Page 1 of 17 12311 ... LastLast
Results 1 to 15 of 241

Thread: clouds?

  1. #1
    Registered User
    Join Date
    May 2018
    Location
    vancouver
    Posts
    2

    clouds?

    Howdy folks-
    I just parachuted in, to ask a question. Tho info is hard to come by (people tend to get preoccupied with explosions, fire, or fluids if they get interested in particles at all), i suspect that LightWave is the best general 3D app for skies/clouds.
    Am i right? Or should i be looking elsewhere (like Terragen, or???)?
    I have limited time so my priorities are very good clouds and great ease of use/production, not fabulous clouds but very difficult to make (i.e. Houdini).
    Also, could somebody summarize user reaction to Volumetric Primitives vs. Hypervoxels? I've not seen any A:B screenshots comparing them. I know one thing tho: NewTek shot itself in the foot by discarding the catchy name, HyperVoxels, for the generic "volumetric primitives". (But NT seems to shoot itself in the foot a lot!) Even HyperVoxels2 would've been a better label. Anyway.
    Thanking you in advance...
    Canadian guy looking to make fly-thru's for HMD's.

    What did the sadist do to the masochist? Nothing

  2. #2
    'the write stuff' SBowie's Avatar
    Join Date
    Feb 2003
    Location
    The stars at night are big and bright
    Posts
    19,394
    I have nothing helpful to offer re: rendering clouds, howevah ...

    Quote Originally Posted by infinice View Post
    I know one thing tho: NewTek shot itself in the foot by discarding the catchy name, HyperVoxels, for the generic "volumetric primitives".
    "Voluprims"?

    Quote Originally Posted by infinice View Post
    Canadian guy looking to make fly-thru's for HMD's.
    A Vancouverite will certainly know something about what realistic clouds look like! (jk, my eldest brother lives there, and has since the 60's - lovely, lovely place.)
    --
    Regards, Steve
    Forum Moderator
    ("You've got to ask yourself one question ... 'Do I feel lucky?' Well, do ya, spammer?")

  3. #3
    LW 2018 does have the new volumetric primitives, which for clouds, does render a heck of a lot faster than doing them with Hypervoxels. I do still use the sprite mode of Hypervoxels for some things because sometimes you just need a 2d cheat that renders faster. I do wish that the new volumetrics had a sprite mode also.

    With the new volumetrics, you can even do a fly-through the clouds, which with the old hypervoxels, would send your render times through the roof.

    https://youtu.be/sICnY0c12kg

  4. #4
    Registered User
    Join Date
    Jun 2014
    Location
    Right here
    Posts
    1,535
    Prometheus to the rescue ; -) He's the LWMOC (LW Master Of Clouds)

    The answer very much depends on your requirements like rendering time, control, realism, if you do compositing etc.

    In my opinion good looking clouds can be done with LW2018. Other applications offer similar but they are mostly particle based. What you have to consider is the high rendering time using LW2018 to achieve a grain-free render.

    Specialized apps would be Vue (easiest for simple clouds but complex with nodes for more advanced stuff) or Terragen (best cloud quality in my opinion). They both offer high realism but depending on lighting also take time to render. I use both and with Vue xStream you have the best integration (but not compatible with LW2018).

  5. #5
    Registered User
    Join Date
    Jun 2014
    Location
    Right here
    Posts
    1,535
    I think one issue that was mentioned is the limitation on scale, the new LW2018 volumetrics only work in small (and unrealistic) scales. Another problem is the limited usage because you can only use Nulls and the few Geometry shapes. However they seem to blend together nicely. But if you want to art direct your clouds to a specific shape, not possible n LW as far as I know.

  6. #6
    I found Ozone/CloudFactory (E-on Software), which is a pared down version of Vue focusing on clouds/atmospheres, quite good with LW albeit a tad slow. But since the e-on hack last year and the company's lack of transparency on the issue to their existing customers I have decided not to renew any of my maintenance subscriptions (Vue Infinite, Ozone, and Plant Factory) with them. They are pretty much offline now so any communication including license renewals are handled via email.

