WIPs: CTZn

[yonosoy]

More room this time, a total of 116 mill.

[ENSLAVER]

Wow

[OnoSendai]

Awesome stuff yonosoy!

[CTZn]

Impressive, my rig can barely provide the quarter of this definition ! High enough so I have witnessed these artifacts with view-dependent subdivisions already. What were supersampling level and max triangle length in pixels for this render ?

Now this is crying for shading, do you have plans for this, or instancing yonosoy ?

[yonosoy]

<super_sample_factor>2</super_sample_factor>

<max_num_subdivisions>19</max_num_subdivisions>
<subdivide_pixel_threshold>1</subdivide_pixel_threshold>

There are plans to work around this; no experience in ISL yet; but work around.
In the same way there are math to represent flocks of birds and fishes, etc... (too much to be alone in this).

[CTZn]

Thanks, now I'm not sure wether the pixel threshold is against the framebuffer or the final output...

Have you tried the "Snow Cover Rough" material from Fakeshamus ? http://www.indigorenderer.com/materials ... /materials
His other materials are using a blend wich requires to add the canyon displacement to each submaterial, with parameters.

Can Cindigo do particle instances ? That would be the way to flocks.

[yonosoy]

Let me some days to put this in order. For the next try the solar spectrum must be express as real tabulated.
(It depends on your amount of ram at the moment, and by hand), five geosun in five light layers gives as result five photographs for the same time in atmospheric simulation, a way to study in exteriors. Maybe for indigo fits the possibility of more than one extra atmospheric sun. Some positives test whit perfect spheres.

Yes, Cindigo from windows gives here a range of 40-60 millions of instances (?) maybe. Flocks are not ready yet. Basic shapes for vegetation and birds to start, but in a procedural way, taking some advantages from houdini... then clouds of points and channels of movements aplied for each specimen must be collect. In fact there are single animated bird.

It`s an embraceble start point. Thankfully to this little team with more effort than "giants" and for the posibilities to grow up even without license in this open forum.

[Zom-B]

<super_sample_factor>2</super_sample_factor>

<max_num_subdivisions>19</max_num_subdivisions>
<subdivide_pixel_threshold>1</subdivide_pixel_threshold>

a pixel treshold of 1 with SS of 2 allows the subdiv algo to generate polys with minimum size of 1/2 pixel. Rise that value to save a lot of RAM! value of 4 would allow minimum of 2 pixel polys... be sure to enable view dependend subdiv, otherwise the setting gets ignored

[yonosoy]

<super_sample_factor>1</super_sample_factor>
<max_num_subdivisions>22</max_num_subdivisions>
<subdivide_pixel_threshold>2</subdivide_pixel_threshold>

A hundred of mill. tri. for better result and so much RAM saved !!! Thanks a lot.

A tweak over the first rig, second in study...
A global mesh will be evaluated.

EDITED: 145 mill.

[CTZn]

Well the albedo was initially meant to illustrate the point but it became messy further along, it's bicoloured really.

I'm keen to blend the mossy rock material from Fakeshamus in after all.

edit: nice update ! I'm suspecting that the stairing artifact can be reduced by playing with the pixel threshold further, introducing more decimals

<super_sample_factor>2</super_sample_factor>
<max_num_subdivisions>22</max_num_subdivisions>
<subdivide_pixel_threshold>1.324</subdivide_pixel_threshold>

or

<super_sample_factor>1</super_sample_factor>
<max_num_subdivisions>22</max_num_subdivisions>
<subdivide_pixel_threshold>0.6</subdivide_pixel_threshold>

Blind shots, and ignoring the tuning for max_num_subdivisions. Thanks for the clarification Zom-B, I came to doubt about it.

[CTZn]

possibly the displacement artifact is in fact a shading artifact (pretty sure, I seem to remember).

Also this set would require a setting twice as big for the least; 1Km width is too small for this shader scale.

