PDA

View Full Version : Lightwave clouds top quality



lightscape
09-08-2014, 08:54 PM
First time I saw lightwave clouds of this quality from baked hyervoxels.
I've seen the hv clouds in the forum but this beats all of them including mine.
The resolution looks high and the lighting is gorgeous.

http://vimeo.com/105414053

Wish newtek had more training and presets for lw 11.

CaptainMarlowe
09-08-2014, 10:46 PM
Yep, noticed that yesterday, it's very good. The author has promised a tutorial. Can't wait for it !

prometheus
09-09-2014, 10:00 AM
I really donīt think the quality is good enough, sure it is baked..but it looses detail quality and volume and shadow depth quality.
Itīs a very good sample based on baked hypervoxels though..I myself havenīt messed with it since the lack of quality before..
not sure it will survive animated process either.

lighting and shading is good ..if everyone just learn to use dp_sunsky sunlight ..it isnīt that hard to set up, sunsky together with hypervoxels light and shading can be very naturalistic and beautiful, and the only thing really needed for the atmospherics to match vue atmospherics, that is a true volumetric layer for fog and volumetric godray lights, that is for the sky atmospherics.
hv particle and voxels or infinite volumetric cloud layers will need to be implemented a little more in order to get it to another level.

I do just point out critics about the baked quality, the rest is nicely done, maybe I think he could have worked more on the cloud shapes/cloud density distribution and itīs procedurals...but itīs well done considering.

JohnMarchant
09-09-2014, 10:12 AM
They are great at a distance but the volume looses it as you get closer. It is however the best attempt i have seen for clouds in HV yet.

prometheus
09-09-2014, 10:28 AM
They are great at a distance but the volume looses it as you get closer. It is however the best attempt i have seen for clouds in HV yet.

yepp...I also donīt think it is usable for animated clouds, thus timelapse or moving clouds or fluctuation changes in clouds canīt be made with baking hypervoxels unfortunatly.
I agree with Ahk salīs request for higher voxel baking 400x400x400 is the highest level as of today, I think it was not long ago they actually raised that, but I think it needs a bit more maybe 2000x2000x2000 would be enough? then it would of course
gather a large gigabite file on the disc for it, and it would also be cool if animation could be included, but then again..how large wouldnīt such a file be?

I donīt think Ahk sal used the local density gradient on the dissolve channel, if he would have..I think the clouds could have looked smoother, but maybe he didnīt want that or it got lost in the baking..I just tested and if done right
the cloud edges can be more softer even when baked.

prometheus
09-09-2014, 10:31 AM
Yep, noticed that yesterday, it's very good. The author has promised a tutorial. Can't wait for it !

thereīs really nothing to wait for in terms of tutorial of the baking process, just bake to the highest resolution you can and then use baked object.
the rest is mostly a process of tweaking the voxels for clouds a lot.

prometheus
09-09-2014, 10:36 AM
If one checks the baked mov files, open in after effecs, you can see that if you have a velocity translate hypervoxel effect to make it fluctate, the animation is baked to the mov file, itīs just that replaying of it doesnīt
take that in to account as I know of and thereīs no option to acheive that.

prometheus
09-09-2014, 12:35 PM
checking the baking a little more, the quality might be somewhat acceptable, not for top level work maybe...depends.
Anyway..what I discovered, using rayleigh in the illumination method seem to bring better detail once the voxels are baked, compared to beer illumination which is softer.

I also noticed that it doesnīt bake in sunsky with sunlight in to the baked model, thus it turns up more white and bright and not as you see it with more naturalistic relation to how the sun is positioned in the sky...not sure if I can correct that, so
that part isnīt good.

except for the voxels not being animatable themself, cloning or instance cloning baked voxels and move around or fly through is extremly fast with just some seconds in rendertime, and it doesnīt matter how close you
get to the baked voxels.

I reckon baking out some different voxel tweakings with position changes and some more tweakings, and reloading those different bakins on to particles or nulls and you can acheive a blending mix that looks more naturalistic
than just use one baked voxel type.

PLEASE...check vpr render times at the bottom status bar, compare non baked with baked voxels.


http://forums.newtek.com/attachment.php?attachmentid=124111&d=1410287599 http://forums.newtek.com/attachment.php?attachmentid=124112&d=1410287618 http://forums.newtek.com/attachment.php?attachmentid=124113&d=1410287636

cagey5
09-09-2014, 12:39 PM
Just a quick note to say it's 'loses' quality etc, not 'looses'... grrr. Carry on.

prometheus
09-09-2014, 12:54 PM
Just a quick note to say it's 'loses' quality etc, not 'looses'... grrr. Carry on.

could be temporary dyslexia, could be my laptop not responding fast enough when I type or accidently add two ooīs..could be sloppyness too.
English is my second language, I am excused.:)

erikals
09-09-2014, 03:26 PM
i would've said looses too... http://erikalstad.com/backup/misc.php_files/aiwebs_029.gif

always wondered how HV's mixed with Turbulence Fluids could work... ain't got time for testing, yet.


what u think Prometheus ?

vonpietro
09-09-2014, 09:07 PM
can't you mix a HV hero cloud that moves independently of the baked clouds or would they be mutually exclusive?
separate of the HV bake

you could render some hero clouds and comp them in afterwards, best of both worlds through compositing.

awesome work, the dragon animation is very good, and i like the sunset lighting

lightscape
09-09-2014, 10:40 PM
It is however the best attempt i have seen for clouds in HV yet.

Yep the best one done with lightwave hv. :thumbsup:

erikals
09-09-2014, 11:32 PM
tested, and got some strange render times...

i noticed this >

VPR lies a bit,
when rendering a baked object VPR would finish at say, 20 seconds
when rendering a regular object VPR would finish at say, 50 seconds

however, the VPR preview turned out to actually be finished way before,
at 15 seconds or so...

so instead of going the "save preview" route, an idea can actually be to use a
(don't laugh) printscreen app that takes a printscreen every say 15 seconds

but i was also a bit surprised with the F9 renders,
sometimes they would be a bit slow, sometimes fast, maybe i missed something... :l

in the last VPR vs F9 i did, F9 was "only" twice as slow
(unless you use "printscreen" for VPR, then VPR "save preview" is actually about the same speed as F9)
(so a printscreen automation would be better)

i had to use regular VPR though, not Draft, as Draft VPR renders polygons of low quality
if there are no polys in the background (or you composite) you could use Draft
(probably saves you another 50% rendertime...)

edit: it might be that regular VPR renders jagged edges too, more testing needed...

----------------------------

subnote,
an interesting thing with the baked voxels though, is that they have more of a glow,
this is because the voxels are actually converted to sprites,
which gives a nice subsurface-scattering-like look...

lightscape
09-10-2014, 12:05 AM
so instead of going the "save preview" route, an idea can actually be to use a
(don't laugh) printscreen app that takes a printscreen every say 15 seconds


Where would you use thee screengrab?
You can save vpr animation previews but has no alpha channel.

GraphXs
09-10-2014, 12:11 AM
*I got it to work with a null, must be doing something wrong with the geo. Was using a flat plane with a displacement....Will try again.

I don't understand how to use the bake hv? Can anyone go over the steps?

The steps I did:

I hit baked them under hv shading tab at the400x400x400 and save as a QT.
It baked/render the HV sequence and after it rendered it checked the "use baked object". But I don't see anything in the render when I hit F9???
I can view the HVQT in the QT player and see the flat image maps.

What am I not understanding?

Thanks, and great video!

erikals
09-10-2014, 12:57 AM
conclusions so far >

http://youtu.be/_s3cXHgxRvM - important, watch in fullscreen 720p...




Where would you use thee screengrab?
You can save vpr animation previews but has no alpha channel.

you'd have to "hack it"

erikals
09-10-2014, 03:13 AM
GraphXs, try to use the default, not QT, could fix it...

erikals
09-10-2014, 03:17 AM
can't you mix a HV hero cloud that moves independently of the baked clouds or would they be mutually exclusive? separate of the HV bake

yes, you can mix them, and no comp needed...

prometheus
09-10-2014, 07:31 AM
i would've said looses too... http://erikalstad.com/backup/misc.php_files/aiwebs_029.gif

always wondered how HV's mixed with Turbulence Fluids could work... ain't got time for testing, yet.





what u think Prometheus ?

blending HV and turbulence for clouds? nah...donīt think so, youīd either go straight on with hypervoxels or with pure turbulenceFD cloud simulations to fill geometry and add subgrid noise to get exactly the type of cloud you want..or simulate It procedurally to a certain state, I donīt think it is a proper way to mix that with hypervoxels it will probpably just look oddly and is hard to match the volumetric shading etc, just maybe..maybe for some underlying whispy cloud base perhaps.


Regarding VPR alpha, in preview options to save out to images, tga 32 is an alternative which has alpha channel baked in.

prometheus
09-10-2014, 07:42 AM
yes, you can mix them, and no comp needed...



Problem with baked hv clouds is no animation and not picking up proper light information in relation the the actual scene light since it is already baked, so for a cloud fly through with some clouds that are more or less still it works ok, but not useful for moving clouds.

moving to position clouds around for general setup can be done however we like to, scaling is done with size tool in layot on the null, and not particle size, but here you go again..it will not adapt the scene lights anymore, so it becomes semi realistic in shading and lighting conditions.

