 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Am 12.05.2011 22:25, schrieb TawnyOwl:
> Second thought: 4 points are one to much. Better use only 3 in a equiliteral
> traingle at the borders of the drops. Then the plane between this points on the
> blade is well defined (4 points may not result in a plane) and the slopes can be
> calculated.
For determining the "slope" of the blade, retrieving the normal vector
from a single trace() call should suffice.
Tracing multiple rays is a good idea though to prevent drops from
hanging out too far at a blade edge.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Great texture on the grass, if it wasn't posted on this forum I would
> have just assumed it was a photo but at closer inspection I think
> some of the rain drops are a bit too spherical for their position on
> the grass.
>
Yes, that's another problem apart from the placement on the blade
borders... I'm already using a blob with two components, the one below
having negative strength to make it look less spherical, but it isn't
doing a very good job (I admit that I didn't put much effort into it).
--
Jaime Vives Piqueres
La Persistencia de la Ignorancia
http://www.ignorancia.org
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> Great texture on the grass, if it wasn't posted on this forum I
>> would have just assumed it was a photo but at closer inspection I
>> think some of the rain drops are a bit too spherical for their
>> position on the grass.
>>
>> Sean
>>
>>
>
> There is also some visible fasceting on the blades.
Yeah... I had a very obvious bug on the loop which creates the blades,
and the point from one section to the the next weren't properly matched
(but somehow, I never traced it before at enough resolution to notice
it). It's corrected now for the final render... thanks!
--
Jaime Vives Piqueres
La Persistencia de la Ignorancia
http://www.ignorancia.org
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Second thought: 4 points are one to much. Better use only 3 in a
> equiliteral traingle at the borders of the drops. Then the plane
> between this points on the blade is well defined (4 points may not
> result in a plane) and the slopes can be calculated.
I was going to go for the 4 points approach, until you made this
useful suggestion... thanks! For the slope, as clipcka said, the normal
returned by the first trace() is enough to reorient the drop (and I was
already using it, even if it's not very noticeable because the drops are
too spherical).
--
Jaime Vives Piqueres
La Persistencia de la Ignorancia
http://www.ignorancia.org
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Maybe not *all* grass, but generally the blades are a tiny bit folded
> around a central - longitudinal - nerve. This helps also the blade to
> keep upright. Water droplets then often concentrate in the fold.
The folding is universal, to my knowledge... just some kinds of grass
don't show it very well. In fact, it's something I noticed from day one,
but was too lazy to try it. I just looked now, and it was in fact very
easy to do, due to the way I'm constructing each blade. Thanks!
--
Jaime Vives Piqueres
La Persistencia de la Ignorancia
http://www.ignorancia.org
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Jaime Vives Piqueres <jai### [at] ignorancia org> wrote:
> > Second thought: 4 points are one to much. Better use only 3 in a
> > equiliteral traingle at the borders of the drops. Then the plane
> > between this points on the blade is well defined (4 points may not
> > result in a plane) and the slopes can be calculated.
>
> I was going to go for the 4 points approach, until you made this
> useful suggestion... thanks! For the slope, as clipcka said, the normal
> returned by the first trace() is enough to reorient the drop (and I was
> already using it, even if it's not very noticeable because the drops are
> too spherical).
>
>
> --
> Jaime Vives Piqueres
>
> La Persistencia de la Ignorancia
> http://www.ignorancia.org
Hi Jaime,
thanks for your response. The "more-then-2-points"-approach was not really meant
to determine the slope, but to ensure that the drop is completely on the blades,
the second issue I meant. You can determine the slopes from three vectors, but
of course you need only one. I was to entangled with a similar problem in
placing a mirror orthogonal to the viewpoint of the camera, where I used a
basis-transformation of two vectors with Shear_Trans() from the
transform-library to accomplish the correct rotation. The result is a WIP, which
is stalled for the moment, but may be you recognize the real author of this
scene...
Thank you for this! The trees are from arbaro.
Regards,
Michael
Post a reply to this message
Attachments:
Download 'tierra_first.png' (539 KB)
Preview of image 'tierra_first.png'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> thanks for your response. The "more-then-2-points"-approach was not
> really meant to determine the slope, but to ensure that the drop is
> completely on the blades,
I finally had to use four... just 3 was not working well for some
reason. At the end it didn't come out as nice as I was expecting, but
it's at least half-convincing at a first glance (see final post on this
thread).
> I was to entangled with a similar problem in placing a mirror
> orthogonal to the viewpoint of the camera, where I used a
> basis-transformation of two vectors with Shear_Trans() from the
> transform-library to accomplish the correct rotation. The result is a
> WIP, which is stalled for the moment, but may be you recognize the
> real author of this scene...
Yes, that landscape sounds familiar... ;) BTW, I've not the slightest
idea of what you're doing there...
--
Jaime Vives Piqueres
La Persistencia de la Ignorancia
http://www.ignorancia.org
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> The drops are blobs traced() from above, but they don't take yet
> into account how wide it's the grass blade they rest on... that
> should be the next and final step.
Finally used a 4-point trace() construct, testing if they rest likely
on the same blade by allowing just a little difference in height between
them.
Also added the folding on the blades as Thomas suggested, and
increased the drop count to 30000... they don't look that great, but at
least the "bad spots" are not very distracting.
Here is attached the final pic (and I meant FINAL, I promise),
post-processed to add reflections which were rendered as a separate
layer. Total render time: 2h15m on a Phenom X4.
Regards,
--
Jaime Vives Piqueres
La Persistencia de la Ignorancia
http://www.ignorancia.org
Post a reply to this message
Attachments:
Download 'grass-06pp.jpg' (323 KB)
Preview of image 'grass-06pp.jpg'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 18/05/2011 7:18 PM, Jaime Vives Piqueres wrote:
> Here is attached the final pic (and I meant FINAL, I promise),
> post-processed to add reflections which were rendered as a separate
> layer. Total render time: 2h15m on a Phenom X4.
Wow!. Beautiful work.
zutroi
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Well, well, well... what a nice photograph ;-)
I think this is very convincing.
Thomas
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
