I don't think you should have variable width. The only width that is easy to
do is when it's exactly equal to the number of bins you requested for the
histogram. Support that only, is fine with me.
-- lg
On Jul 3, 2012, at 8:21 PM, Stefan Stavrev wrote:
> I like the "height set by u
Requested feature from work. I think adding this lets somebody get rid of some
other tool.
IBA::render_text() is a very basic "render text into the image pixels"
functionality, exposed on the command line as oiiotool --text. You can fully
trick it out with optional arguments like:
oiiotoo
On Jun 30, 2012, at 3:17 AM, Chris Foster wrote:
> In some ways it bothers me that we have both these cases munged into
> one class. In particular, if we allow a valid "off the edge" state when
> being used for random access, I can't imagine the code being as
> efficient as possible.
It's alread
I like the "height set by user" part, can be done easily. Will need to
specify min and max height though, for example 100 to 512. In Photoshop the
histogram has fixed height 100, smaller heights make no sense, since we
want the thing to be visible.
But not sure about having variable width. Since I
On Jun 26, 2012, at 2:24 PM, Chris Foster wrote:
> Side note: If we have TiledIterator, should we make it part of the
> interface that Iterator visits pixels in scanline order? If not, there
> doesn't seem much merit in even having a distinction.
It does beg the question about whether, if Tiled
On Jun 26, 2012, at 12:32 PM, Brent Davis wrote:
> > As an aside, what do you think about "TiledIterator" instead of
> > "UnorderedIterator" as a more descriptive name?
> I think that "TiledIterator" could imply that "Iterator" doesn't use tiles,
> even though it does (when using the cache, anyw
How about width equal to the number of bins, height set by user?
I worry that fixing width at 256, it's going to be tricky or misleading to have
a different number of bins.
On Jul 3, 2012, at 3:45 PM, Stefan Stavrev wrote:
> Hi Will,
>
> You might have some ideas about this. For simplicity sa
Hi Will,
You might have some ideas about this. For simplicity sake I made a decision
to draw the histogram in a fixed window 256x256. In Photoshop and After
Effects the histogram windows are of fixed size too. But in Matlab I
believe the image adjusts itself in size, it is not fixed. Usually the
h
Yeah, we actually discussed that in an earlier thread. You can see what the
underlying histogram calculating function looks like here:
https://github.com/StefanStavrev/oiio/commit/6c7a7c121ae662e5ae348648c322bf789b95ee24
It has control over the number of bins, min, and max, as well as an option
Should granularity of bins also be a parameter? What is a sane
default for floats? Would it make sense to have such a parameter be
expressed in terms of size of the bin, or number of bins between the
high and low? Histograms are one area where I think things are
actually conceptually cleaner in
Banish the idea of "255" from your head forever. Everything is floating point,
and a 255 in an 8-bit file will map to 1.0.
I would say that you should default to 0-1 but allow manual override of min/max.
On Jul 3, 2012, at 12:20 PM, Stefan Stavrev wrote:
> I am working on histogram drawing.
>
On Wed, Jul 4, 2012 at 5:20 AM, Stefan Stavrev wrote:
> I am working on histogram drawing.
>
> What are all the possible ranges of values in one channel I should consider?
>
> The usual ranges are 0-1 and 0-255.
Everything is converted to a float for our purposes, so these ranges are
the same (bo
I am working on histogram drawing.
What are all the possible ranges of values in one channel I should consider?
The usual ranges are 0-1 and 0-255. Any other ranges I should have in mind?
___
Oiio-dev mailing list
Oiio-dev@lists.openimageio.org
http://l
Am I wrong to think that HSL (using some perceptual basis for getting luminance
from RGB) would be in all cases superior to ad-hoc HSV or HSI, which would not
have perceptually uniform brightness for all hues or saturations?
On Jul 3, 2012, at 11:44 AM, Stefan Stavrev wrote:
> Jeremy, the box
Jeremy, the box filter for color images I told you about needs hue and
saturation. I can't just apply the grayscale version of the box filter to
all channels independently, from what I have read it leads to bad results.
So I will have a general case where you can apply the grayscale box filter
to a
Any time Jeremy says, "it's ok to not care about what the colors mean", it's
fine by me to take his advice. It's also fairly safe to assume linear response
with Rec709 color primaries, 99% that's what people will want anyway, and the
other .9% will be too close to detect any difference between
Stefan, I have a lot of non-color-space-aware conversion code (e.g., RGB<->HSV,
without asking too many questions about what RGB means) lying around. Tell me
what you need, and maybe I'll have it.
-- lg
On Jul 3, 2012, at 9:55 AM, Stefan Stavrev wrote:
> Thanks for helping out Jeremy!
Exactly!
So if your algorithm assumes an HSV like input, just note this super
explicitly in the documentation. (which channels each component is
in). And then leave it up to the user to perform the conversion.
Which algorithm, by the way, are you thinking of that requires hsv?
-- Jeremy
On T
Thanks for helping out Jeremy!
Just a small example to confirm I understand what you say:
Let's say that some algorithm needs hue and saturation. To achieve the
flexibility you talk about, I will need to keep anything color conversion
related outside of my functions. The documentation will say th
My recommendation for all of these ImageBugAlgo implementations
(contrast, histogram adjust, box filters, etc) is to completely ignore
color space.
I have reviewed the book referenced by Stefan, and I agree that it
makes a strong case that many image processing operations achieve
superior result
Sent on the move...
On 2 Jul 2012, at 21:20, Stefan Stavrev wrote:
> You mean it is complicated because conversion formulas are expensive, or lack
> of agreement in the whole color spaces thing? I mean, there is a lot of
> confusion about the color spaces, different names for things, differe
21 matches
Mail list logo
