PDA

View Full Version : Hypervoxels VS New Volumetrics



prometheus
01-01-2018, 09:31 AM
Hi folks..

Just installed Lightwave 2018 some hour ago, and the very first thing I promised to check was the new volumetric system, so here I will put up some initial impressions on it..maybe images later, I only got a few hours left today, then back to bed before a full work weekend.

Since Ivé been a fan of the Hypervoxels system, and used it quite a bit..I thought I should share some findings between them two, though what I initially say may be subject for change during my exploration, where things are not as I expected them to be in hypervoxels etc, and things will work quite differently, some things removed etc.

Initially I have both bad impressions as well as good ones, and I got my first Lightwave crash after an hour using only the volumetrics and increasing particles.
I am not that overall joyfull of how they designed how to work with the volumetrics, I really loved the way you could keep track of things and set things up in the hypervoxels window, but there you go..things change, and it now is more of drop down lists and more of nodes actually required to get a hypertexture going, so for me at this time I feel it to be a non inituive way of working, perhaps slower...but as I said, it can be a part of how I am used to it as well.

PARTICLE SCALE AND BLENDING...
Random scale for hypervoxel size was previously an option in a slider to scale right under particle size in the hypervoxel system, in the new system it has no such option...so after lurking, I tried nodes and used particle Id and random scale plugged in to radius, as you can see..such process is much harder for a newbie..or anyone else to actually get started with, a simple random scale value I think they should try and implement, and able to override if you want nodes.
I also found that you may actually use an actual particle emitters particle size and random size instead, that will also yield random scale for the particles, random scale is often essential for more natural looking clouds for instance.

Regarding blending between two or more volumetric items, as well as particle blending, the blending quality now is finally just as good as old dynamite I think, in previous hypervoxels it was quite poor in comparison, that that is a thumbs up.

QUALITY..
The quality of the density edges/smoothness is now much better, where you previously had to set some gradients in local density to smooth out edges...so that is a thumbs up.
I do have to figure out final quality and render settings though, since it seems it can be noisy if not set right.

SPEED...
We can now clone volumetric items/nulls for instance and overlap them without it halting the system, which was the case of what happened when using two nulls with hypervoxels..so this is a good improvement on how you can design clouds with several volumetric items.

I am a bit confused about the speed, it seems to iterate a bit too slow in vpr with the finer smoothing of the volumetric system..when it tries to refine away the noise, if it were to be 3 -4 times faster, I would have been satisfied..as it is now, it can actually seem to be slower than hypervoxels..depending on what mode you have set hypervoxels at, then again it may be unfair to compare it fully against eachother, since the new system is supposed to be more realistic in itīs properties.

OPEN GL..
I have complained about that we can not see hypervoxels in opengl except a sort of wireframe representation, it now has an option to show it more volume like, but it seems only quite flat, not as good as with dynamite opengl for instance...as I can tell so far anyway.

So many things I have complained about, seem to have improved and taken care of, but it has also changed drasticly how you set up and work with volumetrics in such design that I am not that joyful about how that turned out.

SPRITES...
I havenīt seen where to get to that.

The extremely simple particle scene I set up with an emitter and volumetric..it crashed, and when trying to load that very scenefile, it crashes constantly.

prometheus
01-01-2018, 10:01 AM
To note, you can add the old hypervoxels system to lightwave 2018, under effects tab..legacy volumetrics, you can add hypervoxels or pixie dust, you then have to go to the volumetrics tab and also check the use lecacy volumetrics.
But you can not use them both togeter, but at least it is a backup if you need it to perform with old volumetrics, and in case you need sprites the old way..or hypervoxel style of volumetrics, that said..the new volumetrics end result of blending, and using multiple items, cubic volumetric item..will surely expand on some things, though it seem to be a bit harder to set up and not as fast and easy going to work with as withing the hypervoxels interface...at least for me, for some others it may be the opposity and I would have to await some input from others.

prometheus
01-01-2018, 11:21 AM
The Sunsky plugin from dpont sort of works half way, you can add the sunsky environment, but it seems the sunsky light in the light properties will not change, when you choose it from the list, it still remains as a distant light..thus the color and intensity that affects the volumetric light will not adapt to such changes unfortunately, they are probably not compatible yet.


http://forums.newtek.com/attachment.php?attachmentid=139018&d=1514830810

139018

prometheus
01-01-2018, 11:40 AM
Also rman gardner clouds and all those textures, not working for me here wich is a bit of a bummer, hope dpont may look in to that.

erikals
01-01-2018, 11:41 AM
unsure if DPont will keep developing plugins for LightWave, seem to recall i read that he was busy elsewhere. (?)

good notes on HyperVoxels.

regarding longer rendertime, the quality is much higher, and also includes more possibilities.
LightWave 2018 is basically Made to render Blade Runner exteriors   :)

https://www.youtube.com/watch?v=Cr1PC0iI3IE


> except Octane perhaps might be the render engine.

prometheus
01-01-2018, 12:02 PM
unsure if DPont will keep developing plugins for LightWave, seem to recall i read that he was busy elsewhere. (?)

good notes on HyperVoxels.

regarding longer rendertime, the quality is much higher, and also includes more possibilities.
LightWave 2018 is basically Made to render Blade Runner exteriors   :)

https://www.youtube.com/watch?v=Cr1PC0iI3IE


> except Octane perhaps might be the render engine.

That would be sad for procedural texturing, a workaround for getting gardner clouds in though, that would be to install the standard rman textures, and in the node editor add a color layer instead, and from there you can choose the rman textures such as rman clouds and weather, and feed it in to the texture input of the new volumetric item.
As seen in the image...but the node version doesnīt seem to work though it is there.
For sunsky and the Sk_sun light that doesnīt work .letīs hope Anti can look in to a new model for sky..heard something about that, and make sure we get a decent sky model that works with the newer volumetrics more seamless

Good thing with the new volumetrics, we do not get any round spherical volume falloff no more, this stuff seem to blend better, here it is particles..but it seems to be the same on null items as well, and the softness of the volumetrics is much better.
I still have to adjust how to get lighting and shading properly and how to tweak them best, the asymmetry value is very important and seem to like only small values for good detail results.


http://forums.newtek.com/attachment.php?attachmentid=139020&d=1514833266
139020

MichaelT
01-01-2018, 12:23 PM
I haven't gone very far when it comes to the clouds etc.. I really only had one look at them so far:


http://www.planetrift.com/files/comp1.mp4 (VDB volumetrics)

Personally I haven't had a single crash, other than when I tried to use Octane. Other than that, things are amazingly stable for me at least.
Also, text is improved in this version.

tyrot
01-01-2018, 12:26 PM
how do you use this system with particles ? New volumetrics i mean ..

prometheus
01-01-2018, 12:54 PM
I haven't gone very far when it comes to the clouds etc.. I really only had one look at them so far:


http://www.planetrift.com/files/comp1.mp4 (VDB volumetrics)

Personally I haven't had a single crash, other than when I tried to use Octane. Other than that, things are amazingly stable for me at least.
Also, text is improved in this version.

Huh..thatīs not clouds, thatīs fire..just so you know:D

One crash only so far, still only gotten to try volumetrics ..especially for cloud stuff, a new learning curve with getting the light and volume shading right and looking good etc.

prometheus
01-01-2018, 12:57 PM
how do you use this system with particles ? New volumetrics i mean ..

add your particle emitter as usual, hit "p" for properties, by default the primitive tab should be highlighted, it is here you choose primitive type..by default it will be mesh, change it to volumetric and you will then have turned your particles to volumetrics.

or if you start with a null as volumetric item primitive, adding particles in the property fx tab for the null, then the particles will override the null, and it will be those that are active as volumetric items.

tyrot
01-01-2018, 01:01 PM
thanks mate ... for the info :)

MichaelT
01-01-2018, 01:09 PM
Yeah I know its fire... but in any case, I made a proper cloud too just now:

http://forums.newtek.com/attachment.php?attachmentid=139025&d=1530485
139025

I'm happy with how easy it is though.

prometheus
01-01-2018, 01:11 PM
for improving the quality of the volumetrics, you will actually find that in the lights tab and itīs volumetric sample, it will remove most of the grain if set to 5 or higher, at the cost of longer render times, also to note, the volumetric intensity in the light tab, can
boost the looks quite a bit.
step size in the volumetric item tab is also necessary to keep lower rather than higher for good quality.

- - - Updated - - -


Yeah I know its fire... but in any case, I made a proper cloud too just now:

139025

yep..thatīs good, I do wish however that they introduce a sky model ASAP, or clouds will mostly not react properly to environment, or should I say blend, clouds are only half way of a good sky with clouds.

MichaelT
01-01-2018, 01:14 PM
Hmm.. not so sure about that. It seems to me they they opened up enough things to make motions around mountains etc.. more believable.

prometheus
01-01-2018, 01:37 PM
Hmm.. not so sure about that. It seems to me they they opened up enough things to make motions around mountains etc.. more believable.

Im not taking about mountains, but where clouds usually lives, in the air, it is essential for doing realistic renderings ..if you do not intend to cover the whole sky with just full density clouds.

as far as motions around mountains, not sure what you mean.

