Brilliancy vs Brightness

[Serg]

Hi everyone,
Your help is needed in testing Brilliancy Static Mono metric.
Please have a look at the picture featuring diamonds which have almost same average brightness but differ in brilliancy.

They are sorted and placed according to their brilliancy metric in the descending order, the Brilliancy metric value can be found in the lower right corner of every image.( Brightness value is in the top right corner)
Please let know if your visual perception of the diamonds brightness order differs from the order at the picture

Second image has same diamonds with original brightness

 

[drk14]

This is not an answer to your question, but I think that your analysis will be confounded by the fact that some of the diamond images are saturated, especially in the brightness-normalized images. You may get better results by using a nonlinear transformation (e.g., gamma) to normalize the brightness.

Looks like an interesting project, though!

 

[Serg]

This is not an answer to your question, but I think that your analysis will be confounded by the fact that some of the diamond images are saturated, especially in the brightness-normalized images. You may get better results by using a nonlinear transformation (e.g., gamma) to normalize the brightness.

Looks like an interesting project, though!

• Some facets have to be saturated , same as for naked eye observation
• if gamma image is not similar to inverse monitor gamma then result is wrong
• source data has 12-14 bits

 

[drk14]

Please let know if your visual perception of the diamonds brightness order differs from the order at the picture

Did you mean to ask us to visually assess "brightness" or "brilliancy"? The normalized images have nearly the same brightness, and are not sorted by brightness, so I'm guessing you meant to request for us to assess the "brilliancy". I think an explanation of what you mean by this word would be helpful.

Personally, I associate "brilliance" with scintillation (which cannot be assessed from a static image) and brightness (which has been normalized in your sample). Based on the images, I believe that your brilliance metric is primarily based on contrast.

If I take a more subjective approach to defining "brilliancy", and rank the diamonds according to my best guess as to whether the diamond will have good scintillation when viewed in real life, then my rankings will not match the computed brilliancy metric in several cases. For example, 042150 should be much lower on the list, while MSSRBC16 should be ranked much higher IMO.

But it's hard to answer the question without knowing exactly what you're asking...

 

[Serg]

Drk,
First of all you may just ignore my information about brightness and just check order of diamond according your understanding of Brightness or Brilliancy.

Yes, diamond tilting gives biggest contribution in Brilliancy( temporal rivalry )

other 2 parts of Brilliancy are Stereo rivalry and Intensity Rivalry

more details you may find here
http://www.gem.org.au/ckfinder/userfiles/files/GAA_Journal_V25_No3_web2(1).pdf

Page:90
"
We can see that the essential condition for brilliance to emerge is the presence of a ‘visual rivalry’ for the human brain originating from contradictory information. That is when the human brain combines reasonably different images seen by two different eyes into one form. This rivalry can have the following origins:
1. Intensity rivalry: a certain distribution pattern of areas with various intensities in space, when areas with the same intensity appear as different ones. In this case, the human brain subjectively extends the dynamic range of picture brightness in order to comprehend all the information from a complex picture. 2. Temporal rivalry: when intensity varies with a certain frequency within the same area of space.
3. Binocular rivalry: when two different eyes see two different intensities (or colours) within one space area and the brain combines them into one unstable image.
To resolve these conflicts, the human brain extends the dynamic range of image intensities to visually make individual diamond virtual facets much brighter than they are in reality. They begin to shine, and we observe brilliance. To make it possible for us to see brilliance due to the temporal rivalry, fairly large image segments need to change their intensity slowly as the gem moves. If these are small segments or virtual facets with fast changing intensity, we experience scintillation. Both this motion and stereo vision contribute a lot to the amount of perceived brilliance, which must not be overlooked in designing new cuts to maximize beauty. Moreover, we believe that it is only motion and stereo vision that define the main, and most likely the entire contribution to the perception of brilliance by a consumer. Therefore, for the purpose of developing a first approximation for brilliance metrics, we define brilliance as follows:
Diamond brilliance is an illusion caused by the fact that the perceived brightness of the object significantly exceeds its actual brightness.

