[Cin] Fwd: [heroineworshiper/hvirtual] question about clipping in Histogram plugin (Issue #8)

[Andrew Randrianasulu]

well, Andrea, can I copy your email to GitHub issues or you prefer to
explain it there yourself? I think your use case goes a bit beyond just
testing and into actual workflow category?

---------- Forwarded message ---------
От: Adam Williams <notifications at github.com>
Date: пт, 10 нояб. 2023 г., 21:16
Subject: Re: [heroineworshiper/hvirtual] question about clipping in
Histogram plugin (Issue #8)
To: heroineworshiper/hvirtual <hvirtual at noreply.github.com>
Cc: Andrew Randrianasulu <randrianasulu at gmail.com>, Author <
author at noreply.github.com>


The big problem with histogram is histograms are expected to clamp.  The
graph has to be changed to show a Y range beyond 0-1, you need some more
parameters to define min & max Y.  You could make a new histogram plugin
which implemented the new options.  It gets a bit bloated so the question
is again why you need histogram to support more than 0-1?




Sent: Thursday, November 09, 2023 at 11:14 AM
From: "Andrew Randrianasulu" ***@***.***>
To: "heroineworshiper/hvirtual" ***@***.***>
Cc: "Adam Williams" ***@***.***>, "Comment" ***@***.***>
Subject: Re: [heroineworshiper/hvirtual] question about clipping in
Histogram plugin (Issue #8)





A case of is it worth the effort to support values <0 & >1 & the answer was
no. If you can find a good reason & actually fix everything including the
opengl functions no-one uses, maybe it could work.


well, I looked into playback3d.C and some shaders there surely uses clip?
But biggest problem for me I do not know how to init fp32 fbo/pbo, so it
all remained clipped.

But just switching textures to 32f made big difference with test file
Andrea provided and fade out.

https://www.youtube.com/watch?v=7GOhtJciHBk

I tried to follow advice given in
https://community.khronos.org/t/glx-ati-pixel-format-float-under-linux/47685/2
(but using more modern GLX_RGBA_FLOAT_BIT_ARB).

use-case is exr/tiff from some photography software or rasterizer, where
values > 1.0f exist. (ffmpeg seems to convert most/all video to integer
formats, but may be in last ffmpeg they added it -
https://git.ffmpeg.org/gitweb/ffmpeg.git/commit/479747645f795b6f4f376578ea1556409f943c31
- "avutil/pixfmt.h: add native-endian RGB32F and RGBA32F formats")

—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you commented.Message ID: ***@***.***>

—
Reply to this email directly, view it on GitHub
<https://github.com/heroineworshiper/hvirtual/issues/8#issuecomment-1806198961>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AJSS7TJ5T7MREFWQWTBCADLYDZVO5AVCNFSM6AAAAAA7E3NKDOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMBWGE4TQOJWGE>
.
You are receiving this because you authored the thread.Message ID:
<heroineworshiper/hvirtual/issues/8/1806198961 at github.com>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cinelerra-gg.org/pipermail/cin/attachments/20231110/faaf92de/attachment.htm>