Creating a knitting pattern of an image

Hallo,

first I want to explain my question. I have an digital image, which I want a pattern of it by creating lines in a specific rectangular.

The dimension of the image and the regtangular is 3:4.

The boundary of each side is divided by several points. So the endpoints of the lines which should create the pattern shell cross these points.

In the attachment there is an example of what I want to create.

It would be great if anybody can help me

THX

Attachments:

all15small.jpg, 111 KB

Replies to This Discussion

Permalink Reply by Laurent DELRIEU on June 5, 2017 at 1:33pm

Surely not an easy task. Seems to be very interesting but at the moment you better ask the artist !

Always good to have references of the work you want to copy

Work from Petros Vrellis

http://artof01.com/vrellis/works/knit.html

It is interesting to know that it requires 2 billion of calculations !

Permalink Reply by Tolga on June 5, 2017 at 1:52pm

Hi Laurent,

thank you for your answer but Petros uses openframework as tool, and it's much more complicated then gh.

I hope someone has a solution for this in gh.

Ciao

Permalink Reply by Joseph Oster on June 5, 2017 at 3:40pm

Knitting is done by hand, with step-by-step instructions dictated by a computer.

Until I watched the video and read that page, I didn't realize that the gray tones and "fade to black" effect around the face is all achieved by threads. 3000 to 4000 lines, by hand! Wow. Gives me an idea for a different approach though... :)

Permalink Reply by Petras Vestartas on June 11, 2017 at 3:59am

Was this originally done as drawing by hand, without following any pre-computed pattern?

Permalink Reply by Joseph Oster on June 5, 2017 at 2:30pm

I thought of connecting all the edge points to each other, filtering out the lines more than a certain distance from any of the defined points. Line density is defined by the 'Count' slider at top-left (the number of points along each edge). This model defines four points: left and right eyes, nose and mouth:

I used the following image, cropped from the painting Girl with a Pearl Earring by Johannes Vermeer:

http://www.uisio.com/wp-content/uploads/2015/01/22pearl-2-1000x520.jpg

http://www.uisio.com/golden-age-of-the-dutch-painting/

Cropped and brought into Rhino using PictureFrame command, then referenced in GH as 'Geo' param:

Attachments:

knitting_2017Jun5a.gh, 15 KB

Permalink Reply by Laurent DELRIEU on June 5, 2017 at 3:48pm

I am there now. it is quite long now, I don't know well Bitmap optimization in C#. It was more simple than I thought. I will optimize it. Less than 100 lines of C# !

Permalink Reply by dvdrbls on June 5, 2017 at 6:11pm

wow! it looks awesome!

Permalink Reply by Michael Pryor on June 5, 2017 at 6:40pm

nice

Permalink Reply by Laurent DELRIEU on June 6, 2017 at 3:42pm

I am here now, 20 minutes calculations. It begins to look at something !

The process is super simple. Draw a white line on an image. At the end of the line test all lines from the end to another point on the border of the circle (it could be a rectangle). Take the line that brighten the more the image. Repeat (here 400). See left image what is becoming the right image.

Right Image from Danny Santos

http://blog.grainedephotographe.com/shooting-portraits-of-strangers...

Permalink Reply by Joseph Oster on June 6, 2017 at 4:30pm

Looking very good! I don't quite follow your description of the algorithm though?

Clearly, the base image must show ONLY the face, nothing distracting around it. The hair does that in this image you chose, but the one I used (for example) would work better if all but the face were masked to black.

Permalink Reply by Tolga on June 7, 2017 at 6:08am

Thanks Joseph, that's awesome

Permalink Reply by Tolga on June 7, 2017 at 6:10am

I have another question.

How is it possible to define the grey tones of the image if the image is black-white?