I can say this though, nebula stuff or whispy clouds will more realistic, so they raised that a bit..since it can have so nice smooth density and softness...will not be able to post images today though.
cirrus clouds or other feather cloud stuff can be made nicely, though you may need to feed the procedural texture in to texture density as well.

Ogo taiki will still be the one that could produce the best overall skies..Even with this new volumetric..(except that for hero clouds this new stuff may be better), with proper spectral lighting etc fog etc, though it was too slow to render and to complex regarding quality settings, and setting things up.

MichaelT
01-01-2018, 02:05 PM
Motion over landscape is a big deal. It very much controls how clouds behave.. you if anyone should know this :) Especially since how clouds rolls over, and around mountains is quite spectacular. But that is just my opinion of course.

prometheus
01-01-2018, 02:11 PM
Gardner clouds possible through color layers only it seems to ..so it rules out feeding other procedurals in to scaling, frequencies, position etc..if you want gardner clouds it seems, as I can see it now.
sample image, color of light has envelope and sunsky color to better represent a sun decay when the sun goes down etc, it is however lacking the features of Sk_sun in regards of intensity as well I think.


http://forums.newtek.com/attachment.php?attachmentid=139029&d=1514841066


139029

prometheus
01-01-2018, 02:15 PM
Motion over landscape is a big deal. It very much controls how clouds behave.. you if anyone should know this :) Especially since how clouds rolls over, and around mountains is quite spectacular. But that is just my opinion of course.

no itīs not your opinion, itīs most valid for many.
And no..you really donīt have to tell me that motions over landscapes is a big deal, it is if you need it..and I know this..but that wasnīt the point really, since I instagated the importance of reacting to environment, but obviously in the wrong way where I mostly focused on light and air scattering...but that is not to say that motions over mountains wouldnīt be important, so in a sense you are right to include that with reactions to environments. :)

Vue has that as a special function for cloud motion over mountains..though I never tried that version, what they have made, or what you mean would help out for lightwave in this stuff, I do not know?

MichaelT
01-01-2018, 02:21 PM
it is possible to use modifiers to change the behavior of the clouds when they are changed/moved around. So by that I mean it is probably possible to simulate the behavior of how a cloud is expected to behave in certain situations. Granted, it is not a properly physical model of course.. but it should suffice in most situations I believe. I doubt most people would see if it is incorrect, unless they look for it (or happen to work with such things, then they would probably see it immediately :) )

tischbein3
01-01-2018, 02:31 PM
Interesting thing to notice, you can use one vdb file (from lets say blender) out of the cache rename it (so it does not have a sequence number) and it will be constantly displayed through the scene.

MichaelT
01-01-2018, 02:36 PM
Not sure about those Gardner clouds, I prefer building them by hand:
http://forums.newtek.com/attachment.php?attachmentid=139031&d=1530526
139031

BeeVee
01-01-2018, 02:42 PM
Have you compared the speed of flying through volumetric clouds in 2018 with doing the same in 2015?

B

MichaelT
01-01-2018, 02:45 PM
Nope.. but I can certainly try. But that will be done tomorrow.. it is getting a bit late here in Sweden, so I'll hit the sack soon :)

Julez4001
01-01-2018, 03:27 PM
Let it render over night.

Those Gardner clouds are gorgeous.
Post on LW WIKI

prometheus
01-01-2018, 03:43 PM
Not sure about those Gardner clouds, I prefer building them by hand:
http://forums.newtek.com/attachment.php?attachmentid=139031&d=1530526
139031

Itīs about texture tweaking, and I am not there yet to get a good one up for this new volumetrics, what do you mean tweaking by hand?..you must have a texture donīt you..then what texture you use is another matter.

prometheus
01-01-2018, 04:25 PM
some openvdb clouds, just found it..donīt think they are that great, but do check them out, change your primitive type to openvdp, and load them in, try the scattering channel and load the file there first, and change from none to density and it should pop up.
https://www.dropbox.com/s/xn95ui9zygeuknp/10%20VDB%20Clouds.zip?dl=0

Farhad_azer
01-02-2018, 12:33 AM
Gr8 thread. Keep it up fellas.

MichaelT
01-02-2018, 01:12 AM
1280x720 200frames took 4hrs 24 minutes. So that's OK. What is less ok is the animation that was going on in the volumes. Need to see if I can stop that somehow, or I have a problem. By its own, animation is OK, since clouds do animate.. but that should be something I choose to do. But like I said, need to check and see if I did something wrong before I say anything more. I haven't tried LW2015 but I don't need to.. I know it will be worse than this :)

----

What I mean is that I build the structure of the cloud by hand. I don't use procedurals as much as many others do I suppose. I'm a bit old school that way :)

prometheus
01-02-2018, 07:12 AM
1280x720 200frames took 4hrs 24 minutes. So that's OK. What is less ok is the animation that was going on in the volumes. Need to see if I can stop that somehow, or I have a problem. By its own, animation is OK, since clouds do animate.. but that should be something I choose to do. But like I said, need to check and see if I did something wrong before I say anything more. I haven't tried LW2015 but I don't need to.. I know it will be worse than this :)

----

What I mean is that I build the structure of the cloud by hand. I don't use procedurals as much as many others do I suppose. I'm a bit old school that way :)

Yes of course..but we all build or clouds by hand in some degree.
.either by forming particle clusters..or creating point clusters..or manually adding a couple of nulls.
That is sort of given..when you say you don't rely on procedural. .It sort of sounded like you do not use them at all..
I've tried them all I think...even using subpatch grids to control point density resolution and add voxel to each point in the grid..then a texture on density channel to set overall cloud density over a large area..then turn on hypertext uteservering on top of that for detail.

That is however obsolete now with the new volumetric item in cube mode.

And yes..as you seem to do..Some point creation..and a standard turbulence procedural does a good job without too much work...If that's sort of what you do?

MichaelT
01-02-2018, 07:32 AM
I do use procedurals.. but I tend to not rely on them so much. Like creating entire skies etc.. More the close detail things. Then I use nulls etc.. and build on them like that. So I get things like this:
http://forums.newtek.com/attachment.php?attachmentid=139056&d=1530690
139056

prometheus
01-02-2018, 09:36 AM
I do use procedurals.. but I tend to not rely on them so much. Like creating entire skies etc.. More the close detail things. Then I use nulls etc.. and build on them like that. So I get things like this:
http://forums.newtek.com/attachment.php?attachmentid=139056&d=1530690
139056

Yes..that looks very good, especially the softness low density, It could perhaps use more texture cut in for a slight more detailed cloud, colors and lighting looks very good.
Building several nulls is a huge render clogger, in former lightwave, two nulls overlapping..and bonk, you hit a wall, two particles..not a problem,
so there was one huge issue with former hypervoxels that really prevented us from taking clouds to another level, and especially since you just couldnīt move around single particles,
and I donīt think we can today in layout in a smooth way, except for some workarounds..scanning with particlefx and editFX node etc.

But with the new system that has been taking care of, I was one of those guys asking for improvement on that,
o adding several volumetric items in Lightwave 2018 doesnīt seem to clog the renderer, at least not vpr..so we now have the freedom to design and move volumetric items in the way we need to..
so that was about speed and freedom of designing it.

Then the quality of blending two nulls or particles, though I think I mentioned it earlier here, much better blend between two volumetric items,
best seen when you have texture mode set to non pyroclastic, but without actually adding a texture, itīs a difference between that and texture mode set to none..
so texture mode set to non pyroclastic and no texture active...make two of those and let them intersect and youīll see how much smoother the blend is compared to old hypervoxels, that is also something many of us asked for, it helps smoothing out several cloud items that intersect in the scene.

As I think I said before..I am not that happy about how the UI for setting it up is designed, I will probably tolerate it because of all the other improvements made for a better end result,
and will have to deal with it taking longer to set up and tweak, hypervoxels..you had one module window to see all items at once, and check and uncheck wich ones should be active or not,
and copy and paste settings..that stuff isnīt possible now..and the copy and paste hv settings tool for this new volumetrics seems buggy and not working properly, havenīt read the manual through for using presets though..not sure how that is better or worse.

I also do not like the fact that there is no random radius setting for the volume, unless entering nodes and dividing which nodes are needed for a proper random scale..
even if itīs only one node ..you n eed to add it ..plug it in the right input etc..so what previously took not a second..now takes several seconds, it may not seem a big deal..
but it makes this system really slow to tweak and use, fastest way may be to enter the particle emitter and check show size and use the -+ values of the particle emitters particle scale,
that will propagate through and affect the volumetric scale randomly, but that is also slower and spreading out the tweaking over several other item properties that isnīt directly within the volumetric system.

so many plus up and thumbs up for the quality in blending, softness and speed for some of the stuff in the new volumetrics, tweaking speed and ease of use..I give a thumbs down on.

you mentioned openvdp from blender and single files from cache sims, yes..tried some blender smoke sims yesterday, quite easy to do..
though both I and tibshien had several crashes when choosing various sequence files, both on the first, and some of the following sequence files,
and then it could also work on the first and on the following files depending on..I think he bug reported that.

