Re: [pygame] 10 bits per color

[James Paige <Bob@xxxxxxxxxxxxxxxxxxx>]

On Thu, Oct 01, 2009 at 01:41:28PM -0400, PierreLafrance1@xxxxxxxxxxxx wrote:
> Nirav Patel wrote:
> > René was referring to average cameras doing 8-16 per pixel, not per
> > color.  Though, in a Bayer sensor, I guess you could calculate it
> > either way.  The Foveon sensor is of course different.  Even if the
> > display is not capable of showing 10 bits per color, it makes sense to
> > capture at it to allow for post processing.  You could mess with the
> > levels or extend contrast without running into banding or aliasing or
> > having to dither.
> > 
> > Nirav
> > 
> > On Thu, Oct 1, 2009 at 9:53 AM, PierreLafrance1@xxxxxxxxxxxx
> > <PierreLafrance1@xxxxxxxxxxxx> wrote:
> >> René Dudfield wrote:
> >>> On Thu, Oct 1, 2009 at 2:34 PM, PierreLafrance1@xxxxxxxxxxxx
> >>> <PierreLafrance1@xxxxxxxxxxxx> wrote:
> >>>>> Sounds lik yet another gimmick to get uneducated folk to buy another TV
> >>>> LOL, I like this one.
> >>>> But I'm not sure I understand your statement on NTSC.
> >>>>
> >>>> This is what my research subject is all about.  My boss asked me to
> >>>> optimize my hardware design (0.35u CMOS image sensor) to fit eyes and
> >>>> equipement limitations.  But I need first to create RGB101010 software
> >>>> to see if DeepColor makes sense or not, before optimizing CMOS chips.
> >>>> Since I like Python and I have Pygame experiences, I wanted to do that
> >>>> software with Pygames.
> >>>>
> >>>> Thanks
> >>>>
> >>>> Pierre
> >>>>
> >>> sounds like a fun project!
> >>>
> >>> from my understanding of non-cutting-edge cmos sensors, they give out
> >>> 8-16bit per pixel of range?  Then for color they use a bayer filter,
> >>> to filter out the various wave lengths of light... usually RGB
> >>> filters.  Of course if you took 1/3rd the resolution of that image,
> >>> then it could be 16/16/16 per pixel I guess.
> >>>
> >>> In the same sense you could work with a pygame Surface - but think of
> >>> the colors in a different resolution.  Double the resolution, and
> >>> you've doubled the color depth.
> >>>
> >>> But you might want to check out exr... http://www.openexr.com/  and
> >>> it's python bindings...
> >>> http://excamera.com/sphinx/articles-openexr.html
> >>>
> >>>
> >>> ... thinking of it a bit more... you should be able to display higher
> >>> depths via opengl.
> >>>
> >>> With pygame and opengl you should be able to set a higher bit depth...
> >>> if your driver supports it.
> >>> see: http://pygame.org/docs/ref/display.html#pygame.display.gl_set_attribute
> >>>
> >>> Then send your data to the relevant bit depth gl texture, and you
> >>> should be able to display whatever your driver supports
> >>>
> >>>
> >>> cu.
> >>>
> >>>
> >> Hi all,
> >>
> >>> they give out 8-16bit per pixel of range?
> >> René, you put the fingers exactly in the center of our question: Why
> >> camera gave 12, or even 14 bits per color, if Windows, Display, Grahic
> >> cards are limited to 8 bits per color ?
> >> HDMI 1.3 is 1 year old only, Windows will support this (correctly ???)
> >> in 1 month.  Only specialized software are supporting this.  But who has
> >> PhotoShop at home ?
> >>
> >> Ok, some print pictures, but in my opinion, there are even less dynamic
> >> range on paper than on LCD (this need to be confirm).
> >>
> >> About Bayer, some sensor work differently, if you are interrested, look
> >> at Foveon sensor (bought by Sigma), and this patent:
> >> http://www.google.com/patents/about?id=_mzIAAAAEBAJ&dq=Yves+Audet
> >> That's my boss's patent, the CMOS sensor I'm working on
> >>
> >> And yes it's a fun projet, mixing Pygame with chips design :-D
> >>
> >> Pierre
> >>
> >>
> > 
> > 
> Hi
> Sorry I was not clear.  Most camera now have RAW mode with more than 8
> bits per color (in Bayer color space).  Some hi-end camera (like Nikon)
> capture images with 14bits per pixel
> 
> Why is this, when current display are limited to 8 bits per color ?
> 
> It's interresting to see all these answers from Pygame users.  I'm
> really happy to see there a lot of knowledge people in this mailing list.
> 
> Thanks
> 
> Pierre

Suppose, to rephrase Brian's example, you ued a digital camera to take a 
photograph in a cave (with no flash) The picture would be really dark. 
If you loaded into photoshop and increased the brightness 1000% the 
light levels might look more normal, but if you used an RGB888 camera a 
lot of color depth would be lost. However, if you do the same thing with 
an RGB141414 camera, you will still lose color depth, but the end result 
will still have a heck of a lot more color in it than with the RGB888 
camera.

I actually did something like this myself. When I was on vacation to 
Carlsbad Caverns, I took some pictures with no flash, relying only on 
the ambient light that the Park Service has installed in the cave. I was 
using an old digital camera. I don't know the color depth but it was 
probably RGB888 or RGB666 or something like that. On the camera these 
pictures looked pure-black as if I had taken them with the lenscap on. I 
thought they were total losses, until  I put them on my computer and 
massively increased the brightness with The Gimp. Those photos became 
visible again, but their colors were all skewed to greenish hues.

You can see a few of them here: 
http://gilgamesh.hamsterrepublic.com/gallery/CarlsbadCaverns?page=6
The brightness-adjusted photos are really easy to spot in comparison to 
the flash-photos because of the color-shift.

---
James Paige