This may occur when several Figure 3.2 The Hermann-Hering grid illusion and scintillation grid. Note the dark spots are more vivid when viewed with both eyes than with one eye closed. Hermann L (1870). "Eine Erscheinung simultanen Contrastes". Pflügers Archiv für die gesamte Physiologie 3: 13–15 How diamond performance attributes: Brilliance, Scintillation and Fire depend on human vision features | 82 – 121 91 parts of the object with different brightness are observed within a single space-time domain. In the case of a diamond, these parts are its virtual facets. Two virtual facets belong to the same space-time domain when an observer cannot separate them due to spatial or temporal limitations of the sense of vision.
"

page 94

 

[Serg]

DRK,

while MSSRBC16 should be ranked much higher IMO.

Even higher than MSSRBC13?

 

[gm89uk]

Is there a way you can create it in an online tool where people can click hold and move to reshuffle? You may get far more responses. At the moment you have tediously long names for labelling and reordering is quite tiresome for people that would want to help quickly.

 

[drk14]

Even higher than MSSRBC13?

In my subjective opinion, yes. The bottom half of the image of MSSRBC16 is not very bright, but I "know" (=imagine) that a very small tilt of the diamond will light up the bottom half just as bright as the top, so this is what I "see" when I evaluate the brightness of that stone.

 

[drk14]

Drk,
First of all you may just ignore my information about brightness and just check order of diamond according your understanding of Brightness or Brilliancy.

Yes, diamond tilting gives biggest contribution in Brilliancy( temporal rivalry )

other 2 parts of Brilliancy are Stereo rivalry and Intensity Rivalry

By the information in your post and in your 2013 article in The Australian Gemmologist, I'm guessing we are supposed to perceive some type of brilliance illusion when looking at the diamond images, analogous to the Hermann-Hering grid illusion. The prominence of the HH grid illusion is known to depend on the dimensions of the pattern -- thus, wouldn't we need to control the image size and viewing distance to properly perform the test?

 

[gm89uk]

In my subjective opinion, yes. The bottom half of the image of MSSRBC16 is not very bright, but I "know" (=imagine) that a very small tilt of the diamond will light up the bottom half just as bright as the top, so this is what I "see" when I evaluate the brightness of that stone.

But when assessing the image should you not assess it based on the static image? The algorithm if I understand is outputting a value based on this 1 slice, it'll be impossible to calibrate if you're basing your answers on dynamic presumptions.

Maybe the algorithm would take an average at various tilts before pursuing an answer in a more developed edition which would then compensate for your presumption.

 

[Serg]

By the information in your post and in your 2013 article in The Australian Gemmologist, I'm guessing we are supposed to perceive some type of brilliance illusion when looking at the diamond images, analogous to the Hermann-Hering grid illusion. The prominence of the HH grid illusion is known to depend on the dimensions of the pattern -- thus, wouldn't we need to control the image size and viewing distance to properly perform the test?

Yes: distance, Linear scale , monitor brightness, background , room illuminance are very important for precise sorting. I found best results for retina last generation notebook monitors( OLED 10 bit TV are much brighter of course, but a TV has a lot of internal tone mapping), in dark room, 0.5m-1m distance,
with 10x linear size( angular magnification is 3-5x)
usually we use 1 euro coin to check monitor scale and 0.5m distance . see sample below

 

[Serg]

Is there a way you can create it in an online tool where people can click hold and move to reshuffle? You may get far more responses. At the moment you have tediously long names for labelling and reordering is quite tiresome for people that would want to help quickly.

We use similar sorting instrument in internal tests. we use it to sort movies.
It is not ready yet for public work.
On PS I have not goal to receive precise sorting.
My current task to check Brilliancy observation/ visualization in mono static photos.

 

[Serg]

In my subjective opinion, yes. The bottom half of the image of MSSRBC16 is not very bright, but I "know" (=imagine) that a very small tilt of the diamond will light up the bottom half just as bright as the top, so this is what I "see" when I evaluate the brightness of that stone.

Please do not "Imagine video " when you sort static images. Sorting by video is very different.
See the result video sorting for same samples

 

[arkieb1]

I think you need to use a video of actual stones, the static pic IMHO is flawed because my eyes automatically assess stones with more whiteness or greater/larger white areas in your static photo as having greater brightness when clearly this has more to do with the facet pattern of some of the stones than the actual numbers/ratings you have given them. Therefore, if they seem more bright/white in the static photo you can be mislead into automatically assuming that they would seem more brilliant.....