And yes, that may be expected ..if you rename and remove the sequence number in the file extension, it treats it as one file you can use for flying through clouds,
I will look more in to this later this week as well, turbulenceFD was just about
to pic any cached file and load it ..no renaming there, but it doesnīt have to pass through open VDB if you got turbulenceFD.

I need to go in to depth of shading with nodes in OpenVDB, got some interesting fire shading going on by using nodes and gradients in scattering channel,
or you could use the blackbody radiator which has a temperature value, under node editor tools....though I yet have to learn the best settings for using that,
not sure if I need to check something in blenders fire and smoke sim, it doesnīt provide a temp,or fire kind of channel, as I think houdini may do..if it does, then I have missed it.

About moving clouds around mountains more physicly correct ..if not too complex for the job, one Idea would be to run a smoke simulation in blender, and simply use that openvdp,
or if you would like to have more control on the shading, perhaps trying to import a fluid simulation from an orthographic view in blender, the sim running itīs course around a mountain dummy object,
then try and load it in as a density map in the new volumetric system of lightwave instead of procedural textures, it may work..but it may be just as well better to just use openVDB.

Fun to see we now have so many new options though, and the way we can produce very soft thing gas clouds or whispy thin feather cirrus clouds is nice to see,
I have yet to post samples of that...havenīt worked on it..but based on what I found out accidently by feeding some inputs in various density channels within the new volumetric items,
it will be able to look really really good, probably better than what you can produce with vue for those kind of thin cirrus clouds.
And nebulas..just wait, it may be awesome once digged in to.

Im probably not going to do any more tests today, I was up too late yesterday..so I need to catch up with some sleep, and it is also junior hockey world championships tonight which every swede is obliged to follow ..eyh.:)

Ohh...just one more row of letters to say I am sorry for my lengthy replies.

MichaelT
01-02-2018, 04:41 PM
Yes..that looks very good, especially the softness low density, It could perhaps use more texture cut in for a slight more detailed cloud, colors and lighting looks very good.
Building several nulls is a huge render clogger, in former lightwave, two nulls overlapping..and bonk, you hit a wall, two particles..not a problem,
so there was one huge issue with former hypervoxels that really prevented us from taking clouds to another level, and especially since you just couldnīt move around single particles,
and I donīt think we can today in layout in a smooth way, except for some workarounds..scanning with particlefx and editFX node etc.

But with the new system that has been taking care of, I was one of those guys asking for improvement on that,
o adding several volumetric items in Lightwave 2018 doesnīt seem to clog the renderer, at least not vpr..so we now have the freedom to design and move volumetric items in the way we need to..
so that was about speed and freedom of designing it.

Then the quality of blending two nulls or particles, though I think I mentioned it earlier here, much better blend between two volumetric items,
best seen when you have texture mode set to non pyroclastic, but without actually adding a texture, itīs a difference between that and texture mode set to none..
so texture mode set to non pyroclastic and no texture active...make two of those and let them intersect and youīll see how much smoother the blend is compared to old hypervoxels, that is also something many of us asked for, it helps smoothing out several cloud items that intersect in the scene.

As I think I said before..I am not that happy about how the UI for setting it up is designed, I will probably tolerate it because of all the other improvements made for a better end result,
and will have to deal with it taking longer to set up and tweak, hypervoxels..you had one module window to see all items at once, and check and uncheck wich ones should be active or not,
and copy and paste settings..that stuff isnīt possible now..and the copy and paste hv settings tool for this new volumetrics seems buggy and not working properly, havenīt read the manual through for using presets though..not sure how that is better or worse.

I also do not like the fact that there is no random radius setting for the volume, unless entering nodes and dividing which nodes are needed for a proper random scale..
even if itīs only one node ..you n eed to add it ..plug it in the right input etc..so what previously took not a second..now takes several seconds, it may not seem a big deal..
but it makes this system really slow to tweak and use, fastest way may be to enter the particle emitter and check show size and use the -+ values of the particle emitters particle scale,
that will propagate through and affect the volumetric scale randomly, but that is also slower and spreading out the tweaking over several other item properties that isnīt directly within the volumetric system.

so many plus up and thumbs up for the quality in blending, softness and speed for some of the stuff in the new volumetrics, tweaking speed and ease of use..I give a thumbs down on.

you mentioned openvdp from blender and single files from cache sims, yes..tried some blender smoke sims yesterday, quite easy to do..
though both I and tibshien had several crashes when choosing various sequence files, both on the first, and some of the following sequence files,
and then it could also work on the first and on the following files depending on..I think he bug reported that.

And yes, that may be expected ..if you rename and remove the sequence number in the file extension, it treats it as one file you can use for flying through clouds,
I will look more in to this later this week as well, turbulenceFD was just about
to pic any cached file and load it ..no renaming there, but it doesnīt have to pass through open VDB if you got turbulenceFD.

I need to go in to depth of shading with nodes in OpenVDB, got some interesting fire shading going on by using nodes and gradients in scattering channel,
or you could use the blackbody radiator which has a temperature value, under node editor tools....though I yet have to learn the best settings for using that,
not sure if I need to check something in blenders fire and smoke sim, it doesnīt provide a temp,or fire kind of channel, as I think houdini may do..if it does, then I have missed it.

About moving clouds around mountains more physicly correct ..if not too complex for the job, one Idea would be to run a smoke simulation in blender, and simply use that openvdp,
or if you would like to have more control on the shading, perhaps trying to import a fluid simulation from an orthographic view in blender, the sim running itīs course around a mountain dummy object,
then try and load it in as a density map in the new volumetric system of lightwave instead of procedural textures, it may work..but it may be just as well better to just use openVDB.

Fun to see we now have so many new options though, and the way we can produce very soft thing gas clouds or whispy thin feather cirrus clouds is nice to see,
I have yet to post samples of that...havenīt worked on it..but based on what I found out accidently by feeding some inputs in various density channels within the new volumetric items,
it will be able to look really really good, probably better than what you can produce with vue for those kind of thin cirrus clouds.
And nebulas..just wait, it may be awesome once digged in to.

Im probably not going to do any more tests today, I was up too late yesterday..so I need to catch up with some sleep, and it is also junior hockey world championships tonight which every swede is obliged to follow ..eyh.:)

Ohh...just one more row of letters to say I am sorry for my lengthy replies.

Yeah, I've been looking at TFD. But since I already have Houdini installed (and its free + Blender) I just can't justify the cost. Especially since I don't use it often enough. Then there is also the issue about advection not being in TFD for LW.

No need to excuse your writing.. just means you are passionate about the subject ;)

BeeVee
01-02-2018, 06:00 PM
@Prometheus and MichaelT - make sure you get Feature Requests in for the lacks you perceive in the new volumetrics. You can do it directly from inside LightWave, in the Help menu. Try to make each feature request about a single thing, rather than just piling the whole lot in a single report.

B

MichaelT
01-02-2018, 07:31 PM
@Prometheus and MichaelT - make sure you get Feature Requests in for the lacks you perceive in the new volumetrics. You can do it directly from inside LightWave, in the Help menu. Try to make each feature request about a single thing, rather than just piling the whole lot in a single report.

B

:) Got it. But I will let things sink in before making requests. So I don't end up making a oh-god-why-didn't-I-think-of-that request ;)

MichaelT
01-03-2018, 11:41 AM
@Prometheus: I made a higher resolution one, to see how things are when resolution increases.. and it does it very well I think. The godrays in the clouds are more visible now too:

139133

prometheus
01-03-2018, 12:01 PM
@Prometheus: I made a higher resolution one, to see how things are when resolution increases.. and it does it very well I think. The godrays in the clouds are more visible now too:

139133

Looking darn good Michael, much much better, resolution? you mean step size or something else I missed.


The shading luminosity, color and fine detail is looking just great..as well as the softer parts of the clouds, though the larger scope of the clouds, the main shape seem a bit too stiff, could use a reshape perhaps.
More shadow depth in there in this one, makes it look better I think.
But anyway ...good job there.

I Can not see godrays really in the image though.

MichaelT
01-03-2018, 12:17 PM
You can see the rays in the softer parts of the clouds. But I still have a bit of a problem making them visible. But I'm sure I'll find out the solution soon enough :) I rendered the image as 4K. The image here is downsized to about 2K. I also used the full precision and exported EXR, then processed the image in Photoshop. So a bit of exposure control etc.. was needed to bring the details out. As soon as I solve the godrays issue, so I can make them more clear in the atmosphere, I'll move on to see if I can get the light spectrum coloration that should appear near softer areas.

bazsa73
01-03-2018, 12:25 PM
Bara bra! :) These clouds look great. I had less luck with openvdb fire from Blender.

MichaelT
01-03-2018, 12:43 PM
Thanks :) .. I've only played with OpenVDB once in LW so far... but I'll get around to it.

prometheus
01-03-2018, 12:55 PM
You can see the rays in the softer parts of the clouds. But I still have a bit of a problem making them visible. But I'm sure I'll find out the solution soon enough :) I rendered the image as 4K. The image here is downsized to about 2K. I also used the full precision and exported EXR, then processed the image in Photoshop. So a bit of exposure control etc.. was needed to bring the details out. As soon as I solve the godrays issue, so I can make them more clear in the atmosphere, I'll move on to see if I can get the light spectrum coloration that should appear near softer areas.

