What kind of render times do you reckon would be required for convergence in these skin test Cool?
CoolColJ's test pics thread
Exit portals do help a lot in some scenes though, mainly a room with a tiny window or a dark room with a door slightly open with light coming from the other side
but yeah on the whole, if you don't really need it then, don't slap one on just for the sake of it!
Samplerate is effectively cut to 60% as well
As far as skin goes, I'll let you know, when I finish this render. I decided to create a final larger version of this scene.
I would say at least 3000 samples as needed to get the skin smooth. There are these little pink dots on the skin even at the 2200 sample level
but yeah on the whole, if you don't really need it then, don't slap one on just for the sake of it!
Samplerate is effectively cut to 60% as well
As far as skin goes, I'll let you know, when I finish this render. I decided to create a final larger version of this scene.
I would say at least 3000 samples as needed to get the skin smooth. There are these little pink dots on the skin even at the 2200 sample level
Nice v0.9 test 8 is showing some sign of life in the water caustics front! 
not even 200 samples yet and look at all this stuff
With default settings of MNCR 1000. And they develop much faster and more densely. Although not all of my water scenes show it, still trying to figure out why. Perhaps having textures+bump map changes things
as long as you don't try and use phong material that is! Phong is buggy around specular material
not even 200 samples yet and look at all this stuff
With default settings of MNCR 1000. And they develop much faster and more densely. Although not all of my water scenes show it, still trying to figure out why. Perhaps having textures+bump map changes things
as long as you don't try and use phong material that is! Phong is buggy around specular material
- Attachments
-
- im1191826663.JPG (96.14 KiB) Viewed 4024 times
Well the sun definitely goes through the water now, in some scenes very quickly, in some others not so quickly.
MNCR 1000, works well in some, but MNCR 50,000 works even better in some scenes!
This is without textures and bump map in the pool material.
Now I figured there was something in textures or bump map that causes the sunlight/caustics to not go through water that well. It turns out I was half right - textures no problems, but bump map really seems to slow down the rendering of the light going through the water!
The larger the bump map the worse it gets. Without bump map in the pool material, the shading of the sunlight in the pool floor is fairly quickly and large dense chunks, but once you add the bump map, it really slows down to small patchy bits, if any
Worse with MNCR 1000
I don't know why bump mapping would have such a large effect on how indigo renders these paths. So bump mapping under water is not recommended, but maybe this also means no bump mapping in the scene at all? I haven't tested this to see if this is the case
I'll post some pics later today
MNCR 1000, works well in some, but MNCR 50,000 works even better in some scenes!
This is without textures and bump map in the pool material.
Now I figured there was something in textures or bump map that causes the sunlight/caustics to not go through water that well. It turns out I was half right - textures no problems, but bump map really seems to slow down the rendering of the light going through the water!
The larger the bump map the worse it gets. Without bump map in the pool material, the shading of the sunlight in the pool floor is fairly quickly and large dense chunks, but once you add the bump map, it really slows down to small patchy bits, if any
Worse with MNCR 1000
I don't know why bump mapping would have such a large effect on how indigo renders these paths. So bump mapping under water is not recommended, but maybe this also means no bump mapping in the scene at all? I haven't tested this to see if this is the case
I'll post some pics later today
no bump at all?
Not before we get MPD or the other thing, where I forgot the name xD
(that Vector thingy, that also copes with tunnels /etc. O.o
[Imagine a HUGE map, that represents the earth's surface {without flora, fauna, water or buildings} + a second one {animated, this one is for water} down to at least chemical structures, if not, even the more accurate physical ones O.o {THAT would be slow to render xD}])
Not before we get MPD or the other thing, where I forgot the name xD
(that Vector thingy, that also copes with tunnels /etc. O.o
[Imagine a HUGE map, that represents the earth's surface {without flora, fauna, water or buildings} + a second one {animated, this one is for water} down to at least chemical structures, if not, even the more accurate physical ones O.o {THAT would be slow to render xD}])
I am finding a max change value of 0.005 working much better for the caustics and general MLT stuff in V0.9 test 8. It clumps the MLT samples closer together so you get a smoother more dense look. Under water stuff converges faster.
With max change of 0.01 (default value) or higher, you get more patchy look and you don't get the caustics showing up in a curvy line in one pass
Only downside is the render has a more muddy look that takes at least 1000 samples to smooth out.
the same render from above using default values of Large mutation probablity 0.4, MNCR 1000 but with max change of 0.005
ignore the shading errors with smoothing on
You can see it looks much better than the render one above.
Although I had water smoothing turned off in the above render.
Still not fully converged, but you get the idea - caustics show in v0.9 test 8!
As long as you don't have bump maps in the pool it seems. I'm doing some tests and will post em up later. Well they sorta show, but just take 4 times longer or something and patchy
Finally looks nice and sunny!
Although when there are textures, the water needs more absorption now to look right.