 

[valeria101]

Nice to see this post in the morning ! Obviously needed metric ...

But when assessing the image should you not assess it based on the static image?

Why ...

 

[flyingpig]

I believe that your brilliance metric is primarily based on contrast.

+1

I have no idea when you guys talk about dynamic range and static, but....

Whenever the diamond industry explains brilliance or perception of brilliance, they completely lose me. I consider brightness, light return and brilliance about the same. They often explain some amount of contrast increases brilliance. I understand what it means, but disagree.

https://www.pricescope.com/communit...ance-diamond-photography.227610/#post-4108627

In which photo, do you think the diamond is more brilliant?? with contrast? or without contrast?
I say without contrast, in the photo AND in real world, at least according to my definition of brilliance.

 

[valeria101]

Need to start the day (no time), let me just say -

The static brilliancy matter reminds me of the impossible task of appreciating opal in static shots - it ended in video, I feel. [not that history is ever straightforward ... ] Of course, nobody can controll much about the optics of the object there, only the presentation ... I imagine you are looking for a metric to target by new cut designs & guidelines (toward that mentioned 'controll'), I am talking about a search tool that fits better with The Eye (like video did for the uncontrollable opal brilliancy); not sure these the twine could meet ...

What a beautiful problem !

 

[Serg]

+1

I have no idea when you guys talk about dynamic range and static, but....

Whenever the diamond industry explains brilliance or perception of brilliance, they completely lose me. I consider brightness, light return and brilliance about the same. They often explain some amount of contrast increases brilliance. I understand what it means, but disagree.

https://www.pricescope.com/communit...ance-diamond-photography.227610/#post-4108627

In which photo, do you think the diamond is more brilliant?? with contrast? or without contrast?
I say without contrast, in the photo AND in real world, at least according to my definition of brilliance.

these 2 images have exactly same objective average brightness. Do they really have similar Brilliancy for you?


To calculate Brilliancy metric we use Brightness, Contrast, Pattern (static and dynamical pattern). just average and contrast are not enough to receive good correlation with Human score .

 

[flyingpig]

The first diamond (1.13) is definitely more brilliant than the second (0.93), because white is more whiter and brighter in the first image than in the second one.

 

[Serg]

The first diamond (1.13) is definitely more brilliant than the second (0.93), because white is more whiter and brighter in the first image than in the second one.

in same time the first diamond( MSS13, score 1.13) has more darker black areas than the second diamond.
Please compare main facets under table.
when white is brighter and black is darker then the contrast is higher. ( with same average brightness)

 

[Serg]

sorting by video with original brightness


Link to similar table with video I am going to publish in few weeks .
We use videos from cutwise like :
https://cutwise.com/~lCQy

 

[OoohShiny]

Link to similar table with video I am going to publish in few weeks .
We use videos from cutwise like :
https://cutwise.com/~lCQy

That video comparison is AWESOME - very high quality and very useful for comparison of cut performance!

 

[arkieb1]

When you compare those two diamonds it's obvious but humour me and now put up the Old European Cut, the Emerald cut and say one of the cushions that scores 1.05 or 1.04 now compare the white area in the Old Euro to the cushions and even the emerald, it's not that easy to rate them any more is it?

 

[Serg]

When you compare those two diamonds it's obvious but humour me and now put up the Old European Cut, the Emerald cut and say one of the cushions that scores 1.05 or 1.04 now compare the white area in the Old Euro to the cushions and even the emerald, it's not that easy to rate them any more is it?

Do you want compare Brilliancy for cut selection like this: https://cutwise.com/~gvC3 ?

Please push Play button then switch off/on greyscale tick to compare movies in B/W mode

 

[drk14]

these 2 images have exactly same objective average brightness. Do they really have similar Brilliancy for you?

OK, I think I'm beginning to understand what you are asking us to do...

The computed brightness score (which is similar for all specimens in the normalized images) is the actual average brightness of all pixels that make up the diamond image. However, as illustrated in the two images (MSSRBC17 vs HM-1136), our brains perceive the overall brightness as some kind of weighted average that appears to be biased towards brighter pixels. Thus, most people would perceive the MSSRBC17 image to have a higher overall brightness.