Great..do that, I am too tired when I get home right now, so it will probably not be until friday when I can continue testing the new volumetrics, I was trying to use distant light as volumetric..and I thought it could be activated as the others, but it doesnīt seem to be possible, so you would have to go with a point light ..which I think you did here, it is a bit harder to set properly and you have to eyeball it in to match a decent result..but it is sort of a hack.

I tested the lw 2018 content and the forrest scene, check that one, and perhaps play around with scattering asymmetry in the volumetric tab/volumetric intergrator, small values only ranging between 0-0,6 perhaps...and also..increase the lights volumetric density.

Since the distant light canīt be volumetric in lw 2015 nor here in 2018 as I know off, and if I do not want the point light, I tried the volumetric trick of simply adding a volumetric primitive to serve as fog, and let it pick up godrays projected trough a cloudplane and transparency, ive done that before with ordinary old hypervoxels, and it works here too with the new volumetric primitive..though you need to set a very very low density...this way the distant light can be made sort of volumetric, but this is the kind of thing I thought the new volumetric intergrator should take care of..so we donīt have to go that route of setting all that up, I will demonstrate that at the end of the weekend I think, though I am doubtful if I would try and use both a cloud volumetric primitive as well as mixing it with a volumetric primitive as fog, I will check..it may be acceptable since we know now that you can add several volumetric primitives unlike hypervoxels without slowing it down too much.

The advantage of using that technique may be that you can gett better control over fog, and increase or decrease godrays in a more flexible way, as well as adding actual various density textures to that fog.

- - - Updated - - -


Bara bra! :) These clouds look great. I had less luck with openvdb fire from Blender.

Fine Fine :)
what was the problem with blender fire?

prometheus
01-03-2018, 03:31 PM
Darn it ...should be in bed now.

Anyway..since we mentioned godray here, well...distant light doesnīt seem to be able to work as volumetric light, and using a point light poses difficulties of posing it right in distance behind clouds etc, and also getting good density ..scattering, itīs not that easy and really not that realistic either, and sunspot modifier doesnīt seem to work on the point light, and manually moving it down and left etc is a mess, thatīs why I wanted the distant light or sk sun to work...Besides..I also feel the volumetric sunllight is too slow to iterate render in VPR when mixing with volumetric clouds.

So what to do, well..I resorted to the volumetric item trick again, simply adding a volumetric primitive and make that a fog with proper density, so now I can use distant light as casting godrays, and it actually also seem Faster than the volumetric point light, so no volumetric light here..just the volumetric item as fog.

What is also cool about it..that is I can control height..like in vue sort of, and thickness etc, density in ways I do not see possible with using volumetric scattering, maybe if I go nodal in there.

So a sneak image, nevermind I havenīt really tweaked the clouds to look the way I want to here, just testing the godray stuff...itīs not perfect, and I think the rays may cut off to fast before finnishing itīs ray length so to speak.

Garnder clouds only available in standard texture layers, so add a scalar layer in node editor and add it from there, nodal wont work.
right cloud also put a scalar value on texture density to test smoothness, and the left is a bit harder in edges.

http://forums.newtek.com/attachment.php?attachmentid=139143&d=1515018487

BeeVee
01-03-2018, 04:02 PM
That's lovely! Reminds me of Miyazaki.

B

bazsa73
01-03-2018, 04:22 PM
Fine Fine :)
what was the problem with blender fire?

If I'd known :) I can succesfully set up fire and smoke in Blender/Cycles but the new LW engine is full of mistery.
It just doesn't look right. Need more tweaking, fine tuning I guess on my behalf.

prometheus
01-04-2018, 11:11 AM
If I'd known :) I can succesfully set up fire and smoke in Blender/Cycles but the new LW engine is full of mistery.
It just doesn't look right. Need more tweaking, fine tuning I guess on my behalf.

When you say set up fire and smoke setup, you do mean also the shading with nodes, absorbtion, scatter nodes, multiply nodes, emission nodes etc? just wondering.

my first try with some additional scatter data gradient, theres a blackbody radiator node in lw 2018, but I need more time to learn how to use it...Properly that is, to much artifacts or too low resolution somehwere as well.

http://forums.newtek.com/attachment.php?attachmentid=139174&d=1515089418

139174

prometheus
01-04-2018, 11:46 AM
Since the thread is hypervoxels VS the new volumetrics, I will try and post samples of how the blending looks like in the new system, and in the old hypervoxels etc.

but for now, just an old lightwave 2015 image of a single null, single voxel to work as a cloud, flat bottom with gradients in node editor, as you can see the edge softness isnīt anywhere near the edge softness you can get with the new volumetric system, neither is the lighting and shading as realistic as the new system is, but it has it charm and it was easy to set up and tweak (for me anyway) and quite easy to control detail, while I find that more tricky and taking longer time in the new volumetric system.

The feather clouds here in this hv image could also be made much better with the new volumetric system, thanks to itīs much better softness, yet being able to keep a kind of depth of shading..if needed.

https://lh3.googleusercontent.com/aoNDOrmBb2YCUI8Nx5l-EDjlKQaN01bchuCDzYctEwlENE2QDAiHF1GJBKdIF_a4ueJVlW JLX3JfIQ=w1816-h1030-no

MichaelT
01-04-2018, 12:31 PM
The problem as I see it is that the clouds are not making much of a show outside its own volume. Which is very strange. A bug? If I use any other object the shadows are clear, but not from the voxels. I could do a hack to fix it, but the thing is that I know very little about how to code things using the SDK. I tinkered a little but, then I had other things to do :/ With the new SDK I'm back to square 1 anyway. But I will have to look at it I think at some point. Now I need to look at other things.

prometheus
01-04-2018, 01:31 PM
The problem as I see it is that the clouds are not making much of a show outside its own volume. Which is very strange. A bug? If I use any other object the shadows are clear, but not from the voxels. I could do a hack to fix it, but the thing is that I know very little about how to code things using the SDK. I tinkered a little but, then I had other things to do :/ With the new SDK I'm back to square 1 anyway. But I will have to look at it I think at some point. Now I need to look at other things.

I donīt follow you here, what sample do you refere too? which clouds...post it please so I can jump in and see if I know how to treat it.

MichaelT
01-04-2018, 06:31 PM
If you take a volume, and a box. Put both in another volume (but less dense of course) shine a light on them. Assuming you have volumetrics active. The box should have a very noticeable shadow on the other side, whereas the volume does not. There is a darker area for the immediate volume, but there is no clear shadow from it. At least not as I could get so far at least. And I even went as far as making the volume so dense and light catching it went black. Still no shadow. That doesn't look right to me. The thing is that I can get shadows from within the volume itself.. but not outside of it. But like I said.. so far... maybe I'm doing it wrong? And I don't want to post a bug report.. since I don't even know if it is a bug to begin with. For all I know it can be crap behind the keyboard :D

prometheus
01-04-2018, 07:13 PM
I can only guess
How you set it up..and i am actually
Sleeping..temporary up for a wee.
Have you tried asymmetry scattering. .perhaps minis values.
You may also ned
D to feed the texture in to various
Cannels in node Editor.

erikals
01-04-2018, 08:04 PM
Still no shadow.
could Scattering or Absorption fix it?


Sleeping..
gosh, none of us Scandinavians have a day-rhythm ?!  :)

:sleeping:

prometheus
01-05-2018, 12:41 AM
could Scattering or Absorption fix it?


gosh, none of us Scandinavians have a day-rhythm ?!  :)


:sleeping:

How could we..where's the sun..:D
Yes...I noticed that you..me and MichaelT should be sleeping
Certain hours normally...
I'm off t
Will have to look at this later.

UnCommonGrafx
01-05-2018, 04:38 AM
Sunsky env for the sun.

prometheus
01-05-2018, 05:02 AM
Sunsky env for the sun.
That's a response on our northern conditions of lacking the sun?...or
Do you refer to any connection it has
With volumetrics?

MichaelT
01-05-2018, 05:35 AM
Nodes etc.. might help.. haven't tried. But the main take away here is that the other volume I put inside the larger volume is actually black. That in itself should cast a shadow. But it doesn't. Which is the strange part.

prometheus
01-05-2018, 06:12 AM
Nodes etc.. might help.. haven't tried. But the main take away here is that the other volume I put inside the larger volume is actually black. That in itself should cast a shadow. But it doesn't. Which is the strange part.

In the vpr settings, have you checked the volume shadows on? if not...a volume item inside of a volume item will not cast any shadow, while a geometry object will cast a shadow inside of a volume item, this can be useful if you only want to use the old trick I do with transparent textures on a cloud plane, and use a volume item to pick up the sunlight for godrays, but it wouldnīt work if you instead of a cloud plane would use a volumetric cloud.
Left item is a volumetric primitive, and the right a simple cube geometry, and having volume shadow off in the vpr settings, if I had it turned on, I would get a shadow from the volume item, though not as strong as the geometric cube.


http://forums.newtek.com/attachment.php?attachmentid=139190&d=1515158042

UnCommonGrafx
01-05-2018, 06:17 AM
That's a response on our northern conditions of lacking the sun?...or
Do you refer to any connection it has
With volumetrics?