Apart from that..messing with single nulls always are based on a spherical falloff & texture appliance...so either use one single huge hv cloud and let the hypertexture eat in the volume to get the "hero" cloud....I wouldnīt recommend using two nulls anymore since that requires double calculation time over two hv groups, the good thing about baking them is that you can clone a lot of nulls and mix together without that additional time which happens with non baked hv clouds, the bad thing is they are static misses light and shading information from the original hypervoxel and also miss the scene information if you change anything in position etc..
particle or point clusters are probably better to go for since they are within the same hv shading group and therefore a lot of particles donīt have the same calculation time as if it were several null hvīs.

hereīs a lot of experiments with everything from hypervoxels,turbulenceFD, ogo taiki, volumetric lights and hv for godrays,hv for planet clouds, hv to fake volumetric grid item etc...some houdini cloudfx try outs too.
https://plus.google.com/photos/100944643113557837045/albums/5978023753291712209

erikals
09-10-2014, 08:01 AM
Regarding VPR alpha, in preview options to save out to images, tga 32 is an alternative which has alpha channel baked in.

yep, but no alpha info is saved... :/

HV and turbulence, i guess i just have to try one of these days, see if it works,
but yes, could very well get tricky... http://erikalstad.com/backup/misc.php_files/smile.gif

-------

agree, baked does provide challenges... and are probably best for static background clouds

prometheus
09-10-2014, 08:12 AM
here might be one of the reasons I dropped looking in further to baking of hvīs
working with a huge particle cluster at a 3 km size and 2000 particles didnīt survive the baking process well, I need to understand why that is, will need to check if it has something to do with the particles and
compare with static point cloud and bake that instead,

124119


and after baking it is screwed up like this...


124120

prometheus
09-10-2014, 08:24 AM
yep, but no alpha info is saved... :/



ughh..thanks for the heads up, that is correct.

prometheus
09-10-2014, 01:29 PM
Ahh..
Ahk Sal and I had some quick conversation on vimeo, discussing and researching how baking is working, and maybe could be improved...I will await for his tutorial and what he thinks is going on when you try to bake to large scale and the loss of quality.

It has triggered a little more of my curiousity on some hv stuff though, I wonder what can be pushed with the post processing of the baked slices in photoshop or after effects, quite easy to bake a simple volume without hypertextures and then open the mov file in after effects and apply turbulent distortion etc.

Now that applied to the whole sequences of slices will just go through one axis, so it will only look good from initial camera point of view once you export out from after effects and use that mov as the baked hv object, if we could pull such processing off withall the slices in proper axis, that would be cool.
It may be possible to correct and fix softer hv baked cloud edges with the help of using after effects to blur or soften the slices up.

xevious2501
09-10-2014, 03:02 PM
Hey guys, This is Ahk Sal.. (xevious2501) (didnt want to use my full name on vimeo ) not yet anyways. First.. thanks for all the compliments about the Animation tests. Again.. im sorry for the delay with the tut, we are moving into out new Post Facility so we're all chickens running around with our heads cutoff. internet and power is up and down as the construction crew work, so ill try to do it tonight. Or at least answer some questions, but in the meantime ive attached several scene files ready to bake. Most things will be straight forward and norm but a few may be not.

Surly enough as others have mentioned they're not flawless. Bakes are a cheat after all but a necessary one if you dont have access to a crazy render farm. Unbaked voxel still do look better but a bakes solution enables creating a library of baked clouds ready to go with any project. as the the flaws... So increased baking resolutions would definitely help the process ; the drawback would be a larger file as noted. Ive been outputing uncompressed avi. the loads take a sec or two and the files do get big but it doesn't effect LW at all when imported. I think a solution to that is taking the file and compressing it in a post app like AE into a smaller size file hopefully retaining the quality. LW needs to be able to import MOV (as it once did), h246, and certainly a few vector file image and anim formats like svg ai. the Idea of a vector image output is something i though of as well, a bake that goes out as a vector would never loose it resolution when being scaled up or down as it does now due to the 400x400 max size.

As i said earlier size is a major issue when it comes to getting clean bakes. currently with only a 400x400 resolution max and x400 cross scans to sample the voxel, its all pretty tricky to get it rite.
Also is the issue of the blooming or gain affect, when the lit area of the cloud multiply either upon itself of another cloud. Ultimately others have made great clouds via VPR, but unless im just utterly clueless vpr doesnt output passes, nor alpha which makes it kinda useless for production renders. and if you have to do a full alpha only render then your back to square one. Actually what LW really needs is Openvdb which is in houdini. its great and fast but not baked fast, and does not offer a baking feature (at least i haven't found any). Openvdb in lw with the option to bake would be killer. As many of you know it was used for HHTYD 1&2, puss and boots etc etc. anyways gota run *clients*. ill post up as soon as i can.

oh.. Remember Do NOT output anything other than avi uncompressed. its very important.

- - - Updated - - -

having some trouble uploading the rar file. mabie a size limit.. 21mb anyone knows?
Ahk

prometheus
09-10-2014, 03:09 PM
if you click on the attachment button here in your post options, thereīs a little icon "?"
click on that and you can see what type of files and the size you can post..
otherwise I have this copied for you...where 7z is the file that is the largest type of 28.61mb, generally zip is good enough, but only up too 19.07 mb.

I think you should be able to use jzip or something to pack it to 7z format, or use dropbox service..free but register with mail.
https://www.dropbox.com/

Filetype Max File-size Max Width Max Height
7z 28.61 MB - -
bmp 5.00 MB - -
dae 5.00 MB - -
doc 5.00 MB - -
fbx 5.00 MB - -
gif 5.00 MB - -
jpe 5.00 MB - -
jpeg 5.00 MB - -
jpg 5.00 MB - -
lwo 5.00 MB - -
lws 5.00 MB - -
mov - - -
pdf 47.92 MB - -
png 5.00 MB - -
psd 5.00 MB - -
rar 5.00 MB - -
swf 5.00 MB - -
txt 5.00 MB - -
zip 19.07 MB - -

xevious2501
09-10-2014, 04:25 PM
Ok so ive Dl'd dropbox.
here's the link to the file.

[Link removed by moderator]

prometheus
09-10-2014, 04:58 PM
[QUOTE=xevious2501;1398641]Ok so ive Dl'd dropbox.
here's the link to the file.

[Link removed by moderator]]

just checked it, about the content..I was eager to check it without reading the read me notes...sorry, so it couldnīt locate the baked voxel,
we do have to unpack it...for those wondering, the baked voxels are in the zipfile final bakes...I notice that the first scene is looking for Ogl puffy clouds or something, which isnīt in the folder...but I just loaded a clip and checked.
it also requests for dp floodlight which I dontīt have installed, guess I should.

Anyway..for the baking, interesting to see the settings of 200 opacity and hardly no luminosity, I guess you experimented forth that finding, and based on how to reduce the baked brightness effect, good to know.
I can also see you are using textured shadows, is that really necessary...using another illumination method and lower shadow strenght can often acheive very similar shading without textured shadows.

I also see you do have constant illumination, when I baked a cloud based on that...it gave me not the same detail as when I used rayleigh...it was way too smooth.
to soften up clouds I often use a local density gradient in the dissolve channel, so at the edges it will be softer...I wonder if you should not try that?

Turbulent noise and gardner clouds fractals are my favourites for billowing clouds, other clouds with more feathering and flater volumes I prefer the basic turbulence fractal.

You can check my fiddlings with voxels, turbulenceFD, ogo taiki, and houdini cloudfx here...
https://plus.google.com/photos/100944643113557837045/albums/5978023753291712209

Thanks for sharing, Iīll try to do my best to check and research over here and jump in with feedback too.



Michael

xevious2501
09-10-2014, 05:04 PM
Just letme know what missing and ill pack. Dpkit is easy enough to retrieve but i can save everyone the greef of searching. And ill check the Ogl thing.??? Dunno how LW will respond to locating the Final Bakes folder on another computer, i had the folder on my desktop.

another reason bakes should be loaded thru the Image Editor.

And again BE WARNED those bakes were done uncompressed so it unpacks to a 3 gig file.
LW takes a few seconds to load it, but once loaded it operates just fine.

For certain, compressing the files is a must, just as long as you can keep the content clean. i guess its no issue at 400x400.
my concerns were banding.

read the 'Please note!!'

xevious2501
09-10-2014, 05:08 PM
Heres DP plugs sunsky, and Rman collection.

[Link removed by moderator]


needless to say, this collection should be native to LW. so valuable.

Sorry guys i just pulled the plugs...
Though the plugs are free, i dont wanna tick anyone off. so please hit DP's site for them..

prometheus
09-10-2014, 05:09 PM
Just letme know what missing and ill pack. Dpkit is easy enough to retrieve but i can save everyone the greef of searching. And ill check the Ogl thing.??? Dunno how LW will respond to locating the Final Bakes folder. again. another reason bakes should be loaded thru the Image Editor.


Thatīs ok, donīt think you should pack any of his stuff in your packages, donīt think we are allowed to, just make a note of it in read me files,or when you post the file that external plugins is needed, dpsunsky and flood light etc.
I know where to go to find dp stuff so thatīs no problem for me.