[yonosoy]

Blind shots yes, an adventure. Needs a lot of work... Scale 33 (more than 1000 km. square).

This is the last for first impressions.
Could voxels play a role in synthetic vegetation for this scale?

EDIT:
Subdivision algorhytm do necessary n/2 - 1 (aprox) global subdivisions before start the partial ones? Is not possible to try to displace a global mesh in this way ...
Artifacts more over 23 subdivisions.
And seems to not be the same SS 1 pix. thres. 2 than SS 2 pix. thres. 4, ... (so confuse now).
Nonetheless will try to work around this.

[CTZn]

Could voxels play a role in synthetic vegetation for this scale?

Instances are better than voxels. Even voxel engines use instancing !

In the context of this thread it is possible but tedious because voxels are a Maya feature and ISL for the terrain is Indigo's. Getting the displacement data for correct object positioning back in Maya requires to extract a map by rendering it with indigo (normalised base emission, displacement shader as emission, linear tonemapping, ortho camera).

EDIT:

Indigo does n-1 subdivisions if I'm correct.

You want to input the final pixel threshold multiplied per the number of super-sampling levels.

For instance if you want a triangle size of one pixel and that your ssl is 3 you want to input 3.0 as the pixel threshold. This would come down to 1.0 after the framebuffer's reduction to the final output.

Oh and by the way thanks for sharing back yonosoy. A veces mejor que lo digas en español al mejor porque con la traduccion se pierde hasta el sentido a veces and blame me if spanish is not your native speaking.

[yonosoy]

Hi CTZ'n. ¡¡ Muchas gracias ¡¡

Code: Select all

Subdividing and displacing mesh '|canyons_pPlane1|canyons_pPlaneShape1', (max num subdivisions = 22) ...
	Subdividing '|canyons_pPlane1|canyons_pPlaneShape1', level 0...
		num triangles subdivided: 0
		num triangles unchanged: 0
		num quads subdivided: 4
		num quads unchanged: 0
		resulting num vertices: 25
		resulting num triangles: 0
		resulting num quads: 16
		Done.
	Subdividing '|canyons_pPlane1|canyons_pPlaneShape1', level 1...
		num triangles subdivided: 0
		num triangles unchanged: 0
		num quads subdivided: 16
		num quads unchanged: 0
		resulting num vertices: 81
		resulting num triangles: 0
		resulting num quads: 64
		Done.
	Subdividing '|canyons_pPlane1|canyons_pPlaneShape1', level 2...
		num triangles subdivided: 0
		num triangles unchanged: 0
		num quads subdivided: 64
		num quads unchanged: 0
		resulting num vertices: 289
		resulting num triangles: 0
		resulting num quads: 256
		Done.
	Subdividing '|canyons_pPlane1|canyons_pPlaneShape1', level 3...
		num triangles subdivided: 0
		num triangles unchanged: 0
		num quads subdivided: 256
		num quads unchanged: 0
		resulting num vertices: 1089
		resulting num triangles: 0
		resulting num quads: 1024
		Done.
	Subdividing '|canyons_pPlane1|canyons_pPlaneShape1', level 4...
		num triangles subdivided: 0
		num triangles unchanged: 0
		num quads subdivided: 1024
		num quads unchanged: 0
		resulting num vertices: 4225
		resulting num triangles: 0
		resulting num quads: 4096
		Done.
	Subdividing '|canyons_pPlane1|canyons_pPlaneShape1', level 5...
		num triangles subdivided: 0
		num triangles unchanged: 0
		num quads subdivided: 4096
		num quads unchanged: 0
		resulting num vertices: 16641
		resulting num triangles: 0
		resulting num quads: 16384
		Done.
	Subdividing '|canyons_pPlane1|canyons_pPlaneShape1', level 6...
		num triangles subdivided: 0
		num triangles unchanged: 0
		num quads subdivided: 16384
		num quads unchanged: 0
		resulting num vertices: 66049
		resulting num triangles: 0
		resulting num quads: 65536
		Done.
	Subdividing '|canyons_pPlane1|canyons_pPlaneShape1', level 7...
		num triangles subdivided: 0
		num triangles unchanged: 0
		num quads subdivided: 65536
		num quads unchanged: 0
		resulting num vertices: 263169
		resulting num triangles: 0
		resulting num quads: 262144
		Done.
	Subdividing '|canyons_pPlane1|canyons_pPlaneShape1', level 8...
		num triangles subdivided: 0
		num triangles unchanged: 0
		num quads subdivided: 262144
		num quads unchanged: 0
		resulting num vertices: 1050625
		resulting num triangles: 0
		resulting num quads: 1048576
		Done.
	Subdividing '|canyons_pPlane1|canyons_pPlaneShape1', level 9...
		num triangles subdivided: 0
		num triangles unchanged: 0
		num quads subdivided: 1048576
		num quads unchanged: 0
		resulting num vertices: 4198401
		resulting num triangles: 0
		resulting num quads: 4194304
		Done.
	Subdividing '|canyons_pPlane1|canyons_pPlaneShape1', level 10...
		num triangles subdivided: 0
		num triangles unchanged: 0
		num quads subdivided: 4194304
		num quads unchanged: 0
		resulting num vertices: 16785409
		resulting num triangles: 0
		resulting num quads: 16777216
		Done.
	Subdividing '|canyons_pPlane1|canyons_pPlaneShape1', level 11...
		num triangles subdivided: 0
		num triangles unchanged: 0
		num quads subdivided: 85849
		num quads unchanged: 16691367
		resulting num vertices: 17088651
		resulting num triangles: 0
		resulting num quads: 17034763
		Done.