Both, on retrospect. First thing I saw and, as it was something I was looking for, shared my findings.
Then noted the conversation trend... shrugged my shoulders and said, "inclusive or". Umm... said, "eh, applies to both.".

So I left it.

A LOT of goodness in this release on the volumetrics side.
Robert

prometheus
01-05-2018, 06:35 AM
Both, on retrospect. First thing I saw and, as it was something I was looking for, shared my findings.
Then noted the conversation trend... shrugged my shoulders and said, "inclusive or". Umm... said, "eh, applies to both.".

So I left it.

A LOT of goodness in this release on the volumetrics side.
Robert

sunsky is good to use, but it is half broken since we can not use the dedicated sk_sunlight, and it had properties I think would be hard to apply to the distant light.
as far as volumetrics, yes a lot of goodness, but it still seems HALFBAKED, and UNFINISHED...In the sense that it doesnīt feel mature enough, and some things doesnīt seem right either as the topic goes here, and when implementing things like this..I donīt think it was good to release it without a proper sunsky model natively.
That is my harsh judgment about it, that said, great quality in softness and how it blends and how we can work with several null volume items.

I think this may be the issue MichaelT referes too? the left volume item does not cast itīs shadow long enough and cuts it off too abruptly, that is also what I mentioned in my godray image which affects the realism and quality of the cloud, it shouldn behave like that.
Again..note..this is no volumetric light, just using volume item as fog, and one volume item cube, and one geometric cube to the right.

http://forums.newtek.com/attachment.php?attachmentid=139192&d=1515159704

prometheus
01-05-2018, 07:06 AM
forgot to say, if you use sunsky environment, and pitch rotate your distance sun, it may be a good Idea to enter the distant lights color channel envelope, and add the sunsky suncolor modifier ..so you get a more realistic color decay distributed on to the clouds.

UnCommonGrafx
01-05-2018, 08:06 AM
Good tip.
I would say half fixed for where we are today. :)
There is more to come with this foundation laid.
Robert

MichaelT
01-05-2018, 09:11 AM
I've now managed to cast shadows in another volume (I didn't use the sunsky) but the effect is very limited. I'm actually a bit concerned by this. because the volume in question is very dark, and very absorbent. But the thing is.. even white clouds casts shadows. But I'll have to test a bit more.

prometheus
01-05-2018, 10:11 AM
I've now managed to cast shadows in another volume (I didn't use the sunsky) but the effect is very limited. I'm actually a bit concerned by this. because the volume in question is very dark, and very absorbent. But the thing is.. even white clouds casts shadows. But I'll have to test a bit more.

That figures, sunsky has nothing to do really with the shadows I believe, but anyway good to know the process, though I would like to see images of the issues, if you find it worth to post with remarks or something.

I got home early today, whipped up the volumetrics again, and was testing the old planetry model, in this case I just throwed in two volume items, one for the earth surface which should actually be replaced with real geometry and surfacing..but it was a quick setup, and I got the second volume item for the clouds..quite big scale of the planet, but not in real life scale I would say...I saw someone post a planet fly through on lightwave wiki, looked quite good..except unproportionally scaled clouds, especially in height.

Here I was experimenting with colors, and made some experimental node feeding ..which happened to end up in one image yielding a colourful effect that looks a bit like a spectral model (though I suspect itīs the wrong approach) will have to learn and look more at the emission, absorbtion and scattering input.
The color of scattering and absorbtion is tightly connected, so balancing right values makes the final color get synced.

This model lacks a proper atmosphere as well, will look in to that as well..if it doesnīt slow down the rendering too much.

Images, same gardner clouds model with previous gradient layer to control cut off, but I will post some more images with the weather procedural as clouds later maybe, itīs probably a better cloud model.


http://forums.newtek.com/attachment.php?attachmentid=139196&d=1515172039
http://forums.newtek.com/attachment.php?attachmentid=139197&d=1515172074
http://forums.newtek.com/attachment.php?attachmentid=139198&d=1515172089




139196
139197
139198

MichaelT
01-05-2018, 10:55 AM
The spectral effect looks right actually. The problem here would be that you probably need a second volume to act as atmosphere. So you get a more coherent effect towards the horizon. I don't think the shadow effect would be a problem here either, as the shadows within the same volume would probably still be quite visible. The only thing I would say is that the clouds are too thick from that distance.

prometheus
01-05-2018, 11:06 AM
The spectral effect looks right actually. The problem here would be that you probably need a second volume to act as atmosphere. So you get a more coherent effect towards the horizon. I don't think the shadow effect would be a problem here either, as the shadows within the same volume would probably still be quite visible. The only thing I would say is that the clouds are too thick from that distance.

I just did...though the atmosphere may be better of fwith a larger radius, or a bit thicker perhaps, I have to try various tweaks later, I also added a light info node with shadow color output fed in to the emission of the volume_item called earth surface, since that has an emission of itīs own so to speak, and previously the surface didnīt turn dark enough as it should in space, and the clouds on top of that became illuminated in the wrong way, so this is better I think.

Mind You...all this is done by how I would think it may look like in the back of my head so to speak, without actually even looking at references, thatīs where I should start before going in to deep with experimenting.


http://forums.newtek.com/attachment.php?attachmentid=139199&d=1515175546
139199

MichaelT
01-05-2018, 11:15 AM
Hmm, interesting... looks like the shadows show more details than the clouds themselves (bottom left) Almost like the transparency is going on overdrive for the shadows. Node adjustments? Or is that just a trick of the eyes, and you have texture on the planet too?

prometheus
01-05-2018, 11:42 AM
Hmm, interesting... looks like the shadows show more details than the clouds themselves (bottom left) Almost like the transparency is going on overdrive for the shadows. Node adjustments? Or is that just a trick of the eyes, and you have texture on the planet too?

Not really more detail, and its not shadows I think, itīs just the clouds going dark, but not completly dark..so it isnīt properly set yet I think.
No textures on the earth surface volume item at all, just volume item.

MichaelT
01-05-2018, 11:54 AM
Ah, so a little of both then. A trick of the eye + some adjustments :)

prometheus
01-06-2018, 01:56 PM
Crossposting..but should be both here and in the opinion thread..

Some positives...
And a sideeffect of the new render engine I suspect, what do you know, hypervoxels are now able to render multiple hypervoxels instances ..without it hanging the system as it did in the lw 2015, not sure if it ever was possible to bypass in the old system, but here in lightwave 2018 it works just as galant as with the new volumetrics, itīs still the same hypervoxels without any new changes otherwise it seems, ambient light is working though ambient light is removed overall for lights, so this was a pleasent surprise should I need the characteristics of hypervoxels.

And to note..hypervoxels has sprites and sprite clips if that should be required, which doesnīt exist in the new volumetrics I think, so they complement eachother, if you need fast demonstration smoke, go for sprites, if you need full realism of clouds, go for the new volumetric system, not sure if this means you can also fly through hypervoxels in the same speed as the new volumetrics or not.
I havenīt checked up on baking hypervoxels if thereīs any additional improvements there.

So some positives about that.

http://forums.newtek.com/attachment.php?attachmentid=139236&d=1515272174

MichaelT
01-06-2018, 07:22 PM
My main problem is the precision. When using decimals we only have like 0.12345 and that is waaaaaaaay to little. This they should absolutely fix.. because that is like half the precision of a float even. A far cry from let's say a double. Let alone 128bits (long double.. or high precision) I can live with it.. but I absolutely don't like it.
Because it affects (or rather limits) what is possible, in terms of controlling other limits and other aspects of the engine.

Anonymous101
01-07-2018, 07:00 AM
Thx for this interesting Thread, learned something by reading this. Let me say i have no experiance with HV in LW 2015, so i am a rookie in such things. For me it was easy to open the HiRes Earth from LW2018 content and play a little bit with it...next week i hope there is time for me to try some things with the new HV Primitives and adding some clouds Textures to the Node to have a better real looking cloud cover around the global earth. As i see it in the moment there are problems to map the texture (added in HV Primitive Node) correct, so i tryed to use a second camera as front projector while the main cam moves, but this is not the best result for all cases...i will try how textures will be mapped with a checker texture and see what happend with the corner areas...excuse me for my bad english please...looking forward,

Photogram
01-07-2018, 05:06 PM
Here is my first animation with the new volumetrics lights!

45 min for the rendering of 162 frames in HD

https://vimeo.com/250046457

prometheus
01-07-2018, 06:37 PM
Huhhhh...whatīs this?

Just realised that the new volumetrics doesnīt seem to be capable of recognizing point clusters..if that is the case, that is BAD, sometimes I use fill solid or just create point clusters and apply hypervoxels directly on it.
Not sure is this needs node item info somhow when setting up a null then assign other data, or do I need particles on those points...hmm.

Added a particle emitter, set it as volumetric primitive..changed particle size to random size...crash.

prometheus
01-07-2018, 07:00 PM
Another minus..I think, just looked at the particle blending, I was kind of fooled I think by using two nulls volumetric items, they blended nicely much better than previously, but the particle blending..still the same it seems unfortunately, it doesnīt compete with the old dynamite blending per particle.:cursin:

avkills
01-08-2018, 02:00 AM
prometheus, where are you getting that Gardner procedural from?