    In light of the E-on fiasco I've auditioned Terragen Pro and it seems really good and I might go that route since I really like good clouds. Admittedly have not had time (it was a 30 day trial, now expired) to use Terragen in conjunction w/ LW. I've heard Planetside offer a "side-grade" discount, if coming from another package, to buy Terragen which might be a good incentive.

    However, I also have been using, erm...learning, Houdini Indie which is supposedly very good at particles/dynamics/fluids/fx, but have not tried clouds per se yet. Still coming to grips with the learning curve really.
    Humans are pretty much monkeys with car keys.

  7. #7
    Registered User
    Join Date
    Jun 2014
    Location
    Right here
    Posts
    1,535
    Quote Originally Posted by Signal to Noise View Post
    I found Ozone/CloudFactory (E-on Software), which is a pared down version of Vue focusing on clouds/atmospheres, quite good with LW albeit a tad slow. But since the e-on hack last year and the company's lack of transparency on the issue to their existing customers I have decided not to renew any of my maintenance subscriptions (Vue Infinite, Ozone, and Plant Factory) with them. They are pretty much offline now so any communication including license renewals are handled via email.

    In light of the E-on fiasco I've auditioned Terragen Pro and it seems really good and I might go that route since I really like good clouds. Admittedly have not had time (it was a 30 day trial, now expired) to use Terragen in conjunction w/ LW. I've heard Planetside offer a "side-grade" discount, if coming from another package, to buy Terragen which might be a good incentive.

    However, I also have been using, erm...learning, Houdini Indie which is supposedly very good at particles/dynamics/fluids/fx, but have not tried clouds per se yet. Still coming to grips with the learning curve really.
    Yep exactly the same here.

    Actually e-on got hacked about 5-6 years ago and they only found out last December. I even tried to renew my Vue xStream maintenance via Mail but they got confused and sent me my license details instead so I didn't bother again.

    Got Terragen Pro with a decent sidegrade discount too.

    Both tools have their strengths and weaknesses anyway so I use both. Unfortunately both don't support the new LWO3 format or native 2018 integration (and it would actually be up to NewTek but I doubt they will do anything).

    Houdini is also on the radar but not this year because I want to get more advanced with all the other tools I have first and Houdini needs alot of time.

  8. #8
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    15,103
    Quote Originally Posted by Marander View Post
    I think one issue that was mentioned is the limitation on scale, the new LW2018 volumetrics only work in small (and unrealistic) scales. Another problem is the limited usage because you can only use Nulls and the few Geometry shapes. However they seem to blend together nicely. But if you want to art direct your clouds to a specific shape, not possible n LW as far as I know.
    Not quite right, you can use the volumetrics in 2018 on particles, but not point clouds as far as I now..unless some node can extract the info.

    I may have done some cloud stuff before...getting old, hard to remember

    Check this thread specificly dealing with the old volumetrics/hypervoxels and the new volumetrics.
    http://forums.newtek.com/showthread....ght=hypervoxel

    And this thread may give you some hints..
    http://forums.newtek.com/showthread....d-Clouds-Again

    Some cloud experiments here...
    https://plus.google.com/photos/10094...23753291712209



    To sum it up a bit short...best clouds are also a matter of integrating it fully with sky atmospherics, and Lightwave of today do not have that close to realistic physic volumetrics ..like an old plugin that was ogo taiki, if only that developer would have been in contakt with newtek..I think they tried ...and if they would have improved speed and settings, Lightwave would have a system that would be almost close to Terragen, Probably better han vue since the cloud fractals are not that great..and Lightwaveīs procedurals offer quite good stuff.

    Ogo taiki pluign is only 32 bit, it wouldnīt work with 2018 since 2018 now is only 64 bit.
    Sure it may be a dead horse, but there is nothing In Lightwave that comes close to a complete sky realism , as well as planetery render as Ogo Taiki did, the problem was render speed VS quality and also setting it up, too many quality settings as well.