- - - Updated - - -

[QUOTE=xevious2501;1398647]Heres DP plugs sunsky, and Rman collection.

[Link removed by moderator]

donīt think you should pack them, not sure maybe denis is ok with it..might ask him.

xevious2501
09-10-2014, 05:36 PM
Yes, i havent checked every feature yet, your scenes are awesome particularly the backlit clouds further down the page. Are they baked as well? I do see most of your scenes the clouds are at a distance, my concern is always close up clouds. flying thru. thats when rendering takes a crash dive. and im running off of a 3dBoxx workstation w8980 dual e5-2687w xeons. We recently had a client of a local small volume airline. They were interesting in the typical airplane flying through clouds and off into the horizon. simple. So i first though, Vue or Ozone. Great clouds, and not the most terrible render times until you go full HD or higher and consider flying thru them. Then its, renderfarm or no go. So i looked into Terragen. same thing.. crazy realism but crazy render times. and least we forget, its not simply a one shot render. ya gota consider revisions. now we're talking either weeks render or using a render service online. well with crazy clients you just might get burned because of the render times. so i wondered, If LW could get me at least 60% of the quality, i would get the rest in post via AE or FLame. speaking of Flame, theres a few cloud GLSL shaders that are just killer at Shadermonkey.com anyways. LW bakes began to hold up fairly well, just as long as you stray from bad bake clouds.

- - - Updated - - -

yeah though so.. thats why i didnt do it at first. ill see if i can kill the link.

As for everyone else its an easy find.

vonpietro
09-10-2014, 05:36 PM
wow, theres some nice work with all those cloud images and the rock in the middle of the water, very nice.

the first picture in that google list shows a particle cloud thats very colorful, is that the high point cloud hv image you mentioned awhile back. Its really a nice abstract art piece =)
the colors are really vibrant too. Nice.

I recalled larry shultz telling me to subtract HV's using a point cloud of hv's. the simplest being a single hv that is being subtracted from another. Sculpting an HV !!! Was a good tip.
do you know if you can subtract hvs when baked? or before baking?

prometheus
09-10-2014, 06:02 PM
Thanks Xevious for bringing production insight, and what is a struggle for everyone it seems, if you can not afford renderfarms, pixelplow is appearantly official terragen renderfarm at what is supposed to be lowcost, but I donīt know more than that.
terragen I just couldnīt stand working in and tweaking cloudscapes, itīs previewer for clouds is still extremly slow even though they mentioned it should have been improved, it just so slow compared to vue or working with hypervoxels.

Though terragen has the best clouds fractals hands down, maybe helios plugin from dax phandi is matching that now?


von pietro...the first picture ..this?

https://lh3.googleusercontent.com/--qsAKoosIb4/UvaC7YJ7PhI/AAAAAAAACP0/sfACJOh_bgc/w1280-h720-no/3%2Bmil%2Bturbulence%2B%26%2Bcolor%2Bcorrection%2B and%2Bglow%2Bcopy.jpg

https://plus.google.com/photos/100944643113557837045/albums/5978023753291712209/5978109511988887058?pid=5978109511988887058&oid=100944643113557837045



that is a 3 million particle field with fractal texture in the particles velocity vector, and its image processed in after effects with color correction and some turbulent distortion,
and some careful glow setting to enhance the light vibrancy, but the image is derived from this animation...


https://www.youtube.com/watch?v=OYpvZyikRqY&list=UUDxRvcWi0V7RgW69O5Ax5Og

vonpietro
09-10-2014, 06:13 PM
YES, thats pretty interesting
i dont understand the velocity vector thing, where in lw is that? in the particle panel?

prometheus
09-10-2014, 06:50 PM
YES, thats pretty interesting
i dont understand the velocity vector thing, where in lw is that? in the particle panel?


Weird..this thread I donīt get replies on ..even though I specificly selected this thread, I hope this isnīt start ongoing again like before when got nothing at all from any thread.

as for the topic, a screenshot on where the settings are, the particle emmitter/motion tab and in the velocity (ms) x,y,z) doesnīt matter which axis you choose, it will apply on them all...
birth rate 25000/frame and particle limit 25000 and here it only a simple emitter with 3m in x,3 m in y, and flat..that is 0 in z axis.

but you will need around 3 millions for matching my image, and apply tiny tiny sprites without any hypertextures, if you use lw 64 bit, you could with some paticence even reach 8 millions of particles, but for 32 bit lw..it will crash probably above 3.5..depends on your ram memory too I guess.
the motion..from the textures, can be strenghten by raising the velocity value (not velocity (ms)) or you can set a higher texture layer opacity.
the motion I did for the animation sample was several textures mixing and distorting.

If you want the field to start with distortions at once and not start as a uniformly filled field, you can set birth rate to be fixed at start and use a -50 value for instance, but you need to increase particle life time with a +50 to your original particle life.

124159

lightscape
09-10-2014, 07:57 PM
If LW could get me at least 60% of the quality, i would get the rest in post via AE or FLame.

Looking forward to the tutorial. I'd say the quality is the best I've seen for lightwave hv.
Lighting, mood, overall output is excellent.
Its applicable to most projects involving clouds and aerial showcase for clients. Not that your clients are my clients and vice versa:D. But appreciate you sharing your technique.

vonpietro
09-10-2014, 10:52 PM
thanks for the reply prometheus, i'll give that a try one of these days, i really dont like HV. They had something called steamer before HV and it was great.

I'm curious how he got smooth looking clouds as opposed to the hv blotchy look, so a tutorial would be awesome

prometheus
09-10-2014, 11:43 PM
thanks for the reply prometheus, i'll give that a try one of these days, i really dont like HV. They had something called steamer before HV and it was great.

I'm curious how he got smooth looking clouds as opposed to the hv blotchy look, so a tutorial would be awesome

yes they have a lot to recover, good stuff as the doubler or steamer but in new better implemented tools, they also need to get a better blending mode ala dynamite plugin or just as how modoīs voxels blends..which is better.. and along with that distance between particle gradients and a volumetric item for complete volumetric conversion of any poly object.

regarding smoothness, he could have gotten even smoother look if he would have used local density on the dissolve channel, the smoothness from baking is probably because it creates slices, and thus the end volumetric clipping isnīt so thick and hard perhaps, it becomes more like sprites and those can have a smoother look, but as mentioned, if you use local density on the dissolve channel and have just the right thickness setting, it will look decent.
we could use a better way of thickness control on hv texture edges, as it is now, most gradients do not work in the thickness channel, the falloff values are not working either...I would suggest fixing the gradients and falloff, and also implement a density or thickness curve that we can adjust thickness on the edges of the volumeīs.

Then other stuff like the actual hypertextures themself ..if they could rewrite as many hypertextures as possible and introduce an offset or falloff within the texture itself, so it will not yield sharp circular cutoffs in a single voxel, the only way to avoid those uggly cutoffs, that is to choose a decent hypertexture and also have low density or let the hypertexture eat in to the volume quite a lot to avoid that.

if you were to zoom up the preview texture from the hypervoxels tab...take a print screen and zoom in with photoshop and study the turbulence texture, you can see a not so nice circular pattern within how the fractal is built.
not all have that pattern, some have other not so nice build up patterns.

dpont
09-11-2014, 01:22 AM
Heres DP plugs sunsky, and Rman collection.

[Link removed by moderator]


needless to say, this collection should be native to LW. so valuable.

Sorry guys i just pulled the plugs...
Though the plugs are free, i dont wanna tick anyone off. so please hit DP's site for them..

The readme files of my plugins have an explicit legal stuff
no permission for public redistribution,
I suppose that you lost the possibility of editing the post yourself
so I tried to contact a moderator in this forum,
if you didn't already, I would like also that you remove it from Dropbox.


Denis.

jwiede
09-11-2014, 01:24 AM
if you were to zoom up the preview texture from the hypervoxels tab...take a print screen and zoom in with photoshop and study the turbulence texture, you can see a not so nice circular pattern within how the fractal is built.
not all have that pattern, some have other not so nice build up patterns.

Well, it definitely doesn't help that none of the hypertextures have been significantly updated/changed since inception. Computers have grown vastly more powerful, yet we're still using procedural textures designed for a time when the engine faced single-core constraints in compute power available, etc. Most systems should be capable of realtime-ish generation of much denser/more complex hypertextures and other procedurals these days while retaining quite acceptable render / preview / voxelization times.

Ztreem
09-11-2014, 01:30 AM
Yep the best one done with lightwave hv. :thumbsup:

I think the Gorillaz video for Feel Good Inc is one of the best hypervoxels clouds I seen(and it's quite old).

prometheus
09-11-2014, 04:07 AM
I think the Gorillaz video for Feel Good Inc is one of the best hypervoxels clouds I seen(and it's quite old).


meeh....first time I saw it...looks good, but we donīt see much of it and it is very post processed it seems, not that postprocessing is a bad thing or not accepted...but that look is expected when
post processing.

lightscape
09-11-2014, 07:48 AM
I think the Gorillaz video for Feel Good Inc is one of the best hypervoxels clouds I seen(and it's quite old).

Yeah too much post and the clouds are more cartoony than realistic.

xevious2501
09-11-2014, 05:48 PM
@Dpont. I pulled the plugs off dropbox minutes after posting as Prometheus suggested it probably not a good idea which i agreed. People can find the the plugs themselves on your site, i was trying to make a bit easier for everyone sorry.

