Send Gimp-user mailing list submissions to
gimp-user@lists.xcf.berkeley.edu
To subscribe or unsubscribe via the World Wide Web, visit
http://lists.xcf.berkeley.edu/mailman/listinfo/gimp-user
or, via email, send a message with subject or body 'help' to
gimp-user-request@lists.xcf.berkeley.edu
You can reach the person managing the list at
gimp-user-owner@lists.xcf.berkeley.edu
When replying, please edit your Subject line so it is more specific
than "Re: Contents of Gimp-user digest..."
Today's Topics:
1. Re: Subtracting from image (Joao S. O. Bueno Calligaris)
2. Re: Subtracting from image (Axel Wernicke)
3. Re: Subtracting from image (Rikard Johnels)
4. No JPG recognized on one host: 2.2.8 (Rich Shepard)
----------------------------------------------------------------------
Date: Wed, 24 Aug 2005 16:12:39 -0300
From: "Joao S. O. Bueno Calligaris"
To: gimp-user@lists.xcf.berkeley.edu
Subject: Re: [Gimp-user] Subtracting from image
Message-ID:
In-Reply-To:
References:
Content-Type: text/plain;
charset="iso-8859-1"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Precedence: list
Message: 1
On Wednesday 24 August 2005 14:54, Rikard Johnels wrote:
Hi all!
I have a "problem".
I have two images taken with a few seconds difference, and i want
to subtract the differences between them from one of them.
Its a crowded square, with people walking around, And i want to
"keep" all the stationary object, as benches etc, and remove the
things that have moved. Its a experiment to "enhance" pictures
taken for a panorama of populated areas.
Any pointers on where to look for a solution?
Well...for strcit "subtraction of an image" - the GIMP provides layer
modes taht do that.
Just open both your images as different layers on teh same gimp image,
and change the one on the layer above to "subtract" or "difference"
- that will do a subtraction on the RGB pixel value of each pixel.
That is not your intended final result. But from the "difference"
layer mode you should get everything that is fixed in place (ground +
benchs + billboards) in black or near black (with outlines since it
is little likely that the images will match exactly at pixel level),
with negative or otherwise color distorted sillouetes of the moving
objects (e.g. persons + dogs + vehicles) . You can then copy this
"silloueted" view with "copy visible" and paste it as another layer.
Then yoiu can try different ways of getting this "sillouete layer" to
the layer mask of the top image, to trying to let only the parts
where people appear transparent. You will need to work a lot to find
good combinations of operations to do on the layer masks (you will
have to blur + play with curves or levels on them)
I think what you desire would be easier with three instead of two
photos from the same place - or your sillouete layer will contain
data of people on both photos. With three photos you have to further
generate a second sillouete layer, and then combine both sillouetes
to generate a suitable mask for one of your layers.
It will requiere quite some experimenting - and I know this I am
writing must be sounding confuse. If you have three photos of the
same place, mail me (it can be reduced resolution version) of then,
and I might try to document all the steps in detail.
If you just have two images, send then anyway ... but it may be
harder.
Regards,
JS
->
/Rikard
