PDA

View Full Version : INeed Your feedback for the next generation of Hypervoxels,features,improvements etc.



prometheus
01-07-2014, 11:33 PM
So..Maybe..just maybe I can find the time to put it all together, I am gathering all the new features,improvements,bugs in a mindmap/list, then Hopefully I can present that in a nice pdf format and post up for the Lw team to
take a look at.

I would like to bump back here again and see what suggestions you would have for The Next evolution in Hypervoxels.
Erikals has provided some nice youtube clip I will link to as well, regarding the distance between particle gradient, so that will show up, comparative notes against, modo voxels might show up, I might have to install and test, though
from what Ivīve seen ..blending mode is much better, and it has features hypervoxels donīt.
Comparative notes VS the old dynamite will be made, and also against max afterburn,and houdini cloudFX.

So please jump in and give me some suggestions, but also keep in mind that hypervoxels isnīt a fluid (cfd system) that is sort of another topic, hypervoxels is a sprite,volume,and surface render system applied upon
particle effects, so liquids and fluids is likely to have their own threads, with some exceptions.

So whatever it could be for hypervoxels,Ui, workflow,interaction with other Lightwave elements, speed, quality, bugs, hypertextures,openGL,appliance types,blending modes, hv deformers etc..throw them at me and hoperfully we could give some bones for the lightwave team to put flesh on, while they secretly fiddle with it til next Lw version:)

Michael

Areyos Alektor
01-12-2014, 04:29 PM
It's a good idea, and a vast subject :)

Cold I would say that it would be good to optimize performance, to have a true "Node Editor", to be able to generate geometry.

It's already a lot. But after there are lots of small things to improve. It should be that I think quietly. For example an update of the "Metaball" and a more direct link with the "HV".

papou
01-12-2014, 06:46 PM
- Aspect: no more bubbleized HV, Shrink them, blend them based on distance and velocity.
- Model: freeze them to object.
- Compatibility: import Realflow and Blender blending & weight.
- Fix: add a load clip into the Sprite Clips mode.
- Improve Shading: Same settings like Surface editor.

prometheus
01-12-2014, 07:03 PM
It's a good idea, and a vast subject :)

Cold I would say that it would be good to optimize performance, to have a true "Node Editor", to be able to generate geometry.

It's already a lot. But after there are lots of small things to improve. It should be that I think quietly. For example an update of the "Metaball" and a more direct link with the "HV".

yes I miss a true node editor too, when you want to add some extra touch with nodes, the VPR canīt handle the automatic update as if you were to tweak hypervoxels texture directly ...those respond automaticle, it is
a workflow killer to have the node editor "hack" and not able to tweak properly and see the results.
That is accounted in my to do list for hypervoxels already though.


The metaball blend is accounted for too, I just recently started to fiddle with the old dynamite plugin again, which has a much better blending mode, and more specific shader for fire with itīs nice fire illumination.
Though dynamite is actually much slower when you put dynamite voxels on to particle trails than hypervoxels, also depends on texture choosen of course.

in the modo vs lightwave thread..I showcased the volume item from modo, which allows for any geometry to be converted to volumetrics, and then using textures on that, the sample isnīt ideal though since it isnīt
setup to have the textures affecting the volume edge surface boundary, and it just cuts through the geometry, it should work more like houdiniCloudfx tools for best results.

http://forums.newtek.com/showthread.php?139559-Opinions-Lightwave-vs-Modo&p=1362412&posted=1#post1362412

direct link to hv? not sure what you mean there...if you mean being more acessable, hypervoxels is no acessable in the fx tab/hypervoxels...if thatīs what you mean?

theres quite a lot of improvements to be done, moving or adding acces to the hv particles/set particle weight and color to be accessable directly in the Hv tab, as it is now you have to jump between the two tabs and for
those not knowing exactly how to do it, it will be lost for many useres not being aware on how to use it.
That will also be a need to link to further layout improvements such as a weight display in opengl layout, and a paint weight brush directly in layout, these three things will make the workflow much much easier when
you simply can paint in areas of hypervoxels for rocks, mud, water, or mist or even used on cloud grids to paint cloud density..with a direct feedback on the relation to other scene items.

