Page 1 of 3 123 LastLast
Results 1 to 15 of 38

Thread: Hypervoxels / Fluids / Physical sky

  1. #1
    Experienced User matts152's Avatar
    Join Date
    Oct 2006
    Location
    Wilmington, NC
    Posts
    121

    Hypervoxels / Fluids / Physical sky

    OK, i know this is a subject talked to death sometimes, but I need to post my feelings on it anyways. I hardly ever complain about lightwave's development, but in this case it's just too far behind the rest of the industry not to start screaming about it.

    1. Hypervoxels, For the love of god please update them. volume mode needs a way to blend or "smooth" the voxels together, for clouds, cigarette smoke, mist, etc... They also need to react to light in a more realistic way, less "blobby". Surface voxels just look horrible almost 99% of the time for almost every effect you would try to use them for. Yes you can make rocks with them, but try water, oil, tooth paste, jello, etc.. just not possible.

    2. Fluids, Lightwave needs a fluid sim, buy Turbulence, write your own, what ever works, just needs a real fluid sim as part of the package. This could probably be combined with the HV update.

    3. A real pysical sky model similar to maxwell render, mental ray, etc.. SKytracer 3 might be neat Would like to see more control over HDR range in the sky model, so the sun reflects very brightly in reflective surfaces. right now you have to fake it to a degree.

    Ok doke, I'm done.
    Where I work www.provismedia.com

    Personal Site. www.stewart3d.com

  2. #2
    Registered User stevenpalomino's Avatar
    Join Date
    Jan 2010
    Location
    Kansas City
    Posts
    607
    +1

  3. #3
    Registered User
    Join Date
    Dec 2008
    Location
    Ghana West Africa
    Posts
    862
    +1

  4. #4
    Robert Ireland bobakabob's Avatar
    Join Date
    Feb 2003
    Location
    Nottingham UK
    Posts
    2,757
    Hypervoxels in Lightwave are good but have been terribly neglected. It would be nice to see some updates.
    Art and photography Flickr site
    Animation, facial rigging, modelling ArtStation
    Assorted animations Robert Ireland

  5. #5
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    16,251
    +1

    Ivé been ranting about those 3 topics for years now..And I would set priority order exactly as mentioned...hypervoxels, volume mode blending needs improvement(tension between null for example or two point/particles), hvīs needs a distance between particle gradient in all channels possible.(thatīs been requested for years from some guys)
    Newtek could implement a third Item volume option..such as modo just have done.

    number two...a decent fluid simulator.

    and finally..work something out for a realistic/physical sky model..and include volumetric in there...something like ogo taiki but faster and easier to work with.

    I guess this is in the order of 30th..request ranting of the same requestīs...maybe

    Michael

  6. #6
    Registered User SonicN2O's Avatar
    Join Date
    Aug 2010
    Location
    Chicago
    Posts
    485
    +1

  7. #7
    hypervoxels upgraded would be nice, but how far can it be pushed and at what cost?

    fluids, i'd like it, but i believe it would take some time to develop, maybe a plugin could be written instead.
    LW vidz   DPont donate   LightWiki   RHiggit   IKBooster   My vidz

  8. #8

  9. #9
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    16,251
    I would state that sunsky isnīt really a physicly correct sky model ..merely a procedural sky model, vue has a pretty decent physical spectral mode ( just poor cloud fractal models)l and so does terragen..but terragen has better cloud functions.

    sunsky works great for many stuff and for exterier arch viz, but I donīt think one should talk about physical sky models if it lacks atmospherics with fog, decay, air scattering and godrays etc.

    Michael

  10. #10
    Member
    Join Date
    May 2006
    Location
    France
    Posts
    4,138
    Quote Originally Posted by prometheus View Post
    I would state that sunsky isnīt really a physicly correct sky model ..merely a procedural sky model...
    Sunsky is a spectral physical model,
    based on atmospheric measures and sampling made by scientist,
    including scattering effects, you can read the paper for details,
    same algorithm is used in the physical LuxRender engine,
    but of course Lightwave is a biased engine
    so direct or undirect illuminations are different,
    not a question for a 'simple' feature request I think.

    About Clouds, Fog and Light Volumetric,
    that's not a question about Sky but about Weather,
    not accessible with a pure physical approach
    for a convincing effect in final, especially for clouds,
    main algorithms are analogic and approximated process
    still using the old raymarching process,
    in this matter, more mathematic than physic for the shape
    same thing for the quality of the illumination.

    Vue is good for a global atmospheric look,
    but Terragen 2 is better for its clouds,
    as I can see the best results are made by artists,
    experimenting and tweaking nodal trees
    with more aesthetic than physic consideration.

