unforum

[Arenhart]

[Xkeeper]

So much work...

[AUZ505]

[Xkeeper]

[Arenhart]

So, it leads to an url, with the following text backwards:

[RAD24]

[sequoya]

Awesome work, everyone! And Xkeeper, your explanation page is really helpful So glad this was solved, I was so out of my depth on that one. I almost don't want to ask, but... can we apply a similar mechanism to the other picture, the one of Kekkonen? Perhaps on a different sort, though?

[AUZ505]

@xkeeper

your decoding should also work for the nekkonke image (at least after chnaging the color channels). Could you give it a try?

[Arenhart]

I used xkeeper's solution on the nekkonke, and it worked (though it was based on the blue channel, are we to expect a third green message?) the provides a qr code that leads to an url, the url has an inverted text that, once reversed yelds apparently no clue... I post it below.

[Arenhart]

So what leads are left to follow?

I am currently working on that face modelling thing. So far it seems more like an algorithm problem than a riddle. Anyway, this is what I got:

Uphill techniques is a least squares algorithm, wich means it tries to get the lower possible sums of the square of difference between the data (original image) and the model(be the ellipses, or a copy of the image with simplified colours). This difference can be any numerical property of the data.

The closest application of an uphill method in this context that I found is, as was mentioned by somenone before, in image segregation. What I found were methods to separete a full colour image in a limited group of regions that have similar colours, in a way that the parameters of each group are defined so that the collors grouped together are as similar as possible.

Now, this particular algorith doesen't seem very useful for this, first becouse it segregates the regions based solely on theyr colour, and not on position, so, for exemple, both eyes would be put in the same "white" group, but this wouldn't be useful when assignin the ellipses, as it would take the mass center of the eyes and turn the man into a cyclops. And second becouse this colour segregation could be trivially achieved by simply applying a selective gaussian filter in the image, and then manually limiting the levels of gray.

The real problem is assigning the ellipses, which could benefit from an uphill algorithm, but I believe it needs a good initial set of ellipses before it begins optimizing... So after limiting the number of coulours, the algorithm could identify isolated groups of colours and from the isolated groups of same colour assign the initial ellipses.

Now, this is quite a lot of work, so, does anybody has a more practical idea?

[Jim Sawyer]

Hey, just caught up on this, looks interesting. Just out of curiosity, has anyone gone and checked out the book from the Library? I know it was suggested that there may be another playing card there but what if there is actually a clue in the suggested page. The book is about computer organization so maybe there's something in there about the facial recognition algorithms or something...

Side note: I thought Richard Exaltego might be another anagram and I picked out the words Rich Alter Ego but that leaves the letters 'adx' which doesn't really make a lot of sense.

[sequoya]

[Arenhart]

[sequoya]

[amandel]