They're great btw. ive used them for some time but only recently dug in a bit deeper. Tuts from other users.
Now that i have your attention, i have a question about one.. The SVG importer.
I was wondering if that plugin could be compiled for 64bit use. Currently its only 32bit.

I have idea about LW using a vector based file format which 'in theory' should be non destructive upon scaling.
If thats true, and lighwave had the ability to import and export svg, using svg in conjunction with hypervoxels would be a great asset.
baking with vectors or svg should prevent pixelization upon re sizing voxels, as is the case with raster/bitmap images.

just wanted to run i by ya.

And again, ment no disrespect with your plugs, they're a god-send.

lightscape
10-14-2014, 04:25 AM
Any progress on the tutorial? Will newtek improve hypervoxel from you request?
http://forums.newtek.com/showthread.php?136612-LW-11-7-wishlist

Greenlaw
10-23-2014, 05:00 PM
The other day I finally got around to playing around with baked hypervoxels myself. Jeez, I don't know what took me so long...in my first test, the unbaked HV was taking about 20 minutes to render and the baked HV was done in seconds! That's a jaw-dropping difference in render time. At work we just happen to have a movie in production with lot of flight scenes, so now I'm trying to come up with clever (i.e., cheap) ways to use this tool. I thought I'd share some observations. :)

I think the first thing that anybody trying out baked Hypervoxels notices is that there is a dramatic difference between the looks of the original HV and the baked HV. Here are some tips for resolving that:

1. You need to tweak the settings based on the look of the baked Hypervoxel, not the unbaked result. When tweaking, set the Resolution to a lower value, like 100 or 200 pixels. This will still give you a good idea of what the final high res bake will look like but at a fraction of the bake time. It's not the same as tweaking in realtime but it's still a big timesaver. When you get the look you want, then bake the HV at the max setting of 400.

(I hope they allow users to increase this in the future. I'm surprised by how good the baked HV can look at this resolution but obviously it could be even better for production work.)

2. It may be a little puzzling which setting you need to adjust to make your baked HV look like the unbaked version. If you're trying to match the look, start with increasing Thickness, maybe 2X. This seems most effective if you're trying to preserve the volume of thin and whispy clouds.

3. You can't alter the baked Hypervoxel in Lightwave but, because the bake file is an .avi/.mov, you can bring it into After Effects or Fusion and alter them there. I've been using composting tools to add subtle color gradients, and experimenting with other processing. Just for kicks, I found you can change the orientation of the Hypervoxel through post processing too, although realistically you should probably just rotate the original HV and rebake it. (In case you're wondering, the slices in the video file go from front to back so rotating the image rotates the hypervoxel on its z-axis. That's not especially practical information but it's interesting.)

4. Save your hypervoxel settings and clearly mark them as 'bake' settings. This will save you time when creating variations.

It's curious that the baked HVs can cast a shadow but cannot receive light or shadows themselves. I understand why lighting can't change but I don't understand why you can't change color and other image properties in LightWave since I can obviously do just that using another program. I'm guessing this is a bug because the manual states that you should be able to change color and luminance.

One thing I'm finding very interesting is that with a little tweaking, the baked HV can actually look better than the original unbaked HV. Here's an example:

125192

What really surprises me is how well this trick holds up even when you're inside the Hypervoxel. Render times remain sweet--even when fully filling the screen, most of my 1080p render tests are taking only a few seconds to a couple of minutes at most.

I tried saving some diskspace with the bake files by using the Lagarith codec. Lagarith is a free lossless codec and it preserves the alpha channel. For HV bakes, this appeared to work at first but when I tried to switch between different bake files, Lightwave would crash. So, it kinda works but it does make Lightwave less stable. I haven't tried other codecs but, IMO, better to stick with uncompressed .avi/.mov for now.

Final observation: baked hypervoxels may not be perfect, but man, even with the current limitations this feature going to save us a ton of time and work here.

Well, that's all I can think of for now. If I can manage the time, I'll post some more tips.

Much thanks to the OP for starting this thread. I know this isn't a new feature but for me this has been like a surprise 'update'. :)

G.

prometheus
10-23-2014, 05:11 PM
The other day I finally got around to playing around with baked hypervoxels myself. Jeez, I don't know what took me so long...in my first test, the unbaked HV was taking about 20 minutes to render and the baked HV was done in seconds! That's a jaw-dropping difference in render time. At work we just happen to have a movie in production with lot of flight scenes, so now I'm trying to come up with clever (i.e., cheap) ways to use this tool. I thought I'd share some observations. :)

I think the first thing that anybody trying out baked Hypervoxels notices is that there is a dramatic difference between the looks of the original HV and the baked HV. Here are some tips for resolving that:

1. You need to tweak the settings based on the look of baked voxel, not the unbaked result. When tweaking, set the Resolution to a lower value, like 100 or 200 pixels. This will still give you a good idea of what the final high res bake will look like but at a fraction of the bake time. It's not the same as tweaking in realtime but it's still a big timesaver. When you get the look you want, then bake the HV at the max setting of 400.

(I hope they allow users to increase this in the future. I'm surprised by how good the baked HV can look at this resolution but obviously it could be even better for production work.)

2. It may be a little puzzling which setting you need to adjust to make your baked HV look like the unbaked version. If you're trying to match the look, start with increasing Thickness, maybe 2X. This seems most effective if you're trying to get thin and whispy clouds.

3. You can't alter the baked Hypervoxel in Lightwave but, because the bake file just .avi/.mov, you can bring it into After Effects or Fusion and alter them there. I've been using composting tools to add subtle color gradients, and experimenting with other processing. Just for kicks, I found you can change the orientation of the Hypervoxel through post processing too, although realistically you should probably just rotate the original HV and rebake it. (In case you're wondering, the slices in the video file go from front to back so rotating the image rotates the hypervoxel on its z-axis. That's not especially practical information but it's interesting.)

4. Save your hypervoxel settings and clearly mark them as 'bake' settings. This will save you time when creating variations.

It's curious that the baked HVs can cast a shadow but cannot receive light or shadows themselves. I understand why lighting can't change but I don't understand why you can't change color and other image properties in LightWave since I can obviously do just that using another program. I'm guessing this is a bug because the manual states that you should be able to change color and luminance.

One thing I'm finding very interesting is that with a little tweaking, the baked HV can actually look better than the original unbaked HV. Here's an example:

125192

What really surprises me is how well this trick holds up even when you're inside the Hypervoxel. Render times remain sweet--even when fully filling the screen, most of my 1080p render tests are taking only a few seconds to a couple of minutes at most.

I tried saving some diskspace with the bake files by using the Lagarith codec. Lagarith is a free lossless codec and it preserves the alpha channel. For HV bakes, this appeared to work at first but when I tried to switch between different bake files, Lightwave would crash. So, it kinda works but it does make Lightwave less stable. I haven't tried other codecs but, IMO, better to stick with uncompressed .avi/.mov for now.

Final observation: no, baked hypervoxels are not perfect, but man, even with the current limitations, this feature going to save us a ton of time and work here.

Well, that's all I can think of for now. If I can manage the time, I'll post some more tips.

G.



Great..I will look in to this, the statement of baked can look better than original..well I donīt think so, if you want that softness you just need to set it right from original..meaning proper thickness and a good local density gradient in the dissolve channel..plus never use any quality mode under good.

cloud album...
https://plus.google.com/photos/100944643113557837045/albums/5978023753291712209?banner=pwa

Michael

Greenlaw
10-23-2014, 05:29 PM
...the statement of baked can look better than original..well I donīt think so...

I understand what you mean but because the looks between original vs. baked can be so drastically different, I think this is subjective. It really depends on what you're going for and what you think is 'better' in a given situation.

In the above example, I only increased thickness a bit and I thought this made the baked version look much better than the unbaked hypervoxel. That's because my current need is fast rendering whispy 3D cloud elements. But if I needed magical smoke, I might think the original was better. It's totally subjective.

BTW, increasing the thickness setting actually made the original look kinda bad. Thus, my statement: you should tweak your settings based on the look of the baked render, not the unbaked render.

G.

prometheus
10-23-2014, 06:56 PM
BTW, increasing the thickness setting actually made the original look kinda bad. Thus, my statement: you should tweak your settings based on the look of the baked render, not the unbaked render.

G.

fully understandable..I was refering to setting thickness in the proper way in regards to not have it too thick, I generally set it between 15-25 for clouds,and always have the local density gradient in the dissolve channel to smooth things out.
I havenīt checked more in to bakin the stuff really, partly because I think it is to unfriendly when you have to guess what the baked version will look like in regards to starting from original voxels.

Greenlaw
10-23-2014, 10:49 PM
It's a tradeoff to be sure. Where I'm working now, we'll gladly take a few minutes/seconds per frame for 3D clouds over a couple of hours per frame any day. Yes, quality is important to us too but the usual render times for Hypervoxel clouds make them absolutely prohibitive on our schedules. However, the render times I'm seeing in these tests makes baked Hypervoxels a potential game changer for us.

And to be honest, with a little finessing in bake setting and some comp work, I think the quality of baked Hypervoxels can look really good.

