Re: [CinePaint] Really basic script for batch converting

Frank Peters <[email protected]> Tue, 24 Nov 2009 21:22:40 -0500
Newsgroups gmane.comp.video.cinepaint.user
Message-ID <[email protected]>
On Wed, 25 Nov 2009 00:54:03 -0000 (GMT)
[email protected] wrote:

> 
> I've been trying out adding a minute amount of noise to an 16/32bit image
> converted from 8bit in Cinepaint as per early Photoshop 15bit + 1bit noise
> comments in last email and so far I'm happy with outcome.
> 
> I've taken an 8bit image in Cinepaint, convert to 16 or 32bit, add a layer
> by duplicating the background, use 'noisify' to add .008 noise to each of
> the 3 channels.
> 
> Checking the histogram to see if it is all smooth.
> 

Did it appear gapless?

You must realize that the histogram tool in cinepaint, or in other image
processing software, does have the resolution to always show the gaps.
But the gaps from conversion are there.

Here is an illustration:

I created a grayscale image, with 8-bit depth, containing a linear ramp of
values from 0 to 255.  From this image a data file was created containing
very acrurate histogram values.  This data was plotted using gnuplot.  It is
shown in this link:

http://home.comcast.net/~frank.peters/ramp255.png

Notice that the distribution is perfectly uniform across all values,
as expected.  The histogram tool in Cinepaint would show an identical
distribution for this image.

Now, this 8-bit image file was then converted into 16-bit.  The data
for a histogram was very accurately obtained and plotted in gnuplot.
The result is shown here:

http://home.comcast.net/~frank.peters/ramp255to65535.png

Notice the presence of many gaps all across the range. (The plot
consists of many steps going repeatedly from 0 - 1000.)  Cinepaint
however would not reveal these gaps.  In Cinepaint, the distribution
would appear identical to the previous image.

65536 steps is an enormous range and it is not practical to
create a histogram tool that will always reveal the fine detail.
However, specialized software can show the detail.
 
Even this quick illustration does not do full justice to the
changes.  In fact, if the horizontal axis on the second image
were to be expanded, the gaps would be seen to be even greater,
while the first image would still be perfectly uniform. 
But I did not want to spend a lot of time on this.

Frank Peters


------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july