From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <9670ac1dfa9616d21c1f65ffc570b369@quanstro.net>
From: erik quanstrom <quanstro@quanstro.net>
Date: Thu, 10 Sep 2009 17:45:57 -0400
To: 9fans@9fans.net
In-Reply-To: <9ea8b15ff49a68b8ac5d908202c8757d@hamnavoe.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Subject: Re: [9fans] nice quote
Topicbox-Message-UUID: 6cdd4f88-ead5-11e9-9d60-3106f5b1d025

> term% time resample -x 1600 -y 1200 glenda.pic >/dev/null
> 36.07u 0.01s 36.21r 	 resample -x 1600 -y 1200 ...
> term% time resize -b -s 1600 1200 glenda.pic >/dev/null
> 0.91u 0.02s 1.06r 	 resize -b -s 1600 1200 ...
>
> The -b option is for bilinear interpolation.  Without that, it goes
> a bit faster but you get jaggies.
>
> term% time resize -s 1600 1200 glenda.pic >/dev/null
> 0.56u 0.03s 0.71r 	 resize -s 1600 1200 glenda.pic ...
>
> It's in /n/sources/contrib/miller/resize.c

very nice!

i modified resize to accept the same arguments as resample.
it is blistering on my machine.  unfortuntely, jpg still can't
keep up.

; time jpg -t9 2246.jpg | time resize -bx 500 | page
1.96u 0.08s 2.20r 	 jpg -t9 2246.jpg
reading through graphics...
0.06u 0.04s 2.36r 	 resize -bx 500

(a quick test shows that jpg has about the same performance
with p9p and 50% of the ticks are in colormap:1420/62424)

i have been using a script to use resample to convert
pictures as i load them from the camera.  to images the resample
on a core i7 machine can just keep up with the new usb.
(about 4mb/s.  i believe this is limited by the spi-based
media.)

the resampled images can be displaed at >50/s.  it's
amazing how fast plan 9 video is when you don't read
back from the frame buffer.  :-)

- erik