> A color statistics visualization - very intriguing! A few years ago, I
> started to program in C++ a command-line image processor (yip, "Yadgar's
> Image Processor"), which, besides classic computer retro graphics
> filters (see the Max Canada Lynx adaptions on my YouTube, also is meant
> to do such displays, even for entire videos... currently, it's somewhat
> on halt as I intend to re-write the whole thing in C rather than C++.
>

  C???? Why go back?

Post a reply to this message

"Bald Eagle" <cre### [at] netscapenet> wrote:
> "Norbert Kern" <nor### [at] t-onlinede> wrote:
>
> > here is my attempt to depict the colors in images (code in p.b.s-f.)
>
> Very nice.  These look better than what I was fooling with a while back.
> I'll try to envision the function and the "color space" that you're graphing
> this in - but it's too early and I've not had enough coffee yet.   ;)
>

> > like John Wick gets a new look based on strong purples. A similar effect can be
> > seen at recent Avengers: Endgame posters.
>
> Hmmm.  I don't know much about this.  I've only seen some discussion of "color
> matching" of film frames in the LOTR extended extras.   Is that the kind of
> thing you're talking about?
> "The usual"?  For what?  Most movies, or that series?
>
>

>
> "A POV implementation...."   I'm not sure what you mean by that.
> A way to force the color map to be limited to that range of colors?
>
>
> Something interesting that I noticed with these two is that the top graph has a
> ring of missing black, and the bottom one has the black present, and all the way
> around.
>
>
>
>
http://news.povray.org/povray.advanced-users/thread/%3Cweb.5789343436e895bcb488d9aa0@news.povray.org%3E/
>
>
http://news.povray.org/povray.binaries.images/thread/%3Cweb.58cfc0cf857600e8c437ac910%40news.povray.org%3E/?mtop=4224
76
>
>
> There was also another thread around that time where someone was processing
> video and doing an animation of the color palette - can't find it at the moment.


Many thanks - I wasn't conscious about the second link - very interesting.
First I used a cubic space too, but hsl space is more convincing in the end.
Personally I use it as a tool to detect color errors in my images. Here is an
example with the last version of an old image of mine...

Because I render 16 bit png images, there aren't many defects as in 8 bit jpegs.
But what you mentioned, is something different. I included the web origins in
the source (p.b.s-f).
Perhaps the JW3 person didn't process the poster as heavily as the other one -
DP2.

"Orange-Teal" is one of the most common color-grading effects used at
movie-making or by professional photographers.
There are many variants like enhencing blue colors in shadows and orange ones in
highlights and a bit in midtones.

I made a pov version, which gave ugly results and I cannot even find the file...


Norbert

Post a reply to this message

Attachments:
Download 'i2c_vertraeumt.jpg' (658 KB)

Preview of image 'i2c_vertraeumt.jpg'

"jr" <cre### [at] gmailcom> wrote:
> hi,

>
> very nice.  I tried it with some images and got interesting renderings.  one
> image though (png, 8-bit RGB) gives me a problem, it is an at-night photo of a
> street and I expected mostly black spheres.  instead - nothing, no (visible)
> spheres.  does the code not work with "black and white"?
>
>
> regards, jr.

Hmmh - I don't know, how a perfect black and white is depicted in hsl space.
Perhaps it is not a cylinder, but a double cone with black and white as single
points at top and bottom - but I don't think this is the case.

If you used extremely large images as input, you get very small spheres and if
you really have mostly black colors, you get an almost blank result. Perhaps it
helps, if you increase the sphere radius.

Norbert

Post a reply to this message

=?UTF-8?Q?J=c3=b6rg_=22Yadgar=22_Bleimann?= <yaz### [at] gmxde> wrote:

> A color statistics visualization - very intriguing! A few years ago, I
> started to program in C++ a command-line image processor (yip, "Yadgar's
> Image Processor"), which, besides classic computer retro graphics
> filters (see the Max Canada Lynx adaptions on my YouTube, also is meant
> to do such displays, even for entire videos... currently, it's somewhat
> on halt as I intend to re-write the whole thing in C rather than C++.