G.

lightscape
10-23-2014, 11:15 PM
What's the best way to generate hv cloud based point, mesh?

Greenlaw
10-23-2014, 11:52 PM
Depend on how you want to animate them. At simplest, just make a plane, subdivide it, jitter it, and kill the polygons. If you want to be able to see the points in Layout, convert them to single point polygons. If you want to be able to move the points individually, use nulls or use Edit Points in SoftFX to move them around in Layout. You can also use bones, morphs, whatever. It's even more fun with you use other shapes--about 15 years ago I created a 'hurricane' like vortex using Hypervoxels and kind of a donut shape. It worked out well considering how little time I had to create the effect.

You can also use random points tool in Modeler. After generating the points, you can 'sculpt' the cloud using the magent tool and lasso tool and delete.

The new 3rd Powers lattice is a good way to animate points. You can deform them directly with Lattice or you can make a shape with the lattice tool and push your points through it. I did something like that earlier this year to animate a mass of instanced creatures crawling over a body. You might use the same trick to move HV fog that rolls over mountains.

(BTW, I don't think this trick would work well for baked Hypervoxels--it will likely be at odds with some of the limitations. For example, you can animate their position and scale but not rotation of baked Hypervoxels, so any attempt to guide their movements with deforrmers probably won't look right. Also, since the scene lighting doesn't fully interact with baked Hypervoxels (they cast shadows, but do not receive light or shadows,) you'll have problems there too. If you're going to use baked Hypervoxels, you'll probably want to keep the setup fairly basic.)

prometheus
10-24-2014, 10:03 AM
What's the best way to generate hv cloud based point, mesh?

for my larger scale of clouds, I tend to use large particle emitters as fields, around 12km or more in size to be grand scale, maybe use 2000 and much more in particles depending on what hv size I need.
I also use a procedural texture in the birth rate to distribute the particles in "cloud" forms..otherwise you will just have particles evenly distributed..an important factor is also to use a very large varitation size.

You can also run particle simulations in layout ..prefered to use partigon emitters and then save the simulated state when you think it looks best with save transformed, you can then load it it as static cloud points and alos further tweak the clouds in modeler or with 3dpowers tools in layout if you have that.

as greenlaw mentioned, you can also edit single particles..if you set up an emitter with a few particles and edit them directly with particleFX edit tool, and recomended is to actually have VPR on and use openGL overlay so you can see opengl size of particles while also seeing the actual hypervoxels.

another way to create point clouds in modeler, draw out poly faces with pen tool or shape the face with any other modeling tool, then look for the Hidden command fill solid, that will fill your poly area with points, it will be a little sharp..but you can just jitter it out a little to spread out the points a little...this is based on creating them flat at one level, but you can also simply create spheres or model the cloud shapes and use fill solid, you can not set any amount of points though..so you have to fill it several times if you need more points.

there is also a way of just using a dense subpatch grid, send it to layout, add hypervoxels to the vertices..make the subpatch grid 100% dissolve in itīs render tab, then in hypervoxels tab..use either dissolve channel or density channel to add a texture a shape the cloud density distribution that way.

Hereīs an experimental appliance test with tiny tiny hv sprites(no hypertextures) applied on a dense divided cube, so it is sort of faking a volumetric cube item, then I cut out the shape with different textures.

https://lh5.googleusercontent.com/-j6qkeGp9hFU/UvaHTw6OXvI/AAAAAAAACP0/Phq5RvAYfKY/w899-h879-no/cubic%2Bexperiment.jpg

prometheus
11-05-2014, 12:45 PM
I might reconsider using baked hypervoxels a little bit more....hereīs a test...
used partigon emitter to be able to simulate a particle field with turbulence fractal texture in the velocity channels until I get desired simulated state where the particles have been formed to my liking, I used the partigon emitter(not hv emitter) just to be able to load it back in modeler and I did some additional twisting of the point cluster, reloaded back to layout, and added hypervoxels and baked the voxels at the next highest resolution..and at medium voxel quality, so there is room for much better quality.


I also rendered out at only half HD resolution ( 640x360) ...just for testing, 500 frames total and all rendered within 27 minutes.
then loaded to after effects and a sligth blur on the clouds, and I used two layers of the clouds, where one layer is faked volumetric light ..where I simply used cc fast radial blur, and I Upsized/rescaled the layers 200% before render out to quicktime H.264
Final size of the quicktime mov 8.71mb.

just a simple first test with baked voxels, and there are room for improvements...though they are suitable for fly throughs in fast rendertime ..I am not so sure I will be able to use it for timelapse clouds that also cast shadows on terrain, that I think isnīt doable with baking.


https://www.youtube.com/watch?v=wib86_Rw9s8


Michael

Greenlaw
11-05-2014, 12:58 PM
That's a nice test! :)

My render times have been crazy fast too. Baked may not be as detailed or as flexible as fully rendered HVs but considering the studio I'm working for can't afford the usual HV render times, baked is turning into a great option. As a compositing element for fly-throughs and for distant (but not too distant) clouds, it's fantastic considering the speedy render times.

I hope LW3DG will take a closer look at this feature and gives us some improvements in this area.

G.

prometheus
11-05-2014, 02:24 PM
That's a nice test! :)

My render times have been crazy fast too. Baked may not be as detailed or as flexible as fully rendered HVs but considering the studio I'm working for can't afford the usual HV render times, baked is turning into a great option. As a compositing element for fly-throughs and for distant (but not too distant) clouds, it's fantastic considering the speedy render times.

I hope LW3DG will take a closer look at this feature and gives us some improvements in this area.




G.

thanks for the feedback...some further post processing with turbulence distortions in after effects on clouds only can give a little more realistic cloud edges too,I didnīt do that on these though..but itīs like a secondary advection to the actual hypertextures...in post.

And Indeed so...it is very fast to deal with in terms of rendertimes, I wonder how different baking of hv is from the image3d tech Houdini has for clouds, I have that tutorial somewhere where they showcased a nice mushroom cloud created from l-systems and applied upon metaballs, it had way better control for turblence and the volumetrics adapted to the metaball shape way better than hypervoxels does with particles or points only.

I would recommend the lw group to take a closer look at more improvement on this baking tech too.

But I would rather see them get a that full volumetric item per geometry basis in there, like modo and houdini has, that is probably no1 request, and after that the actual blending mode between two points or particles wich needs to be corrected and improved on.

for baking I would like to see the color getting baked in from dp sunsky light, not sure if I am not doing it right, but it donīt seem to get baked in..have to revisit and test some more though.
And we need higher resolution, I would also like to be able to bake in animation, and bake large particle fields, and finally be able to use baked hv that casts shadows...or find a workaround for that.

I have often used volumetrics in sprite mode and mixing with either hypervoxels sprite or volumetric for Nebula stuff, I think baking of it can work nicely too..did some early test before to get color gradient properly, and it retains that...the downside is
you canīt really continue to tweak and adjust lights with baked stuff, so you loose a little of the magic tweaking there.

I think baking of hypervoxels can produce some nice fog presets too...that you might be able to use nicely in a lake scene for instance as whispy fog/mist over the lake, have yet to try though.
I need to try out baking clouds of fog voxels based on a dense subpatch grid, where the shape/density is produced with textures in the dissolve or density channels so it carves out the whole grid based on those textures...basicly the same as the image I posted just previous the animation clip.

One need to get customed to the way of setting the hvīs right per guessing or per rendered hypervoxel original result, if one do it often...one might get skilled at getting close to final results quite fast.

Michael

prometheus
11-05-2014, 03:06 PM
these clouds I cant go and bake unfortunatly, 46 seconds per frame in vpr, but probably much longer in final render, and fly through I wouldnīt think of with these....I wasnīt completly satisfied with fluctuation and the so called cloud change of rate.
since the particle is only moving in velocity direction and not transforming in shape, the hypertexture speed effect (velocity translate) will only make some movement effect overall while the actuall density shape do not change, and that isnīt looking realistic, so it would require managing of the particles movement in other directions too..but that could also screw it up completly.


https://vimeo.com/92539727

kopperdrake
11-06-2014, 03:44 AM
Being a complete numpty with HV, a tutorial on creating these clouds would be hugely appreciated by many I imagine, free or paid! Those are some gorgeous looking clouds, to my eye, there. Thanks for the thread, I'm sure many are reading it but, like myself, haven't the necessary knowledge to add to it.

scratch33
11-06-2014, 05:40 AM
Being a complete numpty with HV, a tutorial on creating these clouds would be hugely appreciated by many I imagine, free or paid! Those are some gorgeous looking clouds, to my eye, there. Thanks for the thread, I'm sure many are reading it but, like myself, haven't the necessary knowledge to add to it.

+1

tyrot
11-06-2014, 06:44 AM
+1

prometheus
11-06-2014, 07:48 AM
Being a complete numpty with HV, a tutorial on creating these clouds would be hugely appreciated by many I imagine, free or paid! Those are some gorgeous looking clouds, to my eye, there. Thanks for the thread, I'm sure many are reading it but, like myself, haven't the necessary knowledge to add to it.