    The very best clouds per Hero item would be fluid simulated clouds, like from the Plugin for lightwave called turbulenceFD or other software which you then also may import to lightwave as VDB in 2018, it does however require either smoothing operations or very high detailed simulated cache files, so it doesnīt get blocky.

    The very best clouds per Global sky scenery, Hands down Terragen...and itīs quite fast compared to many other stuff..though it is still slow in itīs nature..mostly all volumetrics is.

    The new Volumetrics..

    On the plus side..
    + Now the new volumetrics can look awesome, and with the new ligths and global scattering volumetrics, godrays can be created nicely..a bit slow in conjunction, but at least it is better than trying it with old hypervoxels.
    + Shading is more realistic, but it can be tricky to set it right if you are a newbie, but light and scattering offers much more cloud realism in lighting in clouds, though it may be more trickier to bring forth cloud noise detail within th cloud volume.
    + The edge softness of the new volumetrics is now much much better than hypervoxels.
    + Blending is much better than hypervoxels, but I think this is just when you deal with cloning nulls, when you use a particle emitter, the blending is not the same and not as good, the old dynamite blending was better for particles and point clusters.





    On the minus side..
    - The new volumetrics doesnīt work on point clouds...major let down.
    - The new volumetrics Requires you to enter nodes just to set a hypertexture, the workflow has taken a step back, it takes longer time to just get a simple texture in there, unlike old hypervoxels.
    - The new volumetrics doesnīt have a direct random particle scale, unlike old hypervoxels, the new system requires you to enter nodes and set up a network of various nodes to do such simple task, or you do have to switch to the other item, particle emitter and use itīs -+ settings, very awkward.
    - The new volumetrics require you to select itīs item and no other item while tweaking it, since all the UI properties is build in to that Items property window, so you can not move around other stuff and then Use a seperate window to tweak volumetrics, unlike old hypervoxels which had a more direct Overview of deactivating or activating many voxel items in the same window, and that while moving other objects like aeroplane etc, but still having the controls open for you.
    - The new volumetrics are not acceptable at close to real cloud scales when it comes to renderspeed and quality ratio, the scenes provided are at 1 m scale, and that is probably because of this drawback.
    The old Hypervoxels IN 2018 however, does render Much faster than the new volumetrics in 2018 when it comes to larger scale, the old system has fixed levels of quality, and it isnīt dependend on the scale, while the new volumetrics is dependend on scale, where you set your step quality lower for large volumetrics...otherwise they will becom sooo slow to render, but at the same time that means too bad quality.

    And correct, You can not use any geometry polygon shape to shape your volumetric item cloud, unlike some other software, only the use of cubic, spherical or cone shape.
    If you use TurbulenceFD plugin, you can however "fill" a polygon shape on single framesm and add displacement noise, so that is a workaround and you would need to save that to a cachefile and reloadas VDB if you want it to work
    with the new global volumetric scattering system for godrays, since turbulenceFD do not work with this new volumetric system.






    But if used at lower scale without dependencies to other scene elements, and taking some time, Hero Clouds can look nice...godrays with the new global scattering can be done, but it is tricky, you need to have the light quite close behind the clouds(thatīs not physicly correct or near phyicly good settings), and further on add additional lights so the clouds themself do not get to dark and uggly.




    So Improved render speed at high step quality levels, and make it work on point clusters...Wonder what the dev team thought of here? did they simply not have time to implement the new volumetrics full out?
    That is what I suspect..we may see it soon, but they seem to simply had the need to release it as it was for a deadline.
    and put that D...texture choics directly in a drop down list ..next to where it actually should be one, then it will become really really nice.

  9. #9
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    15,103
    Quote Originally Posted by SBowie View Post
    I have nothing helpful to offer re: rendering clouds, howevah ...

    "Voluprims"?

