Page 2 of 3 FirstFirst 123 LastLast
Results 16 to 30 of 31

Thread: can this be natively done in LW

  1. #16
    Registered User
    Join Date
    Aug 2015
    Location
    Barcelona
    Posts
    75
    Example is too low in resolution.
    Last edited by nez; 09-15-2020 at 02:28 PM.

  2. #17
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    17,070
    Quote Originally Posted by nez View Post
    I kinda followed lw development, and loved that they were including more tools such as vdb. maybe they planned adding on it, but I guess from some time ago big bucks don't like LW. In hindsight old VPR was like a underdeveloped EEVEE, similar hack with diffuse surfaces but without baked reflection and such. Now EEVEE even has SSGI, just nuts. Had they invested on that, maybe we would be talking otherwise.

    Anyhow I'm diverging here, the one I liked was the most calculation expensive one, the one that looks ultimately nicer above else lol. was the most similar one to the hyperion renders.

    2016 Example of what was extremely good VPR for, and the nice render you did.Attachment 148691
    No following you about 2016 sample and VPR? maybe you thought it was a Lightwave render?
    That image you just linked has nothing to do with 2016 and not anything to do with a Lightwave render or VPR, those cloud images on that thread is just rendered in blender cycles with GPU processing.

    So none of those cloud images or this one below are made in Lightwave, so it was the last one posted there you like, ok..but it really misses the powder effect, which I probably could introduce a bit more, the others have more of that, but also more backlighting.
    I am a bit busy a couple of hours from now, but will try and open that scene in blender portraying that last cloud image later.
    Gathered them in all one shot here..



    Click image for larger version. 

Name:	WDA disney Cloud scattering tests.jpg 
Views:	276 
Size:	226.3 KB 
ID:	148693

  3. #18
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    17,070
    Quote Originally Posted by nez View Post
    Example is too low in resolution.
    What sample? do you mean the image you tried to link?

  4. #19
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    17,070
    Nez,

    Since you asked for renderstats..

    Hereīs the renderstats and some settings info over bounces, transparency bounces also affecting rendertime, I havenīt done any extensive optimizations of it, I could probably cut rendertimes with some reduction of settings without loosing quality that much.
    Around 21 minutes and 54 seconds. for a resolution of 1440x810,thatīs 75% of a 1920x1020.
    When starting it I wasnīt aware of that I had it set to 75% ..so that will have to do.

    The vdb sample tested.. is only the quarter resolution sample provided, so not the one with thehighest detail.
    I also have some nodal trickery with light path node, curves, converter math node, contrast nodes etc, and how you set color management is important, this is high contrast and filmic.
    I did some minor changes just with color management and changed mode for the math node, that brought out more of that darkness almost powder effect in there, rendering now.

    Blender 2.83 beta, denoising in post process 3px, strength 0,35 for feature and strength.



    Click image for larger version. 

Name:	wdas clouds great scatter prom10.jpg 
Views:	154 
Size:	213.1 KB 
ID:	148694

  5. #20
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    17,070
    And a last compare image, we should really have this discussed on the blender forums though, since we are going off topic from your thread topic.

    the image to the right is direct interactive render at 100 samples, same samples as finished render, but no noise filtering, but on that image to the right I changed math node to another mode and changed gamma a bit.



    Click image for larger version. 

Name:	changed math mode and gammajpg.jpg 
Views:	195 
Size:	109.9 KB 
ID:	148695

  6. #21
    Registered User
    Join Date
    Aug 2015
    Location
    Barcelona
    Posts
    75
    Quote Originally Posted by prometheus View Post
    And a last compare image, we should really have this discussed on the blender forums though, since we are going off topic from your thread topic.

    the image to the right is direct interactive render at 100 samples, same samples as finished render, but no noise filtering, but on that image to the right I changed math node to another mode and changed gamma a bit.



    Click image for larger version. 

Name:	changed math mode and gammajpg.jpg 
Views:	195 
Size:	109.9 KB 
ID:	148695
    yeah, I wanted to up a gif, but it was too little to appreciate any detail. Why don't have it discussed there, we are just equating software capabilities since I was curious, and up until the clouds LW was holding up just fine. Darn those clouds are good. Have you tried with the new Open image denoising? it works wonders. Also Adaptive sampling can shrink significanly your render times. Anyhow, here's an all quads subpatch hand grenade I did i modeler years ago, rendered in cycles.Click image for larger version. 

Name:	bamb2.png 
Views:	110 
Size:	2.77 MB 
ID:	148699
    Last edited by nez; 09-16-2020 at 06:13 AM.

  7. #22
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    17,070
    Quote Originally Posted by nez View Post
    yeah, I wanted to up a gif, but it was too little to appreciate any detail. Why don't have it discussed there, we are just equating software capabilities since I was curious, and up until the clouds LW was holding up just fine. Darn those clouds are good. Have you tried with the new Open image denoising? it works wonders. Also Adaptive sampling can shrink significanly your render times. Anyhow, here's an all quads subpatch hand grenade I did i modeler years ago, rendered in cycles.Click image for larger version. 

Name:	bamb2.png 
Views:	110 
Size:	2.77 MB 
ID:	148699
    I think I have had adaptive sampling on while not seeing any speed improvements, but I could be mistaken, have to check again.

    No..2.83 only installed, so at least any viewport denoising isnīt working, I will soon install the latest blender versions ..they also improved the sky background now so it actually seems much better than the physical sky we have in
    Lightwave, but I have to check and verify.

    A lot also depends on how the cloud is lit, unfortunately there was no sun location data for how the hyperion render was setup, angle/pitch and such, it matters a lot how the rays are entering the clouds for that right effect.

    Hereīs a backlit version, though I did some additional changes on the shading as well, only viewport render so no denoising here...thus a bit grainy.

    As for posting images, unfortunately these forums doesnīt allow for posting images at a certain size unlike the blender forums, so what you can do is to upload it by attaching it, then go to preview post and open that preview image in a new browser window, then copy itīs url and choose image insert again, click from url and uncheck the box retrieve remove, and paste back in to a new image insert in that preview post in to the url field, that will give you larger images..that is how I hack the image attachements, itīs a pain in the..but thatīs what you can do.

    Yes..great model there of the grenade, I hacked the image posting process.

    For the clouds, what I like in blender is the renderspeed and ease of how to use multiple bounces in the volume, as opposed to how both Lightwave and how houdini do it with various samples and control settings, I just find it way more easier to use blender..apart from that the renders are faster than them as well, then I found my ways in how to use various nodes to get a decent look at least.

    With lightwave and trying the VDB rendering of the disney asset, you will have to confront VPR taking much longer time to iterate the first iterations of the cloud, meaning it takes much longer time to start seeing how the shading
    is turning out in the volume, and it sort of deminishes your power to tweak in a decent time, itīs a bit similar in houdini as well though I think that one may be faster as well, with blender I get a more faster direct feedback.
    And apart from that, controlling the multiple scattering with intensity and ray amount is much more tricker to work with, than just simply raising bounces in blender.

    So even though I could work more with and challenge lightwave with more tweaks in the GI, and in shading...I do not think its worth it because of those issues, and that with the notion on that I do not know if I ever could
    get equally good rendering in the volumes.

    When it comes to noise based non VDB clouds though, with a special feathering look etc, Lightwave is the best option for me(thanks to how easy it is to set up volume items and all itīs fractals, blender not so nice on that apartment), as long as I do not need to match such kind of scattering as we see in the clouds here.

    Since blender is lacking in that apartment, I would have liked a way to export any volume item with a noise created from Lightwaves all fractals, in to blender and use the blender render engine, but as far as I know..there is no such way to export from a volume primitive item, only if you use gas solver or other vdb sims, but not from a static primitive volume item, and blender doesnīt seem to have much in terms of adapting more fractals, so there you go..have to use different software depending on.







    Click image for larger version. 

Name:	Backlit_Cloud.jpg 
Views:	103 
Size:	69.9 KB 
ID:	148700

  8. #23
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    17,070
    To note, the cloud samples I used were the quarter res, around 66 mb in vdb size, so that isnīt much.
    I switched to half res which is around 480 mb, and that worked in the relatively small viewport for interactive cycles rendering, but employing final rendering at full 1920x1080 res, my system and the cuda canīt handle it.
    So that means switching back to CPU for the final renderer, and thus much longer rendertimes, more than the double, as for what I can see going on now when still rendering, then again I also increased samples for both transparency and volume bounces from 50 to 85, the transparency bounces is also important.

    I will have to see how that looks with higher res, a bit of a change in lighting angle.(Oh how I hate the rotational system in blender)
    I prefer it pretty much with a simply heading, bank and pitch which states exactly what it is rather than guessing and keeping track of the proper x,z,y axis rotations.

    I am guessing on 1 hour 20 minutes or so, and that is a bit too much, 45 minutes to 60 minutes for full hd res without too much noise and with such volumetrics is probably to expect with a medium level hardware system as mine, had I been
    in the business with a fairly good salery, high end would probably be a good choice, with a much better CPU.

    Edit..
    My last change on the render, that was a bit too much and my expectation of 1 hour 20 minutes has already passed by 30 minutes, aproximated remaining hours will be several mour hours, if this lands on 6 hours or 8? it could very well be the case.
    Probably the increase of bounces I did, and also the change to cpu as well as the higher resolution vdb.

  9. #24
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    17,070
    Nez, for the cloud topic, I already had one thread topic on the WDA disney asset and rendering of that, so I put up some more there and itīs best to talk about the clouds there, and fire here.
    Check general discussions.

    https://forums.newtek.com/showthread...clouds-and-vdb

  10. #25
    Electron wrangler jwiede's Avatar
    Join Date
    Aug 2007
    Location
    San Jose, CA
    Posts
    7,072
    Quote Originally Posted by prometheus View Post
    How Far did you get a simple fire with gas solver? would be interesting just to understand what level you do not think is acceptable and how much far I would need to perhaps show something else.
    If you could upload your worse..but best gas solver sample as a video?
    Sorry for the delayed reply, been a bit busy. I'll upload my "bad" test scene when I get a chance (has to be using LW2019, unwilling to upgrade to LW2020 over quality issues), though from looks of it, it isn't that much different from yours. I'd used a bit lower temp and buoyancy, IIRC, but otherwise similar.

    I looked at the results you posted, and while they're "better", I still find them a lot more like "colored smoke" than "flames" (most notable to me is continuous soft edges and lack of "flame tips", flame plasma edges tend to be more chaotic). To put it differently, if you can recolor it into grayscale and it's a decent "smoke result", it isn't that flame-like. Difficult to describe, when I get a chance I'll try to find some IRL examples to better demonstrate what I'm trying to describe.
    John W.
    LW2015.3UB/2019.1.5 on MacPro(12C/24T/10.13.6),64GB RAM, NV 980ti

  11. #26
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    17,070
    Quote Originally Posted by jwiede View Post
    Sorry for the delayed reply, been a bit busy. I'll upload my "bad" test scene when I get a chance (has to be using LW2019, unwilling to upgrade to LW2020 over quality issues), though from looks of it, it isn't that much different from yours. I'd used a bit lower temp and buoyancy, IIRC, but otherwise similar.

    I looked at the results you posted, and while they're "better", I still find them a lot more like "colored smoke" than "flames" (most notable to me is continuous soft edges and lack of "flame tips", flame plasma edges tend to be more chaotic). To put it differently, if you can recolor it into grayscale and it's a decent "smoke result", it isn't that flame-like. Difficult to describe, when I get a chance I'll try to find some IRL examples to better demonstrate what I'm trying to describe.
    Yes..I think I understand, I just happen to like that longer conintinuous flame style, or torch style, or May fire as we have here in Sweden
    I guess you saw both of the samples?

    For a more fireplace kind of fire, I would need to set some more cooling..til you see the smoke or fire is getting cut off, this will cut of the fire so it doesnīt continue.
    And increase curl scale, which in fact has nothing to do with the size of the curl, but just the strength..to me a bit oddly why they then named it curl scale, that will increase vorticity which is more prominent inaster fire place flames for instance.
    And perhaps additional forces as well.

    Another important part, that would be to set a density texture on the object you use as an emitter, otherwise it sort of yields the same density overall, it would look better I think by not applying it overall to an object...something like I used with early TFD demo versions..long time ago, but that sample also has smoke..and some other factors that would be changed.

    Yuck..lightwave forum still not able to handle vimeo links to embed...
    https://vimeo.com/35832026

  12. #27
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    17,070
    I suppose you want more of this sort? TFD does a nice job with the thermal convectional curling...
    Lightwave Guru sample, that level I think will be tricky with gas solver and even in blender...So TFD has some things going for it..especially this stuff, the smokey GI and multiple scattering, that is very poor though..worse than blender, not usable unless having octane, or an extremely high end fast CPU.

    I would actually liked to have seen this sample a little faster, a little closer to realtime fire..not entirely but a little faster.


  13. #28
    I made this entirely in lightwave with the new gas solver and the vdb tools




    https://vimeo.com/403016607
    Last edited by ary3d; 09-16-2020 at 10:17 PM.
    David Agüero - freelance 3D generalist & concept design
    Modeling, texturing, shading, lighting, animation
    www.davidaguero.com.ar
    vimeo channel
    Free Tutorials

  14. #29
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    17,070
    Quote Originally Posted by ary3d View Post
    I made this entirely in lightwave with the new gas solver and the vdb tools




    https://vimeo.com/403016607
    Great..darn nice, but itīs only smoke, and denser smoke with dissipation falling or some density scale to allow it to fall, the fire dynamics are different and fireshading is a bit tougher in my opinion.
    So the emission channel is where the difficult part is in my opinion to tweak it right.

    But it is tedious to work with gas solver, add null, add mesh, change null to open vdb add gas solver, add mesh to volume, connect the inputs, density and emission grid, set up the gas solver to not have thick sinking smoke,
    Any additional force, extra work with add make velocity grid, add make vector grid, connect those to eachother, then to the force input.

    The emission channels are not easy either to get exact match of what is actually visually seen in density and emission in openGL, too tricky to get it match decently initially, though you may get there.

    Then I go to blender, select mesh, spacebar for quick smoke (or f3 I think for later version) play and you are done, exept for additional force, you just add the force and arrange direction, the levels of steps needed to get going with gas solver fluids is simply too much uninitiuive that I think it is in danger of dying before it got over itīs fetus status.

    They really need to adress the workflow and performance with the upcoming versions, and not leave it in the cradle state, I can mess around with it..but not somethign I want to mess around with, I may actually rather spend
    money on TFD if itīs not improved, or simply go with blender/Houdini and embergen.

    As for the thread starters question, I showed that it could be done with a lot of job, never said it was easy.
    Gas solver has a very nice whispy look to it which I like though.

    Lightwave simulated gas solver exported to blender for blender rendering in cycles..



    And particle velocity force driven, not generated unlike blender and tfd, this is a feature that seems unique to gas solver, even though you can use particles as well as generators for the gas...


  15. #30
    Not so newbie member lardbros's Avatar
    Join Date
    Apr 2003
    Location
    England
    Posts
    6,041
    Quote Originally Posted by ary3d View Post
    I made this entirely in lightwave with the new gas solver and the vdb tools




    https://vimeo.com/403016607
    That's really neat David!!! Not seen much done with the LightWave smoke stuff, but this is lovely. I really like how his eyes are really dissipating through the smoke.
    LairdSquared | 3D Design & Animation

    Desk Work:
    HP Z840, Dual Xeon E5-2690 v2, 32GB RAM, Quadro K5000 4GB
    Desk Home:
    HP Z620, Dual Xeon E5-2680, 80GB RAM, Geforce 1050 Ti 4GB

Page 2 of 3 FirstFirst 123 LastLast

Bookmarks

Posting Permissions

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