With max change of 0.01 (default value) or higher, you get more patchy look and you don't get the caustics showing up in a curvy line in one pass
Only downside is the render has a more muddy look that takes at least 1000 samples to smooth out.
the same render from above using default values of Large mutation probablity 0.4, MNCR 1000 but with max change of 0.005
ignore the shading errors with smoothing on
You can see it looks much better than the render one above.
Although I had water smoothing turned off in the above render.
Still not fully converged, but you get the idea - caustics show in v0.9 test 8!
As long as you don't have bump maps in the pool it seems. I'm doing some tests and will post em up later. Well they sorta show, but just take 4 times longer or something and patchy
Finally looks nice and sunny!
Although when there are textures, the water needs more absorption now to look right.
- Attachments
-
- at 430 samples
- im1191888648.JPG (78.56 KiB) Viewed 3903 times
-
- at 160 samples - same as above
- 160samples_2.JPG (94.74 KiB) Viewed 3903 times
no water smoothing... looks a bit strange
nice to have the sun go through water!
The water in this scene rendered really fast compared to the one above.
I don't know why either
Maybe it's because the Sun is facing directly down, ie like in a Noon position. So most of the caustics reflections go straight up into the air
nice to have the sun go through water!
The water in this scene rendered really fast compared to the one above.
I don't know why either
Maybe it's because the Sun is facing directly down, ie like in a Noon position. So most of the caustics reflections go straight up into the air
- Attachments
-
- im1191832589.JPG (99.71 KiB) Viewed 3897 times
this one is up to 2700 samples, but yet it's still not converged yet caustics wise. The sunlight renders under the water really slowly compared to the above renders. I don't why either, as it's much lower poly count wise in the water and more uniform wave wise! 4000 polys vs some of the above where they are at least 66,000...
And I deliberately made the scene/objects smaller so it would render faster...
The wave do stick up more, but who knows...
I used default settings with max change of 0.005, perhaps my usual MNCR of 30000 might do it better...
just textures, all diffuse materials and no bump maps
And I deliberately made the scene/objects smaller so it would render faster...
The wave do stick up more, but who knows...
I used default settings with max change of 0.005, perhaps my usual MNCR of 30000 might do it better...
just textures, all diffuse materials and no bump maps
- Attachments
-
- im1191903224.JPG (52.76 KiB) Viewed 3867 times
YAY!!!!
If you have *any* chance, to import those scenes (especially the water) into blender, you can use the edge-split modifier
- Or in C4D, there sure is an auto smoothing option, isn't there?It would be interesting, if it works fine, for those high poly counts (as it fucked up the normals, in blender...) or if it actually doesn't fuck up the normals, in C4D
they are real refracted caustics, they're just not fully resolved and converged. Lots of paths have been cut short with MNCR 1000 as well so your not seeing the full thing. Jacking MNCR up to 30,000 or higher does the trick
here is a photo of real caustics at a beach - water is about 1m deep
http://www.lysator.liu.se/~kand/caustic ... ustics.jpg
did another pic, put the sun in the noon position which seems to help MLT pick up the sun better. More absorption in the water and jacked MNCR up to 30,000 with large mutation probability of 0.66 and max change of 0.01
2700 samples. It's probbaly only half done, but shows a lot more than the pic above with MNCR 1000
Kram - C4D has smoothing with edge breaks and angle limits, but Cinidgo won't read these, it's just either full smoothing or off.
In any case, some of my water meshes have reversed normals and bad structure, but since the mesh has so many polys it's like looking for a needle in a haystack
I'll make some new water meshes soon.
gonna try some renders in plain MLT mode to see how it copes, and off course some bump map tests....
And jack up MNCR even higher
I had a underwater scene where the camera was inside a glass block/room half over/underwater - that should look interesting now
here is a photo of real caustics at a beach - water is about 1m deep
http://www.lysator.liu.se/~kand/caustic ... ustics.jpg
did another pic, put the sun in the noon position which seems to help MLT pick up the sun better. More absorption in the water and jacked MNCR up to 30,000 with large mutation probability of 0.66 and max change of 0.01
2700 samples. It's probbaly only half done, but shows a lot more than the pic above with MNCR 1000
Kram - C4D has smoothing with edge breaks and angle limits, but Cinidgo won't read these, it's just either full smoothing or off.
In any case, some of my water meshes have reversed normals and bad structure, but since the mesh has so many polys it's like looking for a needle in a haystack
I'll make some new water meshes soon.
gonna try some renders in plain MLT mode to see how it copes, and off course some bump map tests....
And jack up MNCR even higher
I had a underwater scene where the camera was inside a glass block/room half over/underwater - that should look interesting now
- Attachments
-
- im1191968834.JPG (53.44 KiB) Viewed 3736 times
Who is online
Users browsing this forum: Majestic-12 [Bot] and 39 guests