guessing that for the likes of their spectroscopy analysis then instrument response, spectral range, resolution etc are the important concerns. What else is important , what are the main data streams that they are looking for - position, velocity, rotation ?  How do they ensure that what they are recording is "authentic" ?  I'm guessing careful and constant calibration of instruments  and use of clever mathematics to reduce uncertainty errors.  Are aesthetically pleasing photographs ever of scientific merit beyond say publicity and outreach?

Jim

Yes even us amateur spectroscpists aim to get as close as possible to the arrival rate of photon per resolution unit above the atmosphere.  This involves careful calibration of the instruments and images using well known methods and more often than not working differentially using standard or secondary spectroscopic standard stars. Similarly with photometry it is done mainly differentially as few of us have photometric skies. 

I am currently trying to improve the way this is done for slitless spectra which have particular issues with flat fielding as the response is wavelength dependent.

Regards Andrew 

[saac]

1 minute ago, andrew s said:

Yes even us amateur spectroscpists aim to get as close as possible to the arrival rate of photon per resolution unit above the atmosphere.  This involves careful calibration of the instruments and images using well known methods and more often than not working differentially using standard or secondary spectroscopic standard stars. Similarly with photometry it is done mainly differentially as few of us have photometric skies. 

I am currently trying to improve the way this is done for slitless spectra which have particular issues with flat fielding as the response is wavelength dependent.

Regards Andrew 

Andrew what is meant by "photometric skies" 

Jim 

[vlaiv]

6 minutes ago, sn2006gy said:

I'm curious about this topic as well, mostly because what we enjoy/share and publish is almost entirely inconsistent

Interestingly enough - even people who take processing very seriously and try not to push their data beyond what it can deliver, that like tight stars and detail and want their denoising to be subtle and unnoticeable - can't make repeatable results.

This is the reason why we have this discussion in the first place and the reason why there are long held beliefs - that "there is no actual true color of the object and you can do what you want" or "no image is authentic" or whatever.

But that is not reality. Two people should be able to produce same looking image of celestial object using different gear if they agree on basic set of rules - like use authentic color, stretch luminosity in certain range (like mag27 is black point mag18 is white point, gamma is set at 4, etc ...)

It is just when we define exact protocol that we can get repeated results. When you take two different DSLR cameras and take image of object, you will get the same image provided you use the same settings and the rest of the protocol is defined at factory.

[andrew s]

2 minutes ago, saac said:

Andrew what is meant by "photometric skies" 

Jim 

They are "uniform" and stable during a nights observations do only simple air mass corrections need be made. Regards Andrew 

[andrew s]

Could not resist this 

Regards Andrew

[saac]

In aviation terms Situational Awareness is defined as:

" ^{is the perception of environmental elements and events with respect to time or space, the comprehension of their meaning, and the projection of their future status"}   

It features as a key part of Crew Resource Management  and pilots are well schooled in the dangers of loss or compromise of situational awareness. In the early days of aviation and,  even today it is  still a recurring feature in accident investigations where pilots have been unable to reconcile what their instruments are telling them and what they are experiencing.  It would appear that when the brain has set itself on a certain course of action/belief  then it can in some circumstances be difficult if not impossible to change that. I've only ever experienced the feeling once where my senses were presenting me with conflicting information and it was scary to say the least.  I was caught in a heavy snow drift travelling up the A9 heading back to RAF Lossiemouth . The snow was falling at such a angle that I was convinced the car itself was drifting across the dual carriageway.  I pulled over into the layby quite shaken at this stage but again as the snow was driving horizontally across the windscreen I was convinced the car was still moving !  No matter how many times I checked the handbrake or looked at the speedometer I eventually had to turn the engine off and get out of the car.  I still get a shiver thinking about it !!  Imagine the stress and confusion something like that can cause a pilot  - no layby at 30,000 ft!! 

Jim 

Edited January 1, 2021 by saac

[sn2006gy]

2 hours ago, vlaiv said:

Interestingly enough - even people who take processing very seriously and try not to push their data beyond what it can deliver, that like tight stars and detail and want their denoising to be subtle and unnoticeable - can't make repeatable results.

This is the reason why we have this discussion in the first place and the reason why there are long held beliefs - that "there is no actual true color of the object and you can do what you want" or "no image is authentic" or whatever.

But that is not reality. Two people should be able to produce same looking image of celestial object using different gear if they agree on basic set of rules - like use authentic color, stretch luminosity in certain range (like mag27 is black point mag18 is white point, gamma is set at 4, etc ...)

It is just when we define exact protocol that we can get repeated results. When you take two different DSLR cameras and take image of object, you will get the same image provided you use the same settings and the rest of the protocol is defined at factory.

You can make repeatable results and there is "relative true color"

I wouldn't focus the repeatability on ensuring every sensor and every scope is the same, I'd get the repeatability on understanding the masters and how those masters were processed to whatever creative means.

On my images, I don't denoise.  I moved  my gear to a B1 sky and take enough data that denoising now seems to distract or cause more harm than good. I realize everyone can't do this, thus I share my data and I share my steps and those steps are repatable that if you follow them, you can have the same image output. One could use scripts to programmatically do the processing as well so its not how much you slide the slider or stretch the stretch, but rather a re-usable algorithm that's precise enough to have the desired effect on a fully calibrated system (monitors and such)

I guess my point is more of, none of this matters.  In one aspect, you're impressing a smaller and smaller group of people that obsess about it and often the end result is they will find something else wrong with your effort or feel hyper competitive to one up you.   On the other side of the spectrum, doing something with literally nothing and getting a nice image finds a lot more traction and a lot more reward for much less effort.

The discussions about true color are funny...  and often times used to confuse objective vs subjective discussions and not the actual experiment or practices that got us to where are are. In the end, if you know that person did good because it wasn't a modded dslr and they got that good of an image, more people probably recognize that than someone who has a top of a line image with a few non round stars that gets no recognition because someone else has just as sharp of an image with a slightly better reducer and used a name brand camera that the people upvoting give more street cred to.

It's really fascinating and odd...

Its's science and it's art and making it only art or only science is worse than letting it be.