So the way I interpret the stated task is for us to attempt to guess/predict the unweighted average brightness that would be determined by an unbiased computer, and to rank these predictions relative to each other. Per the example above, because of the idiosyncrasies of human vision, our predictions will not match the actual computed ranking. By comparing our rankings to the computer rankings, you can then improve the algorithm for your "Brilliancy Static Mono" metric.

Is this a fair description of what you're after?

 

[Serg]

OK, I think I'm beginning to understand what you are asking us to do...

The computed brightness score (which is similar for all specimens in the normalized images) is the actual average brightness of all pixels that make up the diamond image. However, as illustrated in the two images (MSSRBC17 vs HM-1136), our brains perceive the overall brightness as some kind of weighted average that appears to be biased towards brighter pixels. Thus, most people would perceive the MSSRBC17 image to have a higher overall brightness.

So the way I interpret the stated task is for us to attempt to guess/predict the unweighted average brightness that would be determined by an unbiased computer, and to rank these predictions relative to each other. Per the example above, because of the idiosyncrasies of human vision, our predictions will not match the actual computed ranking. By comparing our rankings to the computer rankings, you can then improve the algorithm for your "Brilliancy Static Mono" metric.

Is this a fair description of what you're after?

I believe you mean Lightness, did not you?( and Webers, Stevens, Krueger laws?)
I used here brightness in diamond as synonym of luminance . ( as bulb brightness)

"Luminous flux is the total amount of light coming from a source, such as a lighting device. Luminance, the original meaning of brightness, is the amount of light per solid angle coming from an area, such as the sky."

Lightness typical depends as cubic root from luminance.( you need increase luminance in 8 times to increase lightness in 2 times)

in any case it is not important to my original question. I asked to check my sorting by brilliancy . it is not very important that brightness have all these diamonds. It was my mistake to mention brightness .

 

[drk14]

I believe you mean Lightness, did not you?( and Webers, Stevens, Krueger laws?)
I used here brightness in diamond as synonym of luminance . ( as bulb brightness)

Actually, I used the word "brightness" to indicate pixel intensity (e.g. 0-255 in an 8-bit image). The images of MSSRBC17 and HM-1136 have an average pixel intensity of 159/255=62% and 154/255=60%, respectively. However, my subjective impression is that MSSRBC17 has a significantly higher average pixel intensity, presumably because I subconsciously weigh the contribution of brighter facets more heavily.

...but this is not how you want us rank the images?

I asked to check my sorting by brilliancy

In your 2013 article, you include 10 different definitions of brilliance (Box A), and you also make the claim "We know that neither a static image (a photograph) of a diamond on a monitor, nor its printed image has brilliance" (page 89). So I guess I will need a little more explanation about what criterion you want us to use to sort the diamond images.

 

[Serg]

Drk,
I do not ask you to sort diamonds according my Brilliancy definition . I am asking you to sort diamonds according you feeling/understanding of Brilliancy . If you do not see any brilliancy in above photos than my sorting is not wrong.

 

[drk14]

I am asking you to sort diamonds according you feeling/understanding of Brilliancy .

But I have already tried two different ways to evaluate the images based on my own feeling/understanding of "Brilliancy" (first, estimating whether the diamond will have bright scintillation based on its appearance; second, estimating the average pixel intensity), and both attempts were not the "correct" type of assessment that you are looking for...

Here's my third (and probably last) attempt at an operational definition for "brilliancy": I propose to rank the diamonds based on which stone I (subjectively) perceive to be more attractive/desirable/well cut.

 

[Serg]

Here's my third (and probably last) attempt at an operational definition for "brilliancy": I propose to rank the diamonds based on which stone I (subjectively) perceive to be more attractive/desirable/well cut.

It is good option.
Static B/W image definitely has not any Scintillation , Fire, Dead zones.
in same time subjective perceive of diamond beauty depends from such positive scores as Brilliancy, symmetry, Brightness, Shape, pattern and negative score as black zones.
there are 2 options :
1) Include Black zones in to total Brilliancy score
2) use negative score for black zones independently from positive grade for brilliancy