Did you mean my clouds? or the baked ones from the thread starter?
commercial, that could be an idea...probably a very low cost tutorial in such case, but covering more than clouds maybe, will have to go and invest in a good mic first, will look in to it.

tyrot
11-06-2014, 09:22 AM
prometheus you have lots of amazing quality work at your youtube channel - i guess - you must release a particle video for us...

prometheus
11-06-2014, 09:31 AM
prometheus you have lots of amazing quality work at your youtube channel - i guess - you must release a particle video for us...


have to see about that, I sort of promised to do some stuff before, but some things got in the way.
will try and find that particle sample for you later..I havenīt forgot.

tyrot
11-06-2014, 10:27 AM
no no :) i can wait for a more extensive commercial tutorial set.. we gotta update at least our own skill set - while waiting..

kopperdrake
11-07-2014, 03:15 AM
Did you mean my clouds? or the baked ones from the thread starter?
commercial, that could be an idea...probably a very low cost tutorial in such case, but covering more than clouds maybe, will have to go and invest in a good mic first, will look in to it.

Both were great, but those you did, viewed as if from a road whilst driving, were the type I could use, for architectural scenes. It'd be great if you could find the time, and find a decent mic :)

tcoursey
12-08-2014, 03:45 PM
these clouds I cant go and bake unfortunatly, 46 seconds per frame in vpr, but probably much longer in final render, and fly through I wouldnīt think of with these....I wasnīt completly satisfied with fluctuation and the so called cloud change of rate.
since the particle is only moving in velocity direction and not transforming in shape, the hypertexture speed effect (velocity translate) will only make some movement effect overall while the actuall density shape do not change, and that isnīt looking realistic, so it would require managing of the particles movement in other directions too..but that could also screw it up completly.


https://vimeo.com/92539727

Yes it would be great to have some direction in this area for us ArchViz guys. I'm always struggling with environment maps and would love to get some real 3D clouds....they don't have to move and transform like timelapse ones you show. Even if you couldn't create tutorial would you be willing to share your scene file so we could pick it apart for knowledge? You are too hard on yourself sometimes, your work is very good!

prometheus
12-08-2014, 04:13 PM
Yes it would be great to have some direction in this area for us ArchViz guys. I'm always struggling with environment maps and would love to get some real 3D clouds....they don't have to move and transform like timelapse ones you show. Even if you couldn't create tutorial would you be willing to share your scene file so we could pick it apart for knowledge? You are too hard on yourself sometimes, your work is very good!

Thanks...well I am not up to sharing the scenes... those scenes took long time to work out, but I do share the techniques behind it in descriptions and by answering questions, then it is also a matter of having an eye for it.
hard on my self :) donīt think so, I havenīt reached the results I want to, and get to the level where I myself is pleased, Lightwave doesnīt empower me with such tool to that extent unfortunatly, If they someday starts to listen
to my suggestions on ogo taiki and what can be done to improve hypervoxels...that would give me the boost and empower me to reach that level..but currently Lightwave fail to reach the level I want.

Environment maps are easier and gives more realistic result in many cases for stills, you should combine it with proper lighting and add fake cloud shadows that seems to derive from the clouds, just through in a transparent plane or dome and use procedurals in transparency channel to fake that, the rest is a matter to match the direct sunlight with backdrop clouds and sky by eyeballing I would say.

What is it with environment map you are struggling with? lighting, scale?

Michael

lightscape
01-16-2015, 10:37 AM
Did a tuturial come out for this anywhere? Lightwave newletter, youtube?

prometheus
01-16-2015, 10:40 AM
Did a tuturial come out for this anywhere? Lightwave newletter, youtube?

do you mean baking of the clouds from ahk sal? I donīt think so, mabye there was some intstructions on some thread, canīt recall exactly, otherwise greenlaw shared some tips in previous post in this thread.

I havenīt put anything up for my stuff anyway.

lightscape
03-18-2015, 09:33 PM
do you mean baking of the clouds from ahk sal? I donīt think so


Too bad xevious2501 tutorial would have been top quality clouds in lightwave. :D


Cloud tutorial with TFD

https://vimeo.com/121981472

bobakabob
03-19-2015, 03:21 AM
Being a complete numpty with HV, a tutorial on creating these clouds would be hugely appreciated by many I imagine, free or paid! Those are some gorgeous looking clouds, to my eye, there. Thanks for the thread, I'm sure many are reading it but, like myself, haven't the necessary knowledge to add to it.

Yes, some great work here but I'm missing the basics. Could some kindly Lwaver claify below?

1. What's the best way to setup a Hypervoxel to be baked - with even lighting all round and no shadows?
2. Where exactly are the baking controls? I didn't know they existed! Are they in surface baking camera?
3. What confuses me above all is the file is QT movie file. Why movie and not a JPEG? And what do you do with this file... Project onto a flat polygonal surface?

prometheus
03-19-2015, 07:57 AM
Yes, some great work here but I'm missing the basics. Could some kindly Lwaver claify below?

1. What's the best way to setup a Hypervoxel to be baked - with even lighting all round and no shadows?
2. Where exactly are the baking controls? I didn't know they existed! Are they in surface baking camera?
3. What confuses me above all is the file is QT movie file. Why movie and not a JPEG? And what do you do with this file... Project onto a flat polygonal surface?

1. why would you want even lighting around? that doesnīt occour in real life on the clouds, apart from that..you could simply turn of the hv affected lights in the hypervoxels tab, and use the hypervoxels own luminosity instead, but donīt be fooled to think you can use baked even lit clouds and mix with the scenes light after baking, lights in the scene will not affect clouds that are baked...so it would just look rather dull if it is even lit all around.

2...have to sit in front of my voxels to describe it, wait a two-three hours and Iīll get back:) when I get home.

3..quicktime plays back image sequences in the same file, you canīt do that with jpg, you use the quicktime baked file within the hypervoxels tab, check use baked..and you will be promted with a requester to load desired clip.

bobakabob
03-19-2015, 08:59 AM
Thanks, Michael, I've been using LW since 5.6 and didn't know you could do any of this. Any help with Q 2 would be gratefully received.

lightscape
03-19-2015, 09:43 AM
Yes, some great work here but I'm missing the basics. Could some kindly Lwaver claify below?


Would be great if xevious or mr. rid would make some cloud tutorials.

prometheus
03-19-2015, 01:13 PM
Thanks, Michael, I've been using LW since 5.6 and didn't know you could do any of this. Any help with Q 2 would be gratefully received.

Oops...almost forgot to get back to you as I promised, had to take a walk..thatīs more important than sitting in front of the puter all the time.

Baking options can be found in the hypervoxels shading tab/basic..right under thickness and smoothness.

1.just click on bake object and you will be promted with options to set resolution, I would suggest using the highest possible if you want good quality..and that would be 400x400x400.

2.Set desired filename,and also click on the file name button to browse to where you want to store your file.

3. choose file option..I have used quicktime mov and in that file option setting it to animation codec, but it also is recomended to use avi uncompressed I think, then click ok for the process to start baking.

once the baking is done, hypervoxels will auto disengage the hypervoxels all other settings and go in to use baked mode by checking the "use baked object", you can have vpr on actively all the time to watch the process, if you want to load another baked file(assuming you have baked some others previously)...then just click on options under baked object and you will get the file requester promt where you can select another file if you have made such.

if you want to go back to hypervoxels standard none baked mode and tweak your object some more, just uncheck use baked object, and tweak it to your liking..and then rebake a new one.

it is unfortunatly a little hard to get the hypervoxels what you see is what you get, the baked object will often not match in lighting and shading, so you have to experiment and make your tweaks based on how the baked object looked like after the initial bake, so that is the tricky part.

And remember, you are baking shadows, light and texture in to the baked object, do not expect you can change any of that on the baked object...the only way is to uncheck baked object and redo tweaks and redo baking, hypervoxels will ignore any lights in the scene that is changed after the baking is done.

Cheers..

Michael

prometheus
03-19-2015, 01:30 PM
Too bad xevious2501 tutorial would have been top quality clouds in lightwave. :D


Cloud tutorial with TFD

https://vimeo.com/121981472

Yeah..thatīs a good one, and he also got an eye for how reference clouds look like and seem to match it pretty good, I didnīt watch it completly and with full attention, just briefly..but I think the process is fully replicable with lightwave too..with some differences only on how to set up particles maybe...the main trick is to have an eye for it and reallly match those ref clouds in color, light and shading and with matching turbulent/noisy edges.

Eagle66
03-19-2015, 01:47 PM
Here is an 1 hour free TFD cloud tutorial (in C4D) which announce realistic clouds:

https://vimeo.com/121981472

top quality yes, but realism - hmm.

Mr_Q
03-19-2015, 01:51 PM
I still find it moot to even try this in LW anymore and I am a full-time user of course. It doesn't take much to learn the cloud system in VUE and it's 1000x better. You can FBX over your cameras and meshes if needed, render, done. They don't even take all that long to render when considering the results. You don't even need to buy the full version of VUE. Lots of great presets out in the community too.

Heck even Maya's native cloud system isn't hard at all and looks fantastic and renders quite fast.

Eagle66
03-19-2015, 02:05 PM
Yes, Mr. Q, VUE , Ozone or the quirky Terragen can do clouds and real sky much better, the challenge is how much realism do you get in LW :-).