    All kind of CG computations are called 'procedural'.

    Denis.

  11. #11
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    16,251
    Quote Originally Posted by dpont View Post
    Sunsky is a spectral physical model,
    based on atmospheric measures and sampling made by scientist,
    including scattering effects, you can read the paper for details,
    same algorithm is used in the physical LuxRender engine,
    but of course Lightwave is a biased engine
    so direct or undirect illuminations are different,
    not a question for a 'simple' feature request I think.

    About Clouds, Fog and Light Volumetric,
    that's not a question about Sky but about Weather,
    not accessible with a pure physical approach
    for a convincing effect in final, especially for clouds,
    main algorithms are analogic and approximated process
    still using the old raymarching process,
    in this matter, more mathematic than physic for the shape
    same thing for the quality of the illumination.

    Vue is good for a global atmospheric look,
    but Terragen 2 is better for its clouds,
    as I can see the best results are made by artists,
    experimenting and tweaking nodal trees
    with more aesthetic than physic consideration.

    All kind of CG computations are called 'procedural'.

    Denis.
    Understand...but I do not quite agree if we should really talk about differing the sky vs weather, even though you say it is supposed to be scientificaly correct, it doesnīt yield the same natural feel when moving the sun disc down to horizon as if you compare to setting the sun in vue, and how nicely it spreads the light and all air anisotrophy scattering, Again ..that might have to do with weather parts like dust, humidity and such.


    But I guess that is what we argue about ..sky and weather phenomena apart, anyway..I suppose we in such case just need the weather algoritms with dust,fog, etc...I love the way vue responds to sky decay and fog etc.

    do not know the technical facts about the sky environment, so yes I might be talking about something I do not know much about, I just do not see equally realistic sky behavior in combination with the sun.

    spectral you say, how much different might vueīs spectral engine be compared to sky environment? as I see it..the vue spectral mode translates through the whole world affecting depth and clouds, and the sky environment seems merely as background environmental, but if not..I would gladely see it improved more to be fully volumetric.

    We do of course have ozone, but I think it sucks compared to vue due to lack of UI control and cloud density functions missing.
    Ogo taiki would be nice, still slow to render (especially fog layers for godrays) and setting up air properties is a pain in the ..compared to vue, but it has a very nice look and feel to it ..if you start with scene samples, and the sun behaves similar to vue when int comes to the results of air scattering and light behavior from a sun.

    Im about to post some volumetric studies with volumetric lights, your sunsky plugin, and hypervoxels, I having difficulties
    combining them though for a good results, a lot of tweakings need to be done to get anyone near results from vue, and thatīs just for the sky look and volumetrics.

    and agreed about vue cloudīs against terragen, terragen is a winner here, but man is terragen slow..


    will post in another thread soon about my volumetric studies.

    Michael

  12. #12
    Member
    Join Date
    May 2006
    Location
    France
    Posts
    4,138
    Quote Originally Posted by prometheus View Post
    Understand...but I do not quite agree if we should really talk about differing the sky vs weather, even though you say it is supposed to be scientificaly correct, it doesnīt yield the same natural feel when moving the sun disc down to horizon as if you compare to setting the sun in vue, and how nicely it spreads the light and all air anisotrophy scattering, Again ..that might have to do with weather parts like dust, humidity and such..
    Sunsky includes Aerosol effect (Water and Dust)
    Mixed gazes, water vapor, rayleigh scattering..

    When converting to rgb you may get different colors
    depends of the system used for conversion,
    we can also alterate the algorithm, but then
    we are far from the physical base.

    ...spectral you say, how much different might vueīs spectral engine be compared to sky environment? as I see it..the vue spectral mode translates through the whole world affecting depth and clouds, and the sky environment seems merely as background environmental, but if not..I would gladely see it improved more to be fully volumetric...
    ..Sky Spectral radiance in wavelength range.

    Not just an environment, as you know, lighting also
    in concurrence with backgroung radiosity,
    so various ways to express Sun and Sky illumination.
    I would not disagree about your frustration,
    we could say that it is too 'neutral'.

    What do we now about Vue? Is it physical?
    Analogic I think, so more technical drived by aesthetic,
    vue is a good synthetic program when it shows sky + ecosystem,
    it is good, hypnotic, but not sure with just Ozone.

    Volumetric is another matter which can get benefit from any
    lighting system.

    Volumetric is implicitly tied to a 3D layered texturing system
    with its own process for scattering lighting, far from the
    generation of real micro water droplets under pressure.

    Turbulence is good for gazeous,
    dynamic effects but is it frequently used for clouds?
    (I mean for realistic clouds not scifi)
    may be not optimized for that or just not adapted?


    Denis.