Michael

prometheus
01-12-2014, 07:07 PM
- Aspect: no more bubbleized HV, Shrink them, blend them based on distance and velocity.
- Model: freeze them to object.
- Compatibility: import Realflow and Blender blending & weight.
- Fix: add a load clip into the Sprite Clips mode.
- Improve Shading: Same settings like Surface editor.


-Aspect, yes distance I think we have, no velocity though only speed.

same settings like surface editor, yes...at least getting access to shaders and node editor material.
And yes, we always have to load a clip in the image editor first in order to use clips, that is annoying and it should be just as simple as load clip from the sprite clip panel,
that is indeed a fix needed.

Thereīs also a bug that I donīt think might have been reported through the bug report system, hypervoxels applied on particles on a surface will distort certain procedural textures on
that very same surface, for a fact I now it destroys the turbulence texture, but havenīt gone through them all.

Using several nulls as one group using the same hypervoxels group would also help, and also the same with using multi emitters but using the same hv group, that would allow for a workflow where you
can direct nulls or several emitters for clouds and several cloud item or distributed smoke without having hypervoxels freeze since it canīt deal with using multi groups of hypervoxels (more shaders to calculate from)

For example ..having two null voxel clouds each one of them with the size of 200 meter, and intersecting, it will choke lightwave, but using two particles or points with 200 meters size each, and intersecting will
not make that happen, because particles or points in the same object or emitter is using one hv shader to calculate from while nulls are not.
but you cant move points, due to lack of mesh edit in layout, and the same with particles...so these limitations are choking your workflow to work smoothly with designing where to put hypervoxels
and direct them in layout.

I will take note of all this and jump back to before I release my PDF dummy for improved hvīs.

prometheus
01-12-2014, 07:36 PM
Note to myself...
texture falloff for hypervoxels seem to never have worked, changing those values to nothing, using the node editor hack and textures from within that and changing falloff in the node editor works, strangely.
thickness..gradients never seem to have worked either. and we could use a curve map function for this too, and for density channel as well...something like we do have for turbulenceFD if possible.

Michael

Areyos Alektor
01-19-2014, 07:49 PM
For the link I was talking about "HV" / "Metaballs". In the same logic to be able to generate geometry.

prometheus
01-19-2014, 08:02 PM
For the link I was talking about "HV" / "Metaballs". In the same logic to be able to generate geometry.

ahh..I thought it might be something like that, metaballs/metamesh maybe, what situations would you find this useful for? instead of just making metaballs in modeler?

Michael

robertoortiz
01-19-2014, 08:24 PM
Core needs:
Full Bullet Integration (Simulation)
Nodal Control
Artist directable shapes based on a timeline (Lattice deformations)
Liquids simulation based on hydrodynamics physics

Here are some cool papers I would look into:
http://s2012.siggraph.org/attendees/sessions/100-120

http://www.youtube.com/watch?v=PTcLqyQO_4Q

A Simple, Efficient Method For Realistic Animation of Clouds
http://www.youtube.com/watch?v=XLXU1o4F6pE
Paper:

http://www.naturewizard.com/papers/clouds%20-%20p19-dobashi.pdf
http://ken.museth.org/Publications_files/Museth_TOG13.pdf

Paper Libraies:
http://classes.soe.ucsc.edu/cmps262/Winter07/projects/clouds/draftpaper.pdf
http://www.naturewizard.com/
http://kesen.realtimerendering.com/

prometheus
01-19-2014, 08:51 PM
Core needs:
Full Bullet Integration (Simulation)
Nodal Control
Artist directable shapes based on a timeline (Lattice deformations)
Liquids simulation based on hydrodynamics physics