And IMHO you need VUE Inf. for realism clouds. The cheaper Vue versions don't have this clouds or realism render results.

jasonwestmas
03-19-2015, 02:22 PM
I've never liked the look of lightwaves volumes even in the LW 7 days. I don't think I need to explain why.

m.d.
03-19-2015, 03:20 PM
i found the best clouds out of the box surprisingly were from Terragen...

Vue with Quadspinner helios plugin is quite good though

prometheus
03-19-2015, 03:54 PM
I still find it moot to even try this in LW anymore and I am a full-time user of course. It doesn't take much to learn the cloud system in VUE and it's 1000x better. You can FBX over your cameras and meshes if needed, render, done. They don't even take all that long to render when considering the results. You don't even need to buy the full version of VUE. Lots of great presets out in the community too.

Heck even Maya's native cloud system isn't hard at all and looks fantastic and renders quite fast.

Donīt quite agree with you about vue cloud system 1000% better ...depends on what you compare to in lightwave, vue cloud fractals are poorly designed, as opposed to lightwave cloud fractals and terragen fractals, thus many of the presets are often blobby or too much turbulent noise in the clouds, I canīt say anything about the light and shading and the overall sky system in vue though, that is much better.
But there are things you canīt (or couldnīt do) with vue...using particles for cloud clusters at different level and designing special cloud peaks, metaclouds are not that good really, I think modos voxel system can generate similar type of cloud system but with better noise control.
The general sky atmospheric system is superior in vue though.

Havenīt touched mayaīs native cloud system..unless you mean maya fluids? so I leave it at that..


Some old ogo taiki stuff in lightwave...

https://lh4.googleusercontent.com/-P9dNo1Qiiug/UvZ_GZyHqKI/AAAAAAAACL8/4a8BdhrOfqM/w1280-h720-no/ogo%2Btaiki%2B2.jpg

https://lh3.googleusercontent.com/-5ZDd2fp9IxU/UvZ_Jo6ltPI/AAAAAAAACME/e6egeBtKGAI/w1280-h720-no/ogo%2Btaiki%2B5%2B21min%2B21%2Bsec%2Baa%2B3.jpg

- - - Updated - - -


Yes, Mr. Q, VUE , Ozone or the quirky Terragen can do clouds and real sky much better, the challenge is how much realism do you get in LW :-).

And IMHO you need VUE Inf. for realism clouds. The cheaper Vue versions don't have this clouds or realism render results.

screw ozone...honestly.

prometheus
03-19-2015, 03:56 PM
i found the best clouds out of the box surprisingly were from Terragen...

Vue with Quadspinner helios plugin is quite good though

True..with terragen..but man..waiting for the previewer to update the skies and clouds...horrible, how many times I have said that..I donīt know.

Yup..best looking clouds seen with fractal approach..that is Helios, havenīt been able to get a chance to try that one out, Helios Is equipped with som new special designed for clouds fractals as opposed to native vue fractals.

Have mentioned it before...either revamp ogo taiki under lw teams wings, and look in to improve hypervoxels and include a volumetric sky atmosphere, voxels need geometry based volumetrics and blending mode ala modo which is much better, and also the option to use several particle emitters under one hypervoxels shading instance, and nulls the same, so you can place multi emitters in different altitudes and with different simulated shapes, also...lightwave need point manipulation in layout ..please..please lw team...Make it happen and I show you something you havent seen before :)

Mr_Q
03-19-2015, 04:31 PM
I don't know. We have a guy here who kicks out amazing darn near photo real cloud and atmosphere shots in very little time. Recently we did a 500 feet above sea-level to 25,000 feet fly-up and fly-over and it looked stunning. He busted that out in a day and then a night of rendering on 20 nodes. 1920x1080.

I get the "but this is doing it in Lightwave challenge" though.

bobakabob
03-19-2015, 04:45 PM
I'm presently animating in Maya and rendering in LW. LW renderer is so flexible, simple and quick. I'd really rather not stay in Maya.

Development of Skytracer and Hypervoxels for advanced landscape work would really raise the profile of LW software. Just some *minor* tweaks would be a step forward:
- access to nodes in Hypervoxels
- adding more advanced fractal cloud textures
- updated cloud texture Presets (the present ones are beyond ancient)
- A content folder of ready made environmental / sky dome scenes for Arch Viz and budget productions.
- from reading this thread, more hi res QT output.

jwiede
03-19-2015, 07:48 PM
- access to nodes in Hypervoxels
- adding more advanced fractal cloud textures

As Prometheus has pointed out repeatedly, though, just the above isn't enough by itself. Having better cloud fractals won't really help without better HV blending, and ability to group emitters within single HV shading instance. Without the former, you won't get hardly any improvement overall. Without the latter, making realistic cloud fields / skies isn't really feasible, because a single input source to an HV shader just doesn't allow for the variety needed. Last, without the kinds of volumetric geometry fields and sculpting similar to what modo offers, you'll get decent totally-random cloud fields, but won't really have the tools to realistically edit those cloud fields -- any editing you try with existing tools will leave obvious problem areas in the otherwise-realistic cloud field.

That said, and given the work needed to equip LW for realistic cloud field generation is neither trivial in complexity nor tiny in scope, is it really worth doing all that prior to any major retrofit of particles in general (which is also very needed, and needed in a lot more ways that just generating realistic clouds/cloud fields)? Keep in mind that any likely particle system replacement work will almost certainly also include better shading / blending / editing as a matter of course.

Personally, I'd rather they just focus on overall particle system replacement work, benefiting many more customers overall. Ripping open both pfx and HVs for substantial internal improvements, while still leaving pfx / HVs overall in much the same state afterwards w.r.t. non-cloud uses (thus still requiring overall replacement), just seems like a wildly inefficient way to spend resources.

m.d.
03-19-2015, 07:52 PM
- from reading this thread, more hi res QT output.

you don't mean quicktime do you?

m.d.
03-19-2015, 07:55 PM
Have mentioned it before...either revamp ogo taiki under lw teams wings, and look in to improve hypervoxels and include a volumetric sky atmosphere, voxels need geometry based volumetrics and blending mode ala modo which is much better, and also the option to use several particle emitters under one hypervoxels shading instance, and nulls the same, so you can place multi emitters in different altitudes and with different simulated shapes, also...lightwave need point manipulation in layout ..please..please lw team...Make it happen and I show you something you havent seen before :)

Well I hope they listen....never played with OGO TAIKI...but I'll take your word for it.

prometheus
03-20-2015, 04:52 AM
I don't know. We have a guy here who kicks out amazing darn near photo real cloud and atmosphere shots in very little time. Recently we did a 500 feet above sea-level to 25,000 feet fly-up and fly-over and it looked stunning. He busted that out in a day and then a night of rendering on 20 nodes. 1920x1080.

I get the "but this is doing it in Lightwave challenge" though.

sounds great, I am not saying it is impossible to get stunning vue renders, but there are some factors that are limiting the cloud generation due to itīs fractal library nature so to speak, you have to be very careful of how you tweak the noise to get it right.
I think I recall defiance shots with vue skies, which I didnīt like because I thought the fractals looked wrong and with too much blob appearence in the overall cloud density.

would be nice if you could post some shots of that stunning clouds, but I reckon you are not allowed to do so.

I have noticed that above clouds shots in vue are easier to get to look good and more realistic than ground perspective shots of the skies and clouds.
If there was better cloud fractal functions in vue, and also controlable in ozone in the same way as in vue..and if we could tweak it in lightwave with full vpr as we can with ogo taiki, then I might suggest skip development of ogo taiki, but ozones version havenīt been able to overcome the lack of features that are available in vue.

apart from general atmospheric volumetric models inside of vue, or with ozone or with ogo taiki...I think lightwave needs a way of designing clouds with volumetrics in the same manner as modo and houdini ..and to convert mesh to volumetrics.

old ozone 5 test 4 years ago, havenīt really fiddled with it much since then...


https://vimeo.com/21242230






As Prometheus has pointed out repeatedly, though, just the above isn't enough by itself. Having better cloud fractals won't really help without better HV blending, and ability to group emitters within single HV shading instance. Without the former, you won't get hardly any improvement overall. Without the latter, making realistic cloud fields / skies isn't really feasible, because a single input source to an HV shader just doesn't allow for the variety needed. Last, without the kinds of volumetric geometry fields and sculpting similar to what modo offers, you'll get decent totally-random cloud fields, but won't really have the tools to realistically edit those cloud fields -- any editing you try with existing tools will leave obvious problem areas in the otherwise-realistic cloud field.

That said, and given the work needed to equip LW for realistic cloud field generation is neither trivial in complexity nor tiny in scope, is it really worth doing all that prior to any major retrofit of particles in general (which is also very needed, and needed in a lot more ways that just generating realistic clouds/cloud fields)? Keep in mind that any likely particle system replacement work will almost certainly also include better shading / blending / editing as a matter of course.

Personally, I'd rather they just focus on overall particle system replacement work, benefiting many more customers overall. Ripping open both pfx and HVs for substantial internal improvements, while still leaving pfx / HVs overall in much the same state afterwards w.r.t. non-cloud uses (thus still requiring overall replacement), just seems like a wildly inefficient way to spend resources.

You have some interesting notes, will have to study it later and get back on this maybe.