Como puedes ver 22 subdivisiones , como empieza en 0, terminará en 21. Lo que significa que el algoritmo ha realizado 11 subdivisiones globales (muy rápidas) y entonces comienza la subdivisión parcial (lenta).
Estamos intentando subdividir una malla global, en la que un gran espacio mucho mas teselado está definido. Para ésta situación (entendemos que anodina) no más de 4 subdivisiones pueden ser realizadas. Si pudiera controlarse el número de ambas podria intentarse la definición compleja de una extensísima zona (escojeríamos una única subdivion global, tal vez).
Enamorados de la atmósfera, la capacidad ISL ,(se ha tenido en cuenta el desplazamiento de las instancias !!),... Soñando por el momento con la capacidad de proceder através de la matemática para representer la naturaleza.

... thanks for sharing back...

Es intención de base.

Revisarémos otra vez el SSL.
Thanks !

[CTZn]

Yay at artifacts. No wonder, the canyon shader is one twisted simple hack really.

Estamos intentando subdividir una malla global, en la que un gran espacio mucho mas teselado está definido. Para ésta situación (entendemos que anodina) no más de 4 subdivisiones pueden ser realizadas.

Could you describe this setup a bit further please ? Does the mesh come with preliminary subdivisions or is pretty high res ... wich scale and resolution is this mesh ?

Si pudiera controlarse el número de ambas podria intentarse la definición compleja de una extensísima zona (escojeríamos una única subdivion global, tal vez).

I'm really not getting the idea, my bad. Ambas de que para controlar ? No tengo claro el problema practico que enfrenteis, me falta la base (de castillano no, espero!). Did you actually request a new feature ?

Como puedes ver 22 subdivisiones , como empieza en 0, terminará en 21. Lo que significa que el algoritmo ha realizado 11 subdivisiones globales (muy rápidas) y entonces comienza la subdivisión parcial (lenta).

Esto claro.

Es intención de base.

Pues gracias. Estos dias (sino semanas) no voy a ser muy disponible, una lastima ):

You love the atmosphere model and ISL, good on you Search for implicit surface in the forum.