Here are some cool papers I would look into:
http://s2012.siggraph.org/attendees/sessions/100-120

http://www.youtube.com/watch?v=PTcLqyQO_4Q

A Simple, Efficient Method For Realistic Animation of Clouds
http://www.youtube.com/watch?v=XLXU1o4F6pE
Paper:

http://www.naturewizard.com/papers/clouds%20-%20p19-dobashi.pdf
http://ken.museth.org/Publications_files/Museth_TOG13.pdf

Paper Libraies:
http://classes.soe.ucsc.edu/cmps262/Winter07/projects/clouds/draftpaper.pdf
http://www.naturewizard.com/
http://kesen.realtimerendering.com/

yeah..nice and all, but that will never be within hypervoxels I would say, much of it is much different in tech, textured volumetrics for clouds etc, something like ogo taiki or ozone.
have no idea what you mean about bullet, implementation...thatīs something for particles and is a chapter by itself, and I want that too, but it isnīt appliable for hypervoxels as a volumetric renderer.
Same goes for liquid simulations etc, so thatīs what I mentioned about keep it somewhat restricted to what hypervoxels is today and how hypervoxels itself should be improved..not a completly new
system that isnīt a volumetric renderer.
So even though those other things might be needed, it doesnīt fit in with hypervoxels or this thread.
Nodal control better than the procedural hack we have today, that I agree with.

Michael

shrox
01-19-2014, 09:31 PM
-...Thereīs also a bug that I donīt think might have been reported through the bug report system, hypervoxels applied on particles on a surface will distort certain procedural textures on
that very same surface, for a fact I now it destroys the turbulence texture, but havenīt gone through them all...

Yes, I can't use HVs for rocks anymore!!!

prometheus
01-19-2014, 09:45 PM
Yes, I can't use HVs for rocks anymore!!!

yes I know, we have been discussing that you and me, I myself havenīt filed a bug report for it, and donīt know about you or anybody else, so we canīt expect them to deal with it if no one
does it.

a simple scene with some hypervoxels on nulls or particles, and a turbulence texture on color or bump for a ground plane and the texture will be screwed up.
I really donīt like the report system, if itīs ok for me to simply mail a scene to the bug mail adress from my own adress, then I will do so.
The lw 11.6 build previous before the final build still has that bug.

What I said regarding Roberto Ortiz whishes, If I were to put out my list...which isnīt half that advanced and vast if you compare to all the stuff Roberto mentioned, it will still be
maybe too much for them to deal with, and that is just to catch up with getting some of the best features from what dynamite voxel(particle engine..not fluids) had as features.
And also catching up with modoīs voxel engine, with fully volumetric geometry( see previous image of the modo head) and also implement better blending mode and illumination scattering methods.
Remarkable that modo didnīt have a voxel engine not long ago, and now upon the first iterations it has surpassed hypervoxels in most cases it seems, lightwave team focused on
not necessary wrong things(for me maybe) more on other things...while the lux team had a better match in that area compared to what I would like to see.

Note to myself (when I compile all this to the pdf) loading of color gradients, that way you can retain the hypervoxel shape and texture, but change color gradients so you can choose from black body fire gradient, or
nebula gas color gradient, or a cloud color gradient etc or other spectacular gas or smoke colors without destroying what you got hypertexture, it will be a speed workflow saver...just put the load gradient beside or beneath the current gradient layer as a button option.

Michael

vector
01-19-2014, 10:24 PM
It's a pity getting this for no money http://cgterminal.com/wp-content/uploads/2011/12/Blender-Smoke-Simulation-Tutorial.jpg and see that HV haven''t got updated for years. I know TFD rocks and I don't ask for exactly the same results but anything more useful. I agree with every word you say, but those smoke balls effect I got with HV sprites is a specially pain in the b***

prometheus
01-20-2014, 02:29 AM
It's a pity getting this for no money http://cgterminal.com/wp-content/uploads/2011/12/Blender-Smoke-Simulation-Tutorial.jpg and see that HV haven''t got updated for years. I know TFD rocks and I don't ask for exactly the same results but anything more useful. I agree with every word you say, but those smoke balls effect I got with HV sprites is a specially pain in the b***

yeah..but you have to request a native fluid solution for Lightwave to the lightwave team, or buy whatīs available..namely turbulenceFD.
hypervoxels is another type of engine, so a little pointless requesting fluids for hypervoxels.

If you know your way around hypervoxels, you should be able to work that magic anyway, but depends on situations, what could help a little is the distance between particle gradient erikals posted as a video.
and we need a better blending mode and a thickness gradient that works with local density(currently it is lousy, namely it doesnīt work with gradients in thickness channel, only dissolve)
Also ...as I mentioned, the falloff has never worked properly as I know of, it does work to some degree with nodes, if falloff could work ..it could help reduce that spherical hard edge that sometimes can be seen in
hypertextures.
Those things might bring better look to hypervoxels, but getting best smoke is a job for fluids.

avoiding puffballs, needs to taking in account not letting particles stray away, and often hvīs size and dissolve at particle age, and local density, and also let the hypertexture value cut in to the volume enough.
Im not sure many other voxel engines has so much better solution for that, except maybe for those engines using distance between particle gradients, and maybe afterburner which has this type of explosion demon
that space warps the voxels somehow, but I donīt know really how.
That is something I would like to see too, a secondary hypertexture that works decently over the whole particle cluster and deforms it over the whole area, not only hypertextures that works only per particle as
we do have now.

Michael

prometheus
01-25-2014, 04:15 PM
SUBJECT-CIRCULAR PATTERN WITHIN HYPERTEXTURES.

Hereīs one issue regarding hypertextures, i canītell for sure, but I recall hypertextures having a smoother look around lightwave versions 8-9,6 maybe.
The issue is circular patterns or hypertexture with a pattern cut off that leaves emty circular breaks in the general texture, image below is a magnified screenshot of the hv tab and the hv thumbnail where this can be seen if you take a close look, the right image is exactly the same as the first, but I just added some guide circles to showcase where I feel the texture is abrupted, so it requires for some examination with eyeswitching between them two thumbnail previews of the hypertexture.

I think this seem to be pronounced in final hypervoxel renders too at some times, and I have a very faint recallection of Lightwave hypervoxels from older versions beeing smoother.

If someone has lightwave version 7-9.3 maybe and can take a screenshot of the turbulence hypertexture thumb preview, that would be interesting to compare and to conclude if that indeed was the case.

http://forums.newtek.com/attachment.php?attachmentid=119578&d=1390691702


119578


Michael

Areyos Alektor
01-26-2014, 10:15 AM
I don't understand the interest to expect fluids in the "HV", for me it's more than a VFX tool as is "3DCoat" for modeling.

What I told you about the link for the "Metaballs" is the same way that the "SubPatch" evolved (and "FiberFX"). A control that is being updated, management by the "Nodes", it's already a good start.

prometheus
01-26-2014, 10:35 AM
I don't understand the interest to expect fluids in the "HV", for me it's more than a VFX tool as is "3DCoat" for modeling.

What I told you about the link for the "Metaballs" is the same way that the "SubPatch" evolved (and "FiberFX"). A control that is being updated, management by the "Nodes", it's already a good start.

Yeah....but I already mentioned that, fluids isnīt really on topic for hypervoxels, think it was ortiz mixing that in, but it belongs to other feature lists.
Michael

Areyos Alektor
01-26-2014, 11:49 AM
Yes :) I should have been clearer in my response ;)

Julez4001
01-30-2014, 01:21 AM
I just know that LW 12.0 will be a paid upgrade so it better have a leap in render technology and not just another Shadow Catcher.
FiberFX is still unreliable, at least not on the same level of Sasquatch. HyperVoxel needs to emulate Houdini usage of FumeFX.
http://vimeo.com/44638236

prometheus
01-30-2014, 01:36 AM
I just know that LW 12.0 will be a paid upgrade so it better have a leap in render technology and not just another Shadow Catcher.
FiberFX is still unreliable, at least not on the same level of Sasquatch. HyperVoxel needs to emulate Houdini usage of FumeFX.
http://vimeo.com/44638236

I understand we all want great fluids inside of lightwave etc, but sorry Julez4001, this is out of topic for the this thread as I mentioned before, we would have to start with a completly new thread asking
for native fluid solution instead in such case.

Hypervoxels is a different system and has nothing to do with a fluid solution, and I do not want the lightwave team to do nothing with hypervoxels, or remove it...that leave us with the fact that
hypervoxels can be improved a lot on as a volumetric item tool, My references showned against cloudFX and how modo have improved their volumetric system is something to aim for, I donīt excpect them to bypass
it, I would be very happy if they did, but might not be realistic.
All the stuff I listed here, and in my final list will keep them busy if they were to try and follow it all, which I cant expect.

The link you provided is yet another fluid tool, and it is a little pointless to improve hypervoxels that way, since it canīt be done...and it is a completly different system that isnīt
represented in how modo or houdini volumetic items work either in underlying tech.
But do go ahead and post feature requests of new fluid tools for lightwave...but it should belong in another thread, it can do nothing for hypervoxels.

Julez4001
01-30-2014, 01:54 AM
119724

I didn't ask for fluids. Not trying to derail your thread.
This is a old image of Hypervoxel doing smoke and such.
There has to be a way for them to do pseudo-smoke. Something between what it can do now and fluids.
It will take more than hypervoxels but update to the particle system as well.

prometheus
01-30-2014, 02:11 AM
119724

I didn't ask for fluids. Not trying to derail your thread.
This is a old image of Hypervoxel doing smoke and such.
There has to be a way for them to do pseudo-smoke. Something between what it can do now and fluids.
It will take more than hypervoxels but update to the particle system as well.


Ok..Understand, just though it was a little odd you were pointing to a fluid simulation, emulating, well pseudo smoke..alright then, but we might perhaps talk about how that
could be approached, and youré right about the need of a new or enhanced particle system...in such case we should look at extreme particle amount and handling of that in ways that krakatoa does,..and
we could use new improved noise and vortex fields.

But I think I have mentioned all that....when you work on such style of particles with several millions or billions, you donīt throw in any large particle size Hvīs with hypertextures, rather use very small sprites or
pixie dust or some other type of volumetric renderer..and I have pointed to erikals videos showcasing the distance between particle gradient that helps hypervoxels to fade away more naturally as well as
being much more realistic for liquid surface renders.

I donīt think theres any chance to emulate such a fluid as you referenced in a nice way for hypervoxels ..using the large hypertextures voxel approach, another thing would be as mentioned to use
millions or billions of tiny tiny particles, but then we go off topic anyway from this hypervoxels thread, since we can just employ hvīs as they are almost...itīs more up to the particle engine being able
to first handle that amount, then what type of noise or vortex fields that can be implemented to get a near nature fluid look in the dynamics.

so it might be somewhat extreme versions of particle simulations..take this and multiply by thousands and a volumetric renderer, and more advanced noise fields..
http://www.youtube.com/watch?v=8AUguci1lPY
http://www.youtube.com/watch?v=OYpvZyikRqY

sure fiddling with hypervoxels...
http://vimeo.com/3438515

But We are better of with fluids though for that particular stuff...
http://vimeo.com/35832026
http://vimeo.com/user680656/videos/sort:date/format:thumbnail

The problem is that as with your sample hv image, such approach with large voxels size and a particle emitter is a combination with particles that can not be allowed to be simulated
with vortex fields to much, that will screw up everything since the voxels will start to spin around and look all wonky, for such stuff you shouldnīt have too much noise in the particles
and just fake it with hypertexture effects, and that approach isnīt as near as realistic as to using millions off particles instead actually being affected by a noise field..Krakatoa style.
hypervoxels with hypertextures, might get away with some explosions, particle trails/smoke from aeroplanes and maybe some semi realistic nuke and vulcano..but the difference in realism
will show when you line it up against a fluid sim.

Julez4001
01-30-2014, 02:19 AM
I think what they need as a intermediate step before in-house fluid system, is to increase particle system limit and auto-grouping.

Sort of a preset system for secondary and tertiary spawning from the primary particle system.
These groups can be read into Hypervoxel as such and add the wispy variations from the core smoke look in the video I sent before and that first one in your last email.
I think Newtek would have to up the particle count as well create a hub between particle system ad hypervoxels.

I would have also expected at this point since Hypervoxel existed that the rendering would be at least 50% faster.

Julez4001
01-30-2014, 02:22 AM
Also create a hypervoxel light, optimize to increase rendering and nothing else.
It gives you those gorgeous shadows without trying to physically accurate raytracing.
An idea for someone who knows nothing about code but hopefully an idea to ponder.

prometheus
01-30-2014, 02:33 AM
I think what they need as a intermediate step before in-house fluid system, is to increase particle system limit and auto-grouping.

Sort of a preset system for secondary and tertiary spawning from the primary particle system.
These groups can be read into Hypervoxel as such and add the wispy variations from the core smoke look in the video I sent before and that first one in your last email.
I think Newtek would have to up the particle count as well create a hub between particle system ad hypervoxels.

I would have also expected at this point since Hypervoxel existed that the rendering would be at least 50% faster.

I do not know what the particle limit is now, it must be very much dependent on your ram I suppose, I managed to render 8 millions at one time, it was not long ago they removed the 1 million limit in rendering, so 6-8 millions I know it can simulate and render, the problem starts for me at around 2-3 millions of particles depending on if I use just particle velocity noise fields or use a wind noise.
whey you reach 2-3 millions on my system anyway, it takes a long time for the particles to cook/and settle...it isnīt really the rendering that seem to be the biggest issue, at least not with 3 millions particles...sure it will probably take much much longer times for each million with the rendering, but foremost it is the simulation with wind fields that are extremly slow, and the settlement of them when raising from 500 000 to several millions.

Secondary and tertiary spawning...well it would be nice, and I have infact requested something like it, maybe more a way to simply choose the emitter and with some activation button slider, I could write in 1-2-3-4 and it will set up the emitter with several child emitters spawning..instead of adding several emitters, parent them, then set the right parent mode etc.

But I donīt think that would help bring forward such whispy variations from the core smoke in the video you showcased, that just needs advanced noise fields, and particles actually able to use vorticles(partcles themself as vortex fields) And of course a lot of particles and a good fade..distance between particles, or a density curve based on particle density or similar.
Krakatoa seem to have a sort of partition renderer, I think it means simulating at lower amount of particles, but render out higher amount of filled particles at rendertimes.

- - - Updated - - -


Also create a hypervoxel light, optimize to increase rendering and nothing else.
It gives you those gorgeous shadows without trying to physically accurate raytracing.
An idea for someone who knows nothing about code but hopefully an idea to ponder.

That ...I donīt quite follow? sample of gorgeous shadows?

Canīt stress this enough I guess, this has been posted here before..put this in, and also in volumetric and sprite mode, and not only in blend and size mode, but also in thickness and dissolve
and density channels if possible...

http://www.youtube.com/watch?v=xMQlAwlDfqg



and for what I mentioned about Vorticles..that is something that pawel olas fire and smoke generator had, it could also emitt from weight maps not possible with lw particle system today.