  13. #13
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    16,251
    Quote Originally Posted by dpont View Post

    Turbulence is good for gazeous,
    dynamic effects but is it frequently used for clouds?
    (I mean for realistic clouds not scifi)
    may be not optimized for that or just not adapted?


    Denis.
    some interesting thoughts and questions there, about Turbulence and clouds, well..I believe the possibilities are there to
    create realistic clouds, but that is scene dependent and more close up clouds in sections, you simple donīt simulate a 5000 meter cloud plane and get away with it memory vise and in proper time...and forget about infinite cloud plane with that.

    I just think no one has worked on it properly to acheive good clouds, itīs mostly a matter of spending a lot of time
    working on it which most people donīt have, I know the potential is there shapevise..it needs proper blending with a sky environment system though, that is what makes it believable.
    So I believe it can be done nicely for a scenetype of fly through shots, but not over large vistas and landscapes.

    so frustrated that ogo taiki just lives there as a plugin that could have been outstanding, and then ..nothing.
    http://www.asahi-net.or.jp/~pq1a-ogs/samples.jpg
    http://www.asahi-net.or.jp/~pq1a-ogs/taiki_e.html

    a big issue with ozone is that you cant tweak the settings with VPR, but you can with ogo taiki,
    and with ozone you are limited to cloud presets since it lacks vueīs cloud function editing, besides the vue
    cloud functions is bases on old noise and fractal function not developed for cloudīs as opposed to terragens fractal clouds.

    Thatīs why ozone fails to my liking and cloud looks arenīt that nice though the other elements are very nice.

    ogo taiki can use any lightwave procedural, like the procedurals from yourself(weather and gardner clouds) and those
    are better fractals than ozone, or vue has.

    I would use ogo taiki if it had better UI, faster final render, and especially volumetric layers like godray.
    the bad part about ogo is setting up the sky properties in a simple way, and not knowing which layer you sometime work on, and also all the quality settings everywhere.

    otherwise I actually believe ogo could surpass ozone realism, simulated at least..mainly because of cloud fractals.
    A shame such good tool isnīt going anywhere.

    Skytracer is very slow if you want good looking clouds, since using textured shadows is almost necessary to get decent
    shadows in clouds, and that is very very slow, and even so ..skytracer doesnīt react volumetric.

    your sunsky plugins are great for many purposes, but I am a cloudyholic and would like to work with such mattes excluding standard photo or paint mattes.
    would be able to direct a sun and itīs sunlight to cast beams through trees and pull it down on horizon to give a nice bleed scattering with smallest efforts.

    This is very hard to acheive nativly in lightwave even with your plugins, maybe the best bet is to wait a couple of years and see if ozone getīs improved, but time is a factor and I should really looks for other ways outside or in conjunction with lightwave to be able to work with such stuff, the problem is I do not wanīt to leave the Lightwave box, and requesting for improvements in this area canīt be a bad thing..as long as you nag about it so often that it someday getīs itīs attention

    hypervoxels with depth shadows, and also geometry based like Modo now has implemented might be a little help, but for large cover we need an infinite texture layer, that reacts to sun and sky properly.



    Michael

  14. #14
    TrueArt Support
    Join Date
    Feb 2003
    Location
    Poland
    Posts
    7,993
    Problem with HyperVoxels is that (the most probable) they're shaded using old fashion single light sample direction function - that's it, just one dot product between average of light sample directions is calculated for every spot.
    New lights such as area, dome, ies, are sending many light samples for one spot. HV is merging the all directions into one then calculating dot product..
    I found it during mine renderer development by observing which ray-tracing functions are called when there is just HV added in scene, that was in LW v9.x times.

    Do test: use area, or dome, them prepare 1-point use it as HV generator, prepare sphere with equal radius. Their look will be slightly different, although radius is the same. Regular sphere will be looking better, smoother.

    Click image for larger version. 

Name:	Snap4.png 
Views:	144 
Size:	252.1 KB 
ID:	110324

    Click image for larger version. 

Name:	Snap3.png 
Views:	123 
Size:	275.7 KB 
ID:	110325

    Click image for larger version. 

Name:	Snap2.png 
Views:	119 
Size:	300.4 KB 
ID:	110326

    Click image for larger version. 

Name:	Snap1.png 
Views:	128 
Size:	175.7 KB 
ID:	110327

    Spheres have same color and parameters. Left one is true geometry, right one is HV.
    Last edited by Sensei; 01-10-2013 at 05:07 AM.

  15. #15
    TrueArt Support
    Join Date
    Feb 2003
    Location
    Poland
    Posts
    7,993
    Scene attached.
    Attached Files Attached Files

Page 1 of 3 123 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
  •