Sounds nice - can it be implemented in povray? ;-)



> about two complementary colors (albeit not complementary in RGB color
> space)...
>
> See you in Khyberspace!
>
> Yadgar

Orange-Teal look is a color-grading effect to emphasize blue colors in shadows
and orange colors at highlights. It can enhance an mediocre image a lot, if
applied carefully.
Here is a typical example -
http://server.felixhuesken.de/lgg_orangeteal/orange-teal_blockbuster_look_jurassicworld_color_grading.png
..

Norbert

Post a reply to this message

Hi(gh)!

On 13.05.19 20:34, Norbert Kern wrote:

> Sounds nice - can it be implemented in povray? ;-)

Theoretically, yes - but it wouldn't be such a good idea, as it would be 
excruciatingly slooooooow!

> Orange-Teal look is a color-grading effect to emphasize blue colors in shadows
> and orange colors at highlights. It can enhance an mediocre image a lot, if
> applied carefully.
> Here is a typical example -
>
http://server.felixhuesken.de/lgg_orangeteal/orange-teal_blockbuster_look_jurassicworld_color_grading.png

Hmmm... as I assumed, 180 degrees apart on the hue circle!

See you in Khyberspace!

Yadgar

Post a reply to this message

Norbert Kern wrote on 13/05/2019 18:59:
> Many thanks - I wasn't conscious about the second link - very interesting.
> First I used a cubic space too, but hsl space is more convincing in the end.
> Personally I use it as a tool to detect color errors in my images. Here is an
> example with the last version of an old image of mine...
> 
> Because I render 16 bit png images, there aren't many defects as in 8 bit jpegs.
> But what you mentioned, is something different. I included the web origins in
> the source (p.b.s-f).
> Perhaps the JW3 person didn't process the poster as heavily as the other one -
> DP2.
> 
> "Orange-Teal" is one of the most common color-grading effects used at
> movie-making or by professional photographers.
> There are many variants like enhencing blue colors in shadows and orange ones in
> highlights and a bit in midtones.
> 
> I made a pov version, which gave ugly results and I cannot even find the file...
> 
> 
> Norbert
> 

The result looks good.
Paolo

Post a reply to this message

hi,

"Norbert Kern" <nor### [at] t-onlinede> wrote:
> > very nice.  ...  does the code not work with "black and white"?
> Hmmh - I don't know, how a perfect black and white is depicted in hsl space.
> Perhaps it is not a cylinder, but a double cone with black and white as single
> points at top and bottom - but I don't think this is the case.
>
> If you used extremely large images as input, you get very small spheres and if
> you really have mostly black colors, you get an almost blank result. Perhaps it
> helps, if you increase the sphere radius.

thanks for that.  image size is not a factor, they're all (much) smaller than
the images you used.  I'm guessing that the image is simply too ill-lit.  fwiw,
I've been converting some music related images, album covers and such, and
include one "favourite".


regards, jr.

Post a reply to this message

Attachments:
Download 'inexpressible432hz.png' (582 KB)

Preview of image 'inexpressible432hz.png'

Interesting -
There are those very noticeable "absent lines" radiating from the center, and
some fainter annular spaces going around.

Makes me wonder how you'd take those missing colors and reconstruct the image
with only those.

Or something like that.  ;)

Post a reply to this message

hi,

"Bald Eagle" <cre### [at] netscapenet> wrote:
> Interesting -
> There are those very noticeable "absent lines" radiating from the center, and
> some fainter annular spaces going around.

not always though.  no absent lines apparent in the image Norbert posted on the
13th.

I'm fascinated by how the resulting images always (it seems.  after ~20 tests)
appear different from the expectation.


> Makes me wonder how you'd take those missing colors and reconstruct the image
> with only those.
>
> Or something like that.  ;)

:-)


regards, jr.

Post a reply to this message

"jr" <cre### [at] gmailcom> wrote:
> hi,
>
> "Bald Eagle" <cre### [at] netscapenet> wrote:
> > Interesting -
> > There are those very noticeable "absent lines" radiating from the center, and
> > some fainter annular spaces going around.
>
> not always though.  no absent lines apparent in the image Norbert posted on the
> 13th.
>
> I'm fascinated by how the resulting images always (it seems.  after ~20 tests)
> appear different from the expectation.
>
>
> > Makes me wonder how you'd take those missing colors and reconstruct the image
> > with only those.
> >
> > Or something like that.  ;)
>
> :-)
>
>
> regards, jr.



Me too,

it's a "negative" tool IMHO.
You can easily detect lacking colors.

Norbert

Post a reply to this message