Anttij77
01-08-2018, 03:03 AM
I've made a tutorial video you might be interested in:

https://youtu.be/HLohazy4A-Q

prometheus
01-08-2018, 03:53 AM
prometheus, where are you getting that Gardner procedural from?

from dpont, but be aware, if you use lightwave 2018, you can not use the rman collections node textures, they work in 2015 but not in 2018, a workaround which I did, is to use lightwaves native scalar layer node, open that node, then you can add the rman collection textures from there and then of course feed the scalar node output to the volumetric intergrators texture input.

you want to download the 64 bit rman textures if you use lightwave 2018..on top of this page, download the nodes on bottom of the page if you use lightwave 2015

http://dpont.pagesperso-orange.fr/plugins/Download.htm

http://dpont.pagesperso-orange.fr/plugins/Textures.htm

A side note after testing old hypervoxels in lightwave 2018, thanks to the new distant light or spherical light, and size and increasing samples while checking normal on the ligh, it now seems hypervoxels can get much much more detail and depth in the old hypervoxels, thatīs nice, it does however lack the softness and some other features the new volumetric system has, but it can actually be faster if you turn of volumetric shadows in VPR though forsaking some realism, I actually think that some of the old hypervoxels stuff in 2018 seem faster than the new system, it is easier to tweak light and shadow, and easier to copy and paste between nulls, and deactivate or activate.

prometheus
01-08-2018, 04:02 AM
I've made a tutorial video you might be interested in:

https://youtu.be/HLohazy4A-Q

Great..just fast skimmed it, I was actually going to test earth with primitive shapes, but havenīt had the time..I think I actually may do that before even looking at yours, even though you make it easier for us, itīs just the way I want to learn, by trying for myself..and only if I get stuck I resort to training, thanks very much though..will at some point look through it carefully.

You do not happen to know? if I missed something regarding using point clusters and the new volumetrics..I posted previously that it seems the new volumetrics doesnīt recognize sprayed points or points created by fill solid, the old hypervoxels does that..but maybe itīs on the planning, just not yet implemented?

MichaelT
01-08-2018, 07:16 AM
I've made a tutorial video you might be interested in:

https://youtu.be/HLohazy4A-Q

Nice... thanks :)

Anonymous101
01-08-2018, 08:01 AM
In terms of Hypervoxel I'm a Rookie, but nevertheless i will give a short explanation to how i set up a cloud Texture with the Hypervoxel Primitive Node. First i used a Checkered Texture to have a look on how to adjust the Cloudmap on the Cloudlayer. Maybe you can fix this in other way, but for me it works with Spherical mapping, X Axis and rotated like you can see on screenshot. Then i applied the Clouds map instead of the checker and lowered the scattering scale in contrast to the content scene. Played a little bit with the Texture Density in the Object Properties and with the Opacity of the Texture that is pluged to the HV Primitives Node. I am not the one who can give big tutorials, but i hope someone of you can use it for own tries

avkills
01-08-2018, 09:53 AM
I've made a tutorial video you might be interested in:

https://youtu.be/HLohazy4A-Q

Very nice, thank you for this Antti!

Antti the Mac version is in no way acting like what you are doing in the Windows version in that video! Can you get someone at Newtek to follow that tutorial on the Mac and let me know.

prometheus
01-08-2018, 10:06 AM
In terms of Hypervoxel I'm a Rookie, but nevertheless i will give a short explanation to how i set up a cloud Texture with the Hypervoxel Primitive Node. First i used a Checkered Texture to have a look on how to adjust the Cloudmap on the Cloudlayer. Maybe you can fix this in other way, but for me it works with Spherical mapping, X Axis and rotated like you can see on screenshot. Then i applied the Clouds map instead of the checker and lowered the scattering scale in contrast to the content scene. Played a little bit with the Texture Density in the Object Properties and with the Opacity of the Texture that is pluged to the HV Primitives Node. I am not the one who can give big tutorials, but i hope someone of you can use it for own tries

Looking good, I think that is sort of the approach I would go for too, if using texture maps, just havnīt gotten too it and I like to play around with fictional my own design of clouds, so it is often merely procedrual textures, one project I would like to try out, that is doing an Alien planet, meaning something like jupiter etc..with swirling clouds in several different bands crossing the planet, maybe simulate a smoke simulation and map it as a texture map sequence, either rendering out several passes of different swirls and stitch together somehow, or just try and do it with several smoke flow objects in the same domain container, foremost thinking of blender smoke sims, since I do not have any commercial turbulenceFD version, but I am also quite busy trying all the New 22018 lightwave stuff, as much as I can, and I must check all the scene content.
Thanks for posting your screenshots.

Anonymous101
01-08-2018, 10:27 AM
switching the Light on...

Anonymous101
01-08-2018, 10:29 AM
Looking good, I think that is sort of the approach I would go for too, if using texture maps, just havnīt gotten too it and I like to play around with fictional my own design of clouds, so it is often merely procedrual textures, one project I would like to try out, that is doing an Alien planet, meaning something like jupiter etc..with swirling clouds in several different bands crossing the planet, maybe simulate a smoke simulation and map it as a texture map sequence, either rendering out several passes of different swirls and stitch together somehow, or just try and do it with several smoke flow objects in the same domain container, foremost thinking of blender smoke sims, since I do not have any commercial turbulenceFD version, but I am also quite busy trying all the New 22018 lightwave stuff, as much as I can, and I must check all the scene content.
Thanks for posting your screenshots.

I agree...i think this will work and the possibilities to create alien planets with this are great as well....i am exploring LW2018 too and this days i dont know where to beginn...but thats all ok...its pure fun! :)

avkills
01-08-2018, 10:55 AM
Just in case you did not see my edit Antti.

Antti the Mac version is in no way acting like what you are doing in the Windows version in that video! Can you get someone at Newtek to follow that tutorial on the Mac and let me know.

Anttij77
01-08-2018, 02:21 PM
Just in case you did not see my edit Antti.

Antti the Mac version is in no way acting like what you are doing in the Windows version in that video! Can you get someone at Newtek to follow that tutorial on the Mac and let me know.

Can you file a bug report? With content.

Anonymous101
01-08-2018, 03:22 PM
Little Test with Nightlights

Photogram
01-08-2018, 06:11 PM
Huhhhh...whatīs this?

Just realised that the new volumetrics doesnīt seem to be capable of recognizing point clusters..if that is the case, that is BAD, sometimes I use fill solid or just create point clusters and apply hypervoxels directly on it.


There is no particle, null or hypervoxels in my animation, this is just light with nodal. I took the truster example in the new content and after some modifications i get this.

- - - Updated - - -

Here's another animation with some modifications.

https://vimeo.com/250204918

avkills
01-08-2018, 08:41 PM
Can you file a bug report? With content.

Yes once I can nail down exactly why it isn't working the same and can verify. So I think I can chalk this up as operator error in some ways. Did not realize the color picker had been switched over to floating point, and that threw me off quite a bit.

I do have 1 question Antti, the distant light does not seem to physically be in the correct place when the angle is changed to the 5˚; lens flare outer glow is no where near the visible light.

prometheus
01-09-2018, 04:26 AM
There is no particle, null or hypervoxels in my animation, this is just light with nodal. I took the truster example in the new content and after some modifications i get this.

- - - Updated - - -

Here's another animation with some modifications.

https://vimeo.com/250204918

I donīt get what you are talking about? since you replied with quote on my comment on volumetrics on points not working.
what does your volumetric light samples has to do with my statement of the General situation regarding the new volumetric primite system not being able to work directly on point clusters?
That has nothing to do with volumetric lights.

Anonymous101
01-12-2018, 04:25 AM
I did a little Earth Tutorial, hope anyone can use it. Its based on the Earth Content Scene.

erikals
01-15-2018, 08:17 PM
Darn it ...should be in bed now.

Anyway..since we mentioned godray here, well...distant light doesnīt seem to be able to work as volumetric light, and using a point light poses difficulties of posing it right in distance behind clouds etc, and also getting good density ..scattering, itīs not that easy and really not that realistic either, and sunspot modifier doesnīt seem to work on the point light, and manually moving it down and left etc is a mess, thatīs why I wanted the distant light or sk sun to work...Besides..I also feel the volumetric sunllight is too slow to iterate render in VPR when mixing with volumetric clouds.

So what to do, well..I resorted to the volumetric item trick again, simply adding a volumetric primitive and make that a fog with proper density, so now I can use distant light as casting godrays, and it actually also seem Faster than the volumetric point light, so no volumetric light here..just the volumetric item as fog.

What is also cool about it..that is I can control height..like in vue sort of, and thickness etc, density in ways I do not see possible with using volumetric scattering, maybe if I go nodal in there.

So a sneak image, nevermind I havenīt really tweaked the clouds to look the way I want to here, just testing the godray stuff...itīs not perfect, and I think the rays may cut off to fast before finnishing itīs ray length so to speak.