One thing which just popped up, if we could use multi emitters and multi nulls to move around working with the same hv shading instance ..so it donīt slow down the system, that would as mentioned be great, that would however pose a problem with tweaking fractals..they would be the same overall within that shading instance, what just popped up in my mind....how about if you just could assign multi emmiters to the same shading and lighting group, which is the main cause of calculating and slowing down things, then make it possible to assign different textures within that shading group...that would open up for more advanced cloud structures...but itīs a wild card thinking from me, and it may not be possible.

Michael

jwiede
03-22-2015, 06:21 PM
how about if you just could assign multi emmiters to the same shading and lighting group, which is the main cause of calculating and slowing down things, then make it possible to assign different textures within that shading group

Hmm, within the single shading/lighting group, how did you envision being able to select which texture should be applied (and across what scope)? If I can understand better what you mean, I can probably give a decent approximation of raw implementation difficulty / feasibility.

prometheus
03-23-2015, 04:28 PM
Hmm, within the single shading/lighting group, how did you envision being able to select which texture should be applied (and across what scope)? If I can understand better what you mean, I can probably give a decent approximation of raw implementation difficulty / feasibility.


I think it might be too difficult to also manage it to be able to use different hypertextures within the scope of child nulls that uses the same illumination and rest of the shader.
I have to hit the sack now, so I only came up with a mockup of the most important enhancement for using this...
how hypertextures should be set up for multi textures to be able to work on different nulls within that parented group, I donīt know and I would have to tinker with the idea first.

hereīs at least how it should be able to perform on a null which could have several nulls parented as child/slaves...they would all be using the same illumination and shading group, so it will not slow down things as opposed to using it the way we do now, since we cant do this currently.

I believe the old dynamite plugin was able to do that, not sure though.

now...You could try points and particles, but we all know how tricky that is to move around with VPR feedback in layout, I can do particle fields with some shaping with fractal textures in velocity, but manually moving static points and particles..thatīs not easy to work with, that is why we sort of got nulls at least, but then you are stuck with nulls that canīt be used in parented groups and using the same hypervoxel shading instance the way particle does.

So using particles or point...you got trouble.
using nulls...you got trouble.

My solution...make points editable in layout without any hardfx softfx tricks or third party plugins.
Or make nulls working in parented groups in hypervoxels as I have described...then the case can be closed.


http://forums.newtek.com/attachment.php?attachmentid=127605&d=1427149566

jwiede
03-23-2015, 04:35 PM
That doesn't seem like it would be hideously difficult, though it's a bit difficult to estimate without knowing whether/how HVs could access hierarchies. Ofc, that's also on top of other changes, some in areas where LW3DG clearly seem to be reluctant to modify HV code, so... (shrug)

I keep hoping Ogo-Taiki's author will open source the plugins on github, rather than abandon them. Alas, not seeming likely.

Crap. I was about to offer to check Dynamite for you, then realized it was a casualty of my recent HDD catastrophe. Not a huge loss, but still finding annoying little losses.

prometheus
03-23-2015, 04:48 PM
That doesn't seem like it would be hideously difficult, though it's a bit difficult to estimate without knowing whether/how HVs could access hierarchies. Ofc, that's also on top of other changes, some in areas where LW3DG clearly seem to be reluctant to modify HV code, so... (shrug)

I keep hoping Ogo-Taiki's author will open source the plugins on github, rather than abandon them. Alas, not seeming likely.

Crap. I was about to offer to check Dynamite for you, then realized it was a casualty of my recent HDD catastrophe. Not a huge loss, but still finding annoying little losses.

I got some older versions of dynamite, but I donīt think I installed it on this machine and 11.6.3 lightwave version, would have to check where they are installed, might be on some older versions or on another machine, had some issues with crashing from time to time, the blending mode is as it should be for smooth blending tension between nulls or particles, though it is quite slow..at least when I compare to way modo blends itīs voxels, I know dynamite had a master slave version..but that was only available in
pro versions as I recall.

Stop talking to me now, I got to get my self a beauty sleep before it is to late :)
So I will jump in tommorrow next time I reckon.

jwiede
03-23-2015, 05:07 PM
Yeah, I gotta run for the airport anyway, will talk more tomorrow! G'night!

prometheus
03-28-2015, 09:24 AM
might be useful to know, the layout mesh edit tool plugin(free win 32 only) might help editing point clusters in layout when working with hypervoxels for clouds.

it will not recognize points if you just create points or spray points in modeler, you need to convert it to 1 point poly in order for it to recognize as a mesh and then being able to edit itīs points, itīs a bit quirky in selecting and updating, and you canīt really
get it to work directly in vpr and update properly, so I suggest using another window in vertex mode and select and move there, but keep a dual window with vpr realtime feedback on the voxels.

Think that is the closest way of editing points in layout together with hypervoxels, softfx, for editing will be too difficut, and edit pfx with particles is quirky too.

and for more fuel to the voxel enhancements...I think I should be able to post a showcase of the difference between modo voxel blending and hypervoxel blending, and dynamite voxel blending soon.



http://forums.newtek.com/attachment.php?attachmentid=127639&d=1427556207



Michael

jwiede
03-28-2015, 04:21 PM
might be useful to know, the layout mesh edit tool plugin(free win 32 only) might help editing point clusters in layout when working with hypervoxels for clouds.

A workflow that depends on an unsupported, win32-only plugin is very fragile/risky, IMO. It might "work" today, but could break at any time, and likely will when needed most. Requiring 32-bit LW version for particle editing constrains ability to usefully "work" quite a bit as well, and adds another LW install to maintain, since you'd still need the 64-bit version for most everything else. It's like workflows relying on Ogo Taiki or Dynamite, perhaps "doable", but fail the instant you factor in efficiency and risk.

prometheus
03-28-2015, 06:21 PM
A workflow that depends on an unsupported, win32-only plugin is very fragile/risky, IMO. It might "work" today, but could break at any time, and likely will when needed most. Requiring 32-bit LW version for particle editing constrains ability to usefully "work" quite a bit as well, and adds another LW install to maintain, since you'd still need the 64-bit version for most everything else. It's like workflows relying on Ogo Taiki or Dynamite, perhaps "doable", but fail the instant you factor in efficiency and risk.

The only time I found myself using 64 bit, that would be when using over 3 millions of particles or fiddling around with huge polyamounts...other than that..I found 64 bit version lacking so many plugins only available in 32 bit, then again..that will of course change, as for instance with no more support for 32 bit turbulenceFD, so in time 64 bit for sure.

But of course..the next release of lightwave will of course get hypervoxels to treat nulls in groups or make it possible to edit points directly in layout..right:D

Michael

jwiede
03-28-2015, 09:38 PM
But of course..the next release of lightwave will of course get hypervoxels to treat nulls in groups or make it possible to edit points directly in layout..right:D

Yes, and you should definitely hold your breath for this! ;D





...just please don't do so near any furniture with sharp corners. If they do, that's good for LW owners, but I've been asking and hoping for HV improvements since LW9 to no avail, so moved on in that area a while ago.

Currently, I use C4D for most of my particle work, because between X-Particles, Effex, and TFD, the combined toolset is quite thorough, and shortly also Houdini Engine for C4D. That said, modo does have some nice volumetric editing tools as well, so I've been known to move particle data from C4D to modo (or Houdini) for editing. I find I'm doing less of that lately with X-Particles v3, and Houdini Engine in C4D will remove most of the remaining need to leave C4D, I suspect. It would take a huge set of improvements to HVs and pfx to get me to move from C4D back to LW for particle work.

prometheus
03-29-2015, 08:08 AM
Yes, and you should definitely hold your breath for this! ;D




...just please don't do so near any furniture with sharp corners. If they do, that's good for LW owners, but I've been asking and hoping for HV improvements since LW9 to no avail, so moved on in that area a while ago.



yeah...I havent reached the surface to take a new breath, and I reckon I am a few steps behind you in this matter, if My purse could be better..then I might have caught up with you :D


Michael

inakito
03-31-2015, 04:52 AM
Anyone using OGO omnn Lightwave 2015? I am able to load sample scenes but no luck opening the interface window...

ianr
03-31-2015, 07:08 AM
Great Points above,that is why
ALL Wavers should Poll & Vote
fer' Houdini Engine 4 LightWave!!

prometheus
03-31-2015, 10:29 AM
Anyone using OGO omnn Lightwave 2015? I am able to load sample scenes but no luck opening the interface window...

Unfortunatly not me ...I donīt have lw 2015, not sure if erikals upgraded to 2015, he has tested ogo taiki now and then, so maybe pm him.

erikals
04-01-2015, 01:04 AM
nope, no LightWave 2015 here, so cannot test that one, remember OGO is 32bit only though...

JohnMarchant
04-01-2015, 01:56 AM
Yeah just wish we could get an update and a 64 bit version.

Houdini engine would be a great add on for LW

inakito
04-01-2015, 03:35 AM
I am testing here Lightwave 2015 and alsoi 32bits ;) and no luck

Airwaves
12-08-2015, 10:23 AM
I realize this thread is old but I never found the tutorial or scene files. Was it ever done? I am trying to make a cloud scene and those clouds would be perfect and my attempts are just not cutting it.

**Ignore this, I did not see all the pages and thought there was only one, and now after reading the conversation I think I know which way to go*****

Can anybody explain what baking is and can you render a scene with baking? I am all confused as I have never used baking before. Thanks