    A Vancouverite will certainly know something about what realistic clouds look like! (jk, my eldest brother lives there, and has since the 60's - lovely, lovely place.)
    I can imagine the cloud scenery over there, and in the us, the vast landscape and valleys, mountains, plains in all that granduer..is a receipe for dramatic weather changes and phenomena not ever seen here in sweden.
    Oh and by the way...havenīt hardly seen any clouds for more than 15 days now..itīs a record may, not since the earliest recorded temperatures in 1739 have we had such warm month in may (Yay..itīs all global warming)
    so itīs been clear sunny sky all these days, no drop of rain...and temperatures between 24-28 + celsius degrees ...Very Very unusal that itīs been like that almost the whole month, we do occasionally get a few days of very high temps ..but not this long time, so early that is high summer, when we usually have a bit chilly spring still.

  10. #10
    www.Digitawn.co.uk rustythe1's Avatar
    Join Date
    Feb 2006
    Location
    england
    Posts
    1,279
    Quote Originally Posted by prometheus View Post
    - The new volumetrics doesnīt work on point clouds...major let down.

    so if you can use particles, just set an emitter on the point cloud to emit the exact number of points on the first frame and set the to emit from verts as the points are counted as verts in pointcloud,
    like this,
    Click image for larger version. 

Name:	Capture11.JPG 
Views:	104 
Size:	278.2 KB 
ID:	141901
    Intel i9 7980xe, Asus Rampage Vi extreme, 2x NVIDIA GTX1070ti, 64GB DDR4 3200 corsair vengeance,
    http://digitawn.co.uk https://www.shapeways.com/shops/digi...ction=Cars&s=0

  11. #11
    Registered User
    Join Date
    May 2012
    Location
    Virginia
    Posts
    428
    I have been working with LW2018 volumetrics recently for the specific purpose of trying to build realistic clouds.

    Here is my opinion;

    1. Ease of use - Extremely easy to use. The best way to approach it (that I have found worked for me personally) is to download the Lightwave content for LW2018. You will find Lightwave scenes of just clouds in it, and those scenes provide an excellent starting point to build your own custom clouds.

    2. Yes, as mentioned earlier, you do have to use nodes for texturing and that may be a downside if you aren't a fan of nodes. I have used them off and on (although I don't consider myself an expert by any stretch of the imagination) so I felt pretty comfortable using them.

    3. At large scales, they do render slowly. I am currently working with a scene where the main cloudbank is around 4km square and on a dual xeon system it takes about 16 minutes or so to render a single frame.

    Now, I do want to caveat that with the fact that some of my render settings are perhaps too much; first, I'm rendering brute force (no interpolation) so I'm taking a performance hit right there. Second, I'm using 125 rays, which is probably overkill.

    However, I can't argue with the results. They are easily the most realistic clouds I have ever created using Lightwave, and setup was fairly straight forward.

  12. #12
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    15,103
    Quote Originally Posted by rustythe1 View Post
    so if you can use particles, just set an emitter on the point cloud to emit the exact number of points on the first frame and set the to emit from verts as the points are counted as verts in pointcloud,
    like this,
    Click image for larger version. 

Name:	Capture11.JPG 
Views:	104 
Size:	278.2 KB 
ID:	141901
    Two things...
    I know that is a workaround, itīs too messy though.
    I wait for them to fix this, or I simply do not upgrade.

  13. #13
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    15,103
    Quote Originally Posted by RPSchmidt View Post
    I have been working with LW2018 volumetrics recently for the specific purpose of trying to build realistic clouds.

    Here is my opinion;

    1. Ease of use - Extremely easy to use. The best way to approach it (that I have found worked for me personally) is to download the Lightwave content for LW2018. You will find Lightwave scenes of just clouds in it, and those scenes provide an excellent starting point to build your own custom clouds.

    2. Yes, as mentioned earlier, you do have to use nodes for texturing and that may be a downside if you aren't a fan of nodes. I have used them off and on (although I don't consider myself an expert by any stretch of the imagination) so I felt pretty comfortable using them.

    3. At large scales, they do render slowly. I am currently working with a scene where the main cloudbank is around 4km square and on a dual xeon system it takes about 16 minutes or so to render a single frame.

