Christian Froeschlin <chr### [at] chrfrde> wrote:
> I think it doesn't make a lot of difference. What I would expect
> to see is just a thin shadow as it would have been cast by the leftmost
> column were the object not impossible (I think this is also what Thomas
> meant). Gimped version attached for clarity.

Ah, I understand now. That gimped version finally caused the coin to drop.
Thanks. And yes, in retrospect, that's probably what Thomas meant as well. I had
been trying to fake a shadow of the horizontal top bar, as to me that seemed to
be the most correct, but apparently it did not work very well. It was only
confusing. So I assume the attached image is more to your liking?

Cheers,
Erwin

Post a reply to this message

Attachments:
Download 'failure-variant3.jpg' (309 KB)

Preview of image 'failure-variant3.jpg'

Eriban wrote:

> So I assume the attached image is more to your liking?

Yes ;)

BTW the shadows are very hard and dark. If the tricks still
work with area light and radiosity you could try these too.

Post a reply to this message

On 21-8-2011 21:35, Eriban wrote:
> Ah, I understand now. That gimped version finally caused the coin to drop.
> Thanks. And yes, in retrospect, that's probably what Thomas meant as well. I had
> been trying to fake a shadow of the horizontal top bar, as to me that seemed to
> be the most correct, but apparently it did not work very well. It was only
> confusing. So I assume the attached image is more to your liking?
>

Yes, sir! You got it right this time :-)

Shouldn't the shadow of the construction touch also the grass, maybe the 
trees? It is tall enough for that it seems.

Thomas

Post a reply to this message

"Eriban" <pov### [at] spamgourmetcom> wrote:
> Hi all,...
>  Nevertheless, hopefully the resulting image is
> still interesting.
....
> Cheers,
> Erwin

Being a great Escher fan, I love it. One idea did occur to me that I'd like to
share. Many of Escher's illusions worked all the better for the fact that the
images were either monochromatic or rendered in a limited color palette. The
bright blue background, although appealing for such a scenic image generally,
may be wowrking against the illusion. The high-contrast diagonal between the
land and sea cuts right across the bit where the "trick" occurs and tends to
focus the viewer's attention just in the wrong spot.  As the various landscaping
elements used in the land area are quite attractive, why not use them for the
entire background and allow the illusion bit to blend in? Just a thought - great
image either way.

Best Regards,
Mike C.

Post a reply to this message

On 22-8-2011 15:21, Mike the Elder wrote:

> Being a great Escher fan, I love it. One idea did occur to me that I'd like to
> share. Many of Escher's illusions worked all the better for the fact that the
> images were either monochromatic or rendered in a limited color palette. The
> bright blue background, although appealing for such a scenic image generally,
> may be wowrking against the illusion. The high-contrast diagonal between the
> land and sea cuts right across the bit where the "trick" occurs and tends to
> focus the viewer's attention just in the wrong spot.  As the various landscaping
> elements used in the land area are quite attractive, why not use them for the
> entire background and allow the illusion bit to blend in? Just a thought - great
> image either way.

That would be indeed a very Escher-like solution. And with a bit of fog 
or media in the background the construction would stand out like in many 
works of Escher.

Thomas

Post a reply to this message

Christian Froeschlin <chr### [at] chrfrde> wrote:
> Eriban wrote:
>
> > So I assume the attached image is more to your liking?
>
> Yes ;)

Okay, good to hear. :-)

> BTW the shadows are very hard and dark. If the tricks still
> work with area light and radiosity you could try these too.

I think I may be able to get area lights to work. I will give it a try.
Radiosity may be more difficult. Also, the computing power I have at my disposal
is a bit of a limiting factor. Rendering 600x800 with anti-aliasing already
takes more than four hours with the current image (on my six-year old Mac Mini).
And for printing the image on A4 format, which I intend to, I need to go to
2400x3600. The current trial render has been going for more than 14 hours now,
and isn't even halfway. That's without area lights and radiosity...

Post a reply to this message

Thomas de Groot <tenDOTlnDOTretniATtoorgedDOTt> wrote:
> Shouldn't the shadow of the construction touch also the grass, maybe the
> trees? It is tall enough for that it seems.

Yes, it should, really. I had already tried this, but did not get it to look
good. One problem is what the shadow should look like. It could be the top of a
single column (that's according to the shadows on the inner wall). However, what
then about the shadow of the square portal with two pillars? Showing a shadow on
the grass basically draws the attention to the fact that something odd is
happening here. Also, it simply did not look good (probably partly due to the
harsh shadows and lack of radiosity). So for now, I am taking the easy way out
here...

Post a reply to this message

"Mike the Elder" <nomail@nomail> wrote:
> Being a great Escher fan, I love it. One idea did occur to me that I'd like to
> share. Many of Escher's illusions worked all the better for the fact that the
> images were either monochromatic or rendered in a limited color palette. The
> bright blue background, although appealing for such a scenic image generally,
> may be wowrking against the illusion. The high-contrast diagonal between the
> land and sea cuts right across the bit where the "trick" occurs and tends to
> focus the viewer's attention just in the wrong spot.  As the various landscaping
> elements used in the land area are quite attractive, why not use them for the
> entire background and allow the illusion bit to blend in? Just a thought - great
> image either way.

Happy to hear that you like the image. Your suggestion is interesting, and I
like to give it a try. I'll post the result once I have done so (which may be a
while, as I am busy of the next couple of days). It's actually a bit similar to
the plan I first had. My initial idea was to let the grass slowly morph into an
abstract tiling pattern that would slowly fade out. That's also quite Escher
like. I discarded that without actually trying it out, however, because I
thought it would distract from the main illusion. Your suggest would probably
work better in that respect.

Post a reply to this message

On 22-8-2011 22:46, Eriban wrote:
> Thomas de Groot<tenDOTlnDOTretniATtoorgedDOTt>  wrote:
>> Shouldn't the shadow of the construction touch also the grass, maybe the
>> trees? It is tall enough for that it seems.
>
> Yes, it should, really. I had already tried this, but did not get it to look
> good. One problem is what the shadow should look like. It could be the top of a
> single column (that's according to the shadows on the inner wall). However, what
> then about the shadow of the square portal with two pillars? Showing a shadow on
> the grass basically draws the attention to the fact that something odd is
> happening here. Also, it simply did not look good (probably partly due to the
> harsh shadows and lack of radiosity). So for now, I am taking the easy way out
> here...
>
Fair enough. Imo, the shadow should look like the portal.

Thomas

Post a reply to this message

> Christian Froeschlin<chr### [at] chrfrde>  wrote:
>> Eriban wrote:
>>
>>> So I assume the attached image is more to your liking?
>>
>> Yes ;)
>
> Okay, good to hear. :-)
>
>> BTW the shadows are very hard and dark. If the tricks still
>> work with area light and radiosity you could try these too.
>
> I think I may be able to get area lights to work. I will give it a try.
> Radiosity may be more difficult. Also, the computing power I have at my disposal
> is a bit of a limiting factor. Rendering 600x800 with anti-aliasing already
> takes more than four hours with the current image (on my six-year old Mac Mini).
> And for printing the image on A4 format, which I intend to, I need to go to
> 2400x3600. The current trial render has been going for more than 14 hours now,
> and isn't even halfway. That's without area lights and radiosity...
>
>
>
If you try with an area light, don't forget to use adaptive.

Post a reply to this message