Garnder clouds only available in standard texture layers, so add a scalar layer in node editor and add it from there, nodal wont work.
right cloud also put a scalar value on texture density to test smoothness, and the left is a bit harder in edges.

http://forums.newtek.com/attachment.php?attachmentid=139143&d=1515018487

this might be on to something.

tweaked in post, might give some ideas... >
a very rough test here, could look way better if i had more time (with correct alphas etc.)

but this is crazy, i need to hit the sack... [4.20 am]

https://i.imgur.com/t4J2uK8.png

prometheus
01-15-2018, 08:24 PM
this might be on to something.

tweaked in post, might give some ideas... >
a very rough test here, could look way better if i had more time (with correct alphas etc.)

but this is crazy, i need to hit the sack... [4.20 am]

https://i.imgur.com/t4J2uK8.png

yes you really should hit the sack, I on the other hand will soon be off to work and compensate by going to bed directly after work, so I got a few hours before work.
Great post work there, though for my sample, I could perhaps get that result if I just scaled up my fog item higher in the air, as you can see the fog item only goes so far up in the sky, so a stretching or moving of the cloud fog item (volumetric primitive in box shape) should have produced he same results almost.

js33
01-16-2018, 03:40 PM
Little Test with Nightlights

Your Earth is spinning the wrong direction.

Anonymous101
01-17-2018, 12:10 AM
Your Earth is spinning the wrong direction.

Please excuse me for bothering you with such a terrible mistake...

prometheus
01-17-2018, 10:42 AM
One thing that really bugs me right now in terms of workflow with the new volumetrics, unlike hypervoxels that has itīs own window module, the new volumetrics and settings can only be accessed for each item when you have it selected, itīs really destroying a very good workflow where you with the old hypervoxels actually could move around any item and have various items selected and at the same time having the volumetric system settings in front of you and active without having to worry about which item you have selected..

Thatīs a thumbs down by me for the workflow with accessing the volumetrics in 2018:thumbsdow

erikals
01-17-2018, 11:06 AM
one will have to weight it against the thumbs up.

prometheus
01-17-2018, 11:54 AM
one will have to weight it against the thumbs up.

Absolutely, there are sure some good things with the new volumetrics, and despite that I think hit isnīt mature in the workflow in the same manner as hypervoxels, and in fact lacks a lot you can do with hypervoxels, it still can produce softer edges of the volume and more realistic physical shading I think.

If the Dev team listen and think itīs a good idea, and if it is implementable, I think there would be room for improving how we access the new volumetrics, copy and paste between nulls and edit it ..while still being able to selectother items in the scene without the volumetric setting being unaccessable.

It is however not as fast as I wished it to be...To note, I think I have noticed that old legacy hypervoxels (in 2018)under certain circumstances, is the absolutely fastest volumetrics you can put up, in fact it now may be possible to get close to the speed of baked hypervoxels, slower..but far more faster than previously, what you need to do is to turn of volumetric shadows in VPR, do not use shadow textures, but with these settings off, the old hypervoxels is much much faster than with 2015..and it doesnīmatter if you clone several nulls of them, which was a major freezing factor previously in 2015, and not only that it renders multi instances of hypervoxels fast, you can also fly through them quite fast, at least when running a VPR preview.

If I only got time to showcase it, but due to those speed increasements that affects the Old legacy hypervoxels, we may get volumetrics clouds more realistic, thicker and detailed than baked hypervoxels, within a decent time. If you were to use the full new volumetric system, I think it would be considerably slower, the new system may be faster though if you compare to enabling textured shadows and turning on volumetric shadows in vpr, so this speed trick with the old hypervoxels will not yield the most realistic volumetric model, but it will look better than baked Hvīs-

prometheus
01-17-2018, 12:00 PM
As for workflows, personally I have moved the image processing under all other editors so they are always acessable no matter what main tab I choose, and I have done so also with the new volumetric fog options, do not get why they hide so many stuff under more tabs..there is room for them, and they are vital for the workflow to see instantly.
I wonder if it wouldnīt be better to have those in the editor tabs by default ..instead of hiding them, since it is customizable...anyone can move them out of the editor menuīs if they run out of menu space or something.

prometheus
01-17-2018, 12:18 PM
two features from the old dynamite plugin I would especially would want, that would be multi editing of several nulls, neither hypervoxels or the new system allows for that, in dynamite you could set several cloned nulls as slaves, so when tweaking the master, the other null volumes would change as well, but they are independent in the sense that you can move them seperately.
And of course, the openGL preview that actually shows a polygon volume representation, currently neither the new system or old hypervoxels provides a good enough openGL presentation.

prometheus
01-20-2018, 01:24 PM
bumping back..

previously I was often using particle emitters in real life scale, ranging from thousands of meter to several thousands of meter for the particle emitter, then applying around 1000-3000 particles generated per frame, then through in hypervoxels with various random size for a cloud look.

I though having a go at it with the new volumetrics, should be rendering faster and even faster as you get closer to the clouds...But, it seems the new volumetrics to not fit well on particle use at all, as opposed to single nulls, when I try to adapt the new volumetrics on the same particle emitters, it becomes horribly slow in vpr, even if I scale up the step size a couple of meters, and besides..doing that will make the quality become horrible.
and to further that..as mentioned before, you have to use the Emitter particles size random scale in order to get some more natural variation..and it isnīt as nice to work with as with hypervoxels own random control.

I disconnected the new volumetric system, and applied the old hypervoxels on them instead...Itīs blistering fast compared to the new volumetrics, and getting textured clouds result in a fraction of the time it takes to go nodal and set up textures in the new volumetrics system.

prometheus
01-21-2018, 12:10 PM
Regarding applying the new volumetric system on particles, like for aeroplane smoke etc..as I said above when testing on particle fields for clouds, it becomes horrible slow, when I now tried to use it on aeroplane emitter smoke, I noticed that it doesnīt seem to be the volume shade per say, but only when I apply a node texture and feed it to the texture input in nodes, that is when vpr starts to slow down extremely, unconnecting it, and the volume spheres around particles are at a decent speed....so what is going on with the textures passed through the node system?? using the old hypervoxels and from the hypertexture drop down list, that is blistering fast compared.

prometheus
01-21-2018, 01:58 PM
and hereīs some settings for the black body radiator used with particles and the new volumetrics, some notes about missing random scale, and the fact that applying textures when using particles will make vpr come to a halt in rendering..itīs horrible in refreshment, once disconnecting textures and only using volume, it gets back up to speed quite a bit, those are the issues I have currently with the new volumetrics and particles, which makes it unusable ..despite the cool black body temp node.
Nodes for black body is done based on my own gut feeling, havenīt read up on it or studied any other scene for it.


http://forums.newtek.com/attachment.php?attachmentid=139647&d=1516568281

139647

Photogram
01-21-2018, 03:49 PM
I did a little Earth Tutorial, hope anyone can use it. Its based on the Earth Content Scene.

Thanks i think i will read your tutorial to get better result on my first volumetric earth test.

I got so much difficulty achieving the clouds arround the earth and especially when landing on earth the clouds are not realist and appear smear. The gradient node are difficult to setup because the gradient values are so close that is impossible to tweak perfectly. Maybe somebody have good advices i wondering to get better clouds and less flat when viewed from space.

https://vimeo.com/252071811

prometheus
01-21-2018, 04:12 PM
Interesting if anyone could pull off anything close to the old Ogo taiki plugin....
www.asahi-net.or.jp/~pq1a-ogs/taiki2.mpg
http://www.asahi-net.or.jp/~pq1a-ogs/taiki4.jpg

erikals
01-21-2018, 05:08 PM
yes,

no time at the moment.

though some of us will keep having a go at this in the near future... :)

MichaelT
01-21-2018, 05:15 PM
;) Gå och lägg er nu. ... och jag med för den delen ;)

prometheus
01-21-2018, 05:25 PM
;) Gå och lägg er nu. ... och jag med för den delen ;)


Jajuste...fast jag sov för länge idag söndag, så jag behöver inte så mycket sömn, och jag börjar 8:50, så snart bedtime.

TRANSLATION...
Yes ... though I slept too long today Sunday, so I do not need much sleep, and I start at 8:50 as soon as bedtime.

And not quite correct in the end..by a google translate, I think I could manage a proper translation myself, but itīs too late now.:D

erikals
01-21-2018, 05:51 PM
vil ikke  :D

...men må nok om ikke så lenge  :)

https://translate.google.no/?hl=no&tab=TT#no/en/vil%20ikke%E2%80%80%20%3AD%0A%0A...men%20m%C3%A5%2 0nok%20om%20ikke%20s%C3%A5%20lenge%20%20%3A)

prometheus
01-21-2018, 11:33 PM
Goood Morning Lightwavers :) Rise and shine..

I had a nightmare yesterday after a crash, be careful to tweak particle birth with the volumetric nodes open, I need to check where and when that happens, it crashed lightwave and has been doing so when I was tweaking emitters on other scenes, itīs also reported by others, and after the crash, i was cleaning up the folders and re-located my plane object, and when I tried to reload the scene, it just crashed...and I found that copying back the object to where it was, made it ok to open again, usually we get a messaged of can not find object file, and not crashes.

I managed to get some textures going, not sure..but I think I got vpr to refresh better, not sure if I had to much running in backgroun processes, anyway..got it going again, the black body radiator is nice, you can set in in many various temperatures, and also balance the looks overall ..like having the illumination scatter more overall to the end of the smoke, or lesser etc, if you tweak scattering scale, or emission scale.