    Now, I do want to caveat that with the fact that some of my render settings are perhaps too much; first, I'm rendering brute force (no interpolation) so I'm taking a performance hit right there. Second, I'm using 125 rays, which is probably overkill.

    However, I can't argue with the results. They are easily the most realistic clouds I have ever created using Lightwave, and setup was fairly straight forward.
    1. Hypervoxels is easier to use...Even though I may be biased, I know how to use the new volumetrics anyway..I still think hypervoxels is easier to use, and by no Point Extremely easy to use for a newbie.

    2. Itīs a downside to Everyone..not just for those who is not a fan of nodes, itīs a different matter that nodes may expand on what you can do..but it is by no way faster to acess any kind of hypertexture..Regardless if you know nodes as king of the nodes, I dare to challenge Anyone to set up a volumetric item with voxels HYPERTEXTURES faster than what I can do with old hypervoxels, that means however..no use of any kind of preset.

    3. yes...at large scale they do render slowly..unless raising step size, but that defeats the purpose of having clouds with good quality, AGAIN...think twice or thrice on Why the scene content doesnīt provide scenes at more real cloud scale, and instead use 1 m.

    Final act...Agreed, clouds can be..Is more realistic than before....But, the Caveat is scale denpendency, so you either render only the clouds in small scale only, or render small scale together with downsized items, or wait extremely long time for more real cloud scale.

  14. #14
    www.Digitawn.co.uk rustythe1's Avatar
    Join Date
    Feb 2006
    Location
    england
    Posts
    1,279
    Quote Originally Posted by prometheus View Post
    Two things...
    I know that is a workaround, itīs too messy though.
    I wait for them to fix this, or I simply do not upgrade.
    Why is it a work around? You only need to set 3 settings in the first window, and this method opens up so many possibilities, such as being able to then dynamically effect the point cloud with forces/dynamics etc, create instances and clones very quickly, you could just use the l script commander to make the whole thing 1 click
    Intel i9 7980xe, Asus Rampage Vi extreme, 2x NVIDIA GTX1070ti, 64GB DDR4 3200 corsair vengeance,
    http://digitawn.co.uk https://www.shapeways.com/shops/digi...ction=Cars&s=0

  15. #15
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    15,103
    Quote Originally Posted by rustythe1 View Post
    Why is it a work around? You only need to set 3 settings in the first window, and this method opens up so many possibilities, such as being able to then dynamically effect the point cloud with forces/dynamics etc, create instances and clones very quickly, you could just use the l script commander to make the whole thing 1 click
    Why isnīt it a workaround when you obviously have to do many more steps to acheive something that previously took lesser steps?
    And no..you do not only need to set 3 settings in the first window, you need to add particles etc..

    Quite simply, itīs faster to use only point clouds ...rather than having to add emitter, set nozzle, rate etc...it takes too long time compared to plain point clouds, and additionally calculating.
    Now itīs a completely different matter if you want to expand on it with effects as you say...and if you know how and like it, fine but..itīs still not reducing the fact that it is a workaround.

    And you do not dynamicly directly effect the point cloud, you effect the particle cloud inherited from the vertex emission, but now I am picky I get your point of what you can do with that approach, and it is interesting from VFX point of view if you want expandin on that.

    Destroyed functions needs fixing, not standard techniques of old to compensate in more tedious ways ..thereīs nothing new in 2018 in that regards that you couldnīt do in 2015 with particle vertex emission.
    Might add ...Ivé been a fan of using particles for hypervoxels, but that is not the point.
    And ..there is the issue of random particle scale that is Way more tedious now than with Hypervoxels, either directly use partice -+ or go to nodes and mess with that instead of a direct value slider in hv tab.




    For fly throughs I would recommend ..doing fluid sims and cache files and reload various pre-simmed frames, either in Lightwave or in other software directly or use other software and import to Lightwave and render with OPEN VDB.

    There is something really nice coming up for some other software though that looks awesome for fluid sims of clouds and re-loading..which I wonīt be mentioning here.

Page 1 of 17 12311 ... LastLast

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •