Website (google) picture query

vvvv beta question
#1

I want to do a (basically) simple thing.

A) connect to goole (or any webpage)
B) query site
C) fetch images and
D) show them on quads (or whatever) as a dx-texture.

a+b+c is no prob (in the html-renderer). but the step from the html-renderer to a texture seems to be impossible or am I overseeing something very obvious?

Project attached
as always uploaded the version with a bug! skip this file and take the next!

fetchimagesfromURL.v4p (8.7 kB)

Filtering images from websites
#2

This is the real file:

fetchimagesfromURL.v4p (8.7 kB)

#3

Use the node dynamictexture (string) it’s perfect for your patch .

#4

so far so good.

I found “dynamic texture” and got this problem done. Now I am asking if the http-module is spreadable… hm…

fetchimagesfromURLV0.1.v4p (11.0 kB)

#5

I don t think that the module is spreadable , but you don t need to load all in the same time you can load one page and record the url , and load a other page and record …

#6

sanch:
I don t think that the module is spreadable , but you don t need to load all in the same time you can load one page and record the url , and load a other page and record …

You mean making a “container” of the images? Hmm, I have then the images. Do I need a new Quad for every texture? I remember I read somewhere in the forum that textures are not spreadable…

I want this application as dynamic as possible…

#7

You mean making a “container” of the images? Hmm, I have then the images. Do I >need a new Quad for every texture? I remember I read somewhere in the forum >that textures are not spreadable…

Textures are spreadable as well as quads. just count your textures (or filenames, urls) with the count nodes. with that information you can spread your quads (for example use linearspread and a transform node.) then connect your spread of textures to the spreaded quad. et voila…

#8

have a look at Queue (EX9.Texture), load an image from a site, then put it in the queue, load the next texture, put it in the queue, …

#9

alright. I finished the first version. I am sure that there a still some optimisations possible, but afterall it works. Feel free to enhance!

020200

fetchimagesfromGoogleImagesV0.1.v4p (18.4 kB)

#10

cool , works fine , the only things that i see now , is the problem with the gif: v4 can t read gif , maybe in the regexpression node you can just delete it?

#11

hey,

v4 can t read gif , maybe in the regexpression node you can just delete it?

you can use google´s options to filter jpegs (this worked for me):

http://images.google.de/images?as_q=
your query
&svnum=10&hl=de&client=firefox-a&btnG=Google-Suche
&as_epq=&as_oq=&as_eq=&imgsz=&as_filetype=jpg&imgc=as_sitesearch=

#12

hi,

since i need to build up a set of images found @ google img i gonna use 020200´s v4p for that and i added the &as_filetype tag. still there are many black images.

is there a way to check the texture for “i am not showing anything than black” (like the rgb values for all pixels compared to a “0” “0” “0” “1” ?!)

i´m going to build up a image crawl + visual effect alternation tool (big plans… hehe)

tryin´ on…

greetinX, bbb

#13

Pipet should help you analysing the texture. it might be a little slow to check all pixels, but if something like 10 random pixels are completely black the image is likely to be unsuited or fully black

#14

I can’t seem to get the “fetchimagesfromgoogleimages.v4p” to work properly. The HTTP node is giving me the status error “Socket Error # 11001 Host not found.”

I’ve checked the HTTP help file and I do not get socket errors .

What can I change in the googlefetch patch to make this work?

#15

same problem here

the http node which (i guess) should receive the image url just gets
{CODE(ln=>1)}"+(b.b.indexOf…^
but the regexp should be correct

i also tried to change the url before and behind my query to the structure given from my browser, but was not successfull.

some ideas what’s wrong? thanks :)
phil

#16

ah, ok, google seems to have changed their code.

just write the following in the regexp:
{CODE(ln=>1)},"http:// (.*?) .jpg^

and add with the Add(String) note the “http://” and the “.jpg” to the string again.

#17

just change to the following and you dont need the + (String):

,"(http://.*?.jpg)

#18

i’m triing this patch and using ,"(http://.*?.jpg) the output slices are one correct and one wrong,one correct and one wrong etc.
Example of bad slice:
http://t2.gstatic.com/images","1",,"",1,"",,"http://t2.gstatic.com/images?q\x3dtbn:ANd9GcQLWHQ2ORyffRczR6X7m9NeXGoE5pDQZX7HoG3VXvHtiv35BTw_viGjGvyu","","","","","","","",""],["/imgres?imgurl\x3dhttp://www.ibts.eu/images/pr/book-gift-shop1.jpg

#19

You could give Vlc plugin a try, it seems to support gif.