A bit hard to tweak the particle age scale of the new volumetrics, either it isnīt working properly, or I am doing something wrong...still have to work with it some more in order to verify whatīs going on...
the textures look horrible if you do not check world coordinates, with legacy hypervoxels..that was not an issue.
So the scale here isnīt good enough, and One should make smaller flames seperately of course, though I doubt this new volumetrics can do that well, and you just canīt use the old hypervoxels sprite with it either, unless render seperate passes, though that can be the way to go.


http://forums.newtek.com/attachment.php?attachmentid=139654&d=1516602679

139654

Amerelium
01-22-2018, 05:07 AM
vil ikke  :D

...men må nok om ikke så lenge  :)

https://translate.google.no/?hl=no&tab=TT#no/en/vil%20ikke%E2%80%80%20%3AD%0A%0A...men%20m%C3%A5%2 0nok%20om%20ikke%20s%C3%A5%20lenge%20%20%3A)

Åh - flere Skandinaver her ser jeg...

erikals
01-22-2018, 10:07 AM
Åh - flere Skandinaver her ser jeg...
Aha - more Scandinavians i see...

Yes, vært her i 15 år nå... :)
Yes, been here 15 years now... :)

Morgan Nilsson
01-22-2018, 10:46 AM
Härligt med Skandinaviska kamrater här inne ;)

Eng:Nice to see some fellow Scandinavians in here!

MichaelT
01-22-2018, 11:42 AM
:) Yup.

Sensei
01-22-2018, 01:57 PM
If you need random values, but there is no built-in randomization, but you have access to Node Editor,
get Input > World Position,
split them to X, Y and Z, using Vector Scalar,
then multiply by some large primes (or so), each time different one,
add them together, and then use as input to Seed in Random Scalar.

prometheus
01-22-2018, 09:32 PM
Thanks for the tip Sensei.
But it illustrates how the perfect good old hypervoxels workflow was destroyed.
Compare how easy it was and faster
With just a change to a random scale slider.
And I know about getting random scale from -+ variations in the particle
Emitter it self..as I told anti.
The issue remains. .
A system like this ..needs better workflow. .not more complex than we had before.

prometheus
01-23-2018, 01:17 PM
If you need random values, but there is no built-in randomization, but you have access to Node Editor,
get Input > World Position,
split them to X, Y and Z, using Vector Scalar,
then multiply by some large primes (or so), each time different one,
add them together, and then use as input to Seed in Random Scalar.

bumping back..

I Knew I could do it with nodes, but I also new I could do it by using the particle emitters min max variations, that is one way.
As for nodes as I said, I think itīs a showcase of how they destroyed a perfect interface for it as hypervoxels had size variation just under particle scale, such option can not be hard to code can it?

But....if we attempt the more tedious way with nodes, what if instead of the nodes and workflow you suggested...why simply not take the particle id output and feed it to the random scalar seed input, then feed that node to the radius input, isnīt that simpler but still
yielding a good enough random scale of particles?

Sensei
01-23-2018, 01:50 PM
bumping back..

I Knew I could do it with nodes, but I also new I could do it by using the particle emitters min max variations, that is one way.
As for nodes as I said, I think itīs a showcase of how they destroyed a perfect interface for it as hypervoxels had size variation just under particle scale, such option can not be hard to code can it?

But....if we attempt the more tedious way with nodes, what if instead of the nodes and workflow you suggested...why simply not take the particle id output and feed it to the random scalar seed input, then feed that node to the radius input, isnīt that simpler but still
yielding a good enough random scale of particles?


I was giving general advice, not particularly tied to particles,
but if you feed Seed the same particle id, every time you will get the same result.. No matter which frame it is.. Sometimes wanted.. sometimes unwanted.

Same seed = same "random" output.

prometheus
01-23-2018, 01:57 PM
I was giving general advice, not particularly tied to particles,
but if you feed Seed the same particle id, every time you will get the same result.. No matter which frame it is.. Sometimes wanted.. sometimes unwanted.

Same seed = same "random" output.

yes of course, I reflected a bit over that after I wrote my last post, ideally you also want to have control over which particles may have a certain scale, so additional nodes to control that is needed.

prometheus
01-24-2018, 11:41 AM
and a bit more, some node setup schemes for you if you need it, colors can really be changed to how you like it, as well as amount of emissio scale/scattering, so if you want more light penetrating longer away and in the smoke, or shorten it, itīs a combination of setting what temperature you would like, an also balance scattering scale, and emission scale, further on you can put a...

- color tool node in between the black body radiator, and sort of punch enhance brightness or contrast or pull down or raise saturation, apart from initial choosing of temperature of course.
- multiply node to make the relative particle age expand the smoke volume size at the end a little bit more
- invert node is needed to make blackbody show up correctly starting from particle birth.
- tubulent noise texture to texture input is the same as old drag and choose your hypertexture in legacy hypervoxels.

I noted that I can not set the temperature to lower than 1000, which is kind of odd...with turbulenceFD and itīs black body gradient shader, it could be set to whatever low temperature you wanted.
It rendered decently fast now, I will still have to evaluate why it sometimes takes so long to render, and yep..I now about the step size.
This kind of illuminated smoke may be good for space shuttle smoke.

Oh by the way, with some kind of color going on in the volumetrics, and having opengl on here, it looks fairly decent, and the color representation seem to match roughly..so the temperature can be shown, even though I had some changes in the nodes to enhance the original black body, and I got sphereīs roughly the size of what then is rendered, it would however be nice to have a more depth to them as well, just as dynamite had for itīs opengl..which also happened to be better.

http://forums.newtek.com/attachment.php?attachmentid=139706&d=1516819003

Amerelium
01-26-2018, 02:51 AM
...nodes - make my head hurt

prometheus
01-26-2018, 04:08 AM
...nodes - make my head hurt
Nodes I'd by its nature more demanding to understand and learn..For those who do adapt to it well..not much of a problem..They may find it more of their way to work.
But I dare to say there are things in its nature that simply is slower vs more direct access able controls...apart from being harder to learn.

In this case..I think the workflow with particle age was easier with legacy hypervoxel. ..as well as applying textures.

Houdini for instance...is putting efforts in to making the node network have a more direct access with baking them in to tools with special controls rather than adding a node and additional input work as required in such case.

Sensei
01-26-2018, 10:16 AM
Houdini for instance...is putting efforts in to making the node network have a more direct access with baking them in to tools with special controls rather than adding a node and additional input work as required in such case.

The real compound feature should looks like this:
- user has nodal setup in some Node Editor,
- select few nodes
- press "Collapse"
- nodes are copied to inside of Compound,
- the all used by selected range of nodes inputs and outputs are maintained from point of view of the main Node Editor and Compound Node Editor..

prometheus
01-26-2018, 12:14 PM
The real compound feature should looks like this:
- user has nodal setup in some Node Editor,
- select few nodes
- press "Collapse"
- nodes are copied to inside of Compound,
- the all used by selected range of nodes inputs and outputs are maintained from point of view of the main Node Editor and Compound Node Editor..

I donīt get what you mean here, collapse does nothing to copy anything inside of a compound, if you tried to describe what happens, or do you mean "nodes should then be copied inside of a compoud)
I donīt see how that would make anything easier to acess either?
you need a frame work written to be outside of the node editor with controll to select textures, to set a random particle scale (as we could with old hypervoxels), not inside, whatīs the point with that?

Sensei
01-26-2018, 12:25 PM
I am not talking about Edit > Collapse,
but I am describing new feature "how composite nodes should work"..
User selects some nodes, presses "Collapse to Compound", they're gone from Node Editor, and moved to Compound node, they're removed from original, and Compound remains instead.. with the all connections set up..
Ultra fast way to clean up complex node setups.

prometheus
01-26-2018, 12:27 PM
I am not talking about Edit > Collapse,
but I am describing new feature "how composite nodes should work"..
User selects some nodes, presses "Collapse to Compound", they're gone from Node Editor, and moved to Compound node, they're removed from original, and Compound remains instead.. with the all connections set up..
Ultra fast way to clean up complex node setups.

Ok, Understood.

prometheus
01-26-2018, 12:34 PM
yet another node setup,here I added the curve node inbetween invert and black body radiator nodes, it will help tweaking the fireshader more manually as you want, by raising the curve to make a sharper intenser cutoff, or by dragging the curve to the side...pushing the illumination more smoother, but deeper/longer on the whole particle simulation.

I do need to record this though, if I can find time and squeeze that in tomorrow or sunday.

http://forums.newtek.com/attachment.php?attachmentid=139774&d=1516995185

Amerelium
01-30-2018, 10:15 AM
Nodes I'd by its nature more demanding to understand and learn..For those who do adapt to it well..not much of a problem..They may find it more of their way to work.
But I dare to say there are things in its nature that simply is slower vs more direct access able controls...apart from being harder to learn.

In this case..I think the workflow with particle age was easier with legacy hypervoxel. ..as well as applying textures.

Houdini for instance...is putting efforts in to making the node network have a more direct access with baking them in to tools with special controls rather than adding a node and additional input work as required in such case.

Hey, I know - this insane thread shows what one can do once adept (expertly so)

http://forums.newtek.com/showthread.php?137719-Human-Progress/page12