PDA

View Full Version : Urgent help need with weird transparency problem



mike_stening
02-10-2014, 05:14 AM
Hi

just had a really odd problem crop up i haven't seen before and its probably something really simple im missing but its an urgent job so i need to fix it asap

basically i have a plastic bottle with a colours liquid (drink) inside and then a shrink wrap label (also transparent in places but cant be shown here as it still to be released)
Its a Lucozade sport bottle so you should know the type of label i mean.
With just the bottle and no liquid or label on the transparency is fine, same if i just have the label or the drink.

120011 120012

As soon as i combine any of these the transparency goes dark where they overlap

120013
the liquid i made smaller just to make sure it wasn't a poly intersecting issue


Turning off raytrace occlusion and ambient occlusion lessens the problem but not enough and adding luminosity to the liquid also does nothing.
I've tried increasing the ray recursion and the indirect bounces still to no avail.

ive tried putting a light inside the liquid and that also does very little.

Now i have done versions of this bottle before with no problem but this has got me stumped.

also if i make the liquid solid (not transparent) it shows up fine
120014

so any suggestions?

cheers

mike_stening
02-10-2014, 05:50 AM
also just tried loading items from scene into a clean scene
the default distant light casts a very weird picture on this as the shadow on the back plane looks correct but the transparency says otherwise

120015

but its definitely transparent as you can see at the bottom of the bottle:bangwall:

mike_stening
02-10-2014, 08:40 AM
i think ive narrowed it down to being an issue with the models themselves, as creating a couple of simple boxes and applying the textures to them renders fine

now i just need to find what is the issue with the models

Danner
02-10-2014, 09:08 AM
Make sure you don't have duplicate geometry, caused by using paste command twice. If you merge and unify and nothing is removed then Up your ray recursion limit in the Render Globals /render tab.

mike_stening
02-10-2014, 10:23 AM
ok did unify polys and nothing there
was beginning to think it was duplicate geometry but have just built a new bottle and still same issue
this is ridiculous

Danner
02-10-2014, 01:17 PM
what is your ray recursion limit?

spherical
02-10-2014, 02:49 PM
Well, that's weird. How about Exclude From Volume Stack setting on any/all surfaces? What are the IORs of each?

mike_stening
02-11-2014, 02:40 AM
i set the ray recursion limit to 16 and still no luck, just created a new scene and put the model in and its still doing it
exclude from v stack has no effect
running out of ideas here
IORS are 1.33 and 1.5
going to try rebuilding the bottle but a less dense mesh to see if its that that is causing it
got a job to get done and this is killing it
last option is to uninstall and do a clean install or try booting into windows and trying there.

Danner
02-11-2014, 03:13 AM
Have you tried using the dielectric shader? Other things to try: turning on/off "raytrace transparency" in render tab and "Use transparence" in global illumination tab, also self shadow. Also note that VPR and final render do not share the same ray recursion limits.

mike_stening
02-11-2014, 03:31 AM
not tried dielectric but that still wouldn't explain the issue i have
tried all the turning on and off of transparency and refraction
nothing seems to work
am just re installing lightwave now,fingers crossed

mike_stening
02-11-2014, 04:04 AM
re installed and it seems to be working now, though i do need the ray recursion limit very high still but it only darkens a bit when its at default, not the pure black issue i was having
at least i wasn't going mental

spherical
02-11-2014, 05:26 AM
not tried dielectric but that still wouldn't explain the issue i have

I don't see how you could say that for certain. Glad you got some improvement on that which you have, though. Nothing worse than a brick wall.

lardbros
02-11-2014, 05:39 AM
Definitely looks like a VPR issue to me... it has a limited ray recursion unless you turn off draft. Even then, I'm not sure it's follows your render settings.

Did an F9 render fix it? Or did you simply re-install and everything now works fine?

mike_stening
02-11-2014, 07:29 AM
I don't see how you could say that for certain. Glad you got some improvement on that which you have, though. Nothing worse than a brick wall.

the dielectric shader did exactly the same thing, looks like part of my installation was screwed.
all seems to be behaving nicely now

mike_stening
02-11-2014, 07:32 AM
Definitely looks like a VPR issue to me... it has a limited ray recursion unless you turn off draft. Even then, I'm not sure it's follows your render settings.

Did an F9 render fix it? Or did you simply re-install and everything now works fine?

F9 F10 renders also came out black, creating new clean scenes and new geometry also didn't work

complete new install fixed it so corruption somewhere

VPR is also behaving well now, but its a day lost which is very annoying.

spherical
02-11-2014, 04:07 PM
the dielectric shader did exactly the same thing, looks like part of my installation was screwed.
all seems to be behaving nicely now

Ah. Good to know. Thanks for the report back.

lardbros
02-12-2014, 06:22 AM
I've had some strange issues with VPR failing completely, but a deletion of configs seemed to fix it completely! Rather than a total re-install, configs is always a good first step.

I wish Newtek would work on some kind of config tool, where you could remove certain aspects of it via checkboxes. Similar to Google Chrome and its history deletion. It would be lovely simply to delete the obvious bits, but leave the Menu and plugin configs alone.

mike_stening
02-12-2014, 07:37 AM
I've had some strange issues with VPR failing completely, but a deletion of configs seemed to fix it completely! Rather than a total re-install, configs is always a good first step.

I wish Newtek would work on some kind of config tool, where you could remove certain aspects of it via checkboxes. Similar to Google Chrome and its history deletion. It would be lovely simply to delete the obvious bits, but leave the Menu and plugin configs alone.

but in this case it wasnt just VPR failing, the full render was doing the exact same thing

lardbros
02-12-2014, 11:56 AM
Yeah... very weird!!

mike_stening
02-13-2014, 02:52 AM
still having transparency issues though even with ray recursion maxed at 64, have got a version of the bottle with water drops on it and where some of them are over the transparent parts of the label they are going black where as on the solid parts of the label they are perfectly fine.
I would be surprised if a setting of 64 is really needed but any less and i have black water drops on the bottle, also raytrace occlusion cant be on either for the same reason.

Danner
02-13-2014, 03:44 AM
Could you share a simplified model and scene that has this problem so we can look at it?

mike_stening
02-13-2014, 03:54 AM
Could you share a simplified model and scene that has this problem so we can look at it?

just in the middle of doing a render then i'll package it up and remove the Copyright sensitive parts.

mike_stening
02-13-2014, 03:58 AM
here's a screen shot of a low res full render so you can see what i mean

120159

mike_stening
02-17-2014, 05:26 AM
ok heres the scene packaged up, i've just blurred out the sensitive stuff in the images so i dont get sacked.
can others see if they get the dark transparency issues where the drops are over the transparent label/bottle/liquid.
RRL is 32 but setting it maxed out to 64 made no difference.
would be very intereted if anyone can get rid of the dark drops and whats causing it.
Ive tried different HDRI maps just in case there were dark areas in that causing it but even fairly bright maps still saw black in the drops.
Here you can clearly see where the drops are clear over the opaque section of the label and black over the transparent
120222

over to you guys if you think you can help

http://bcl.ftpstream.com/136004/d9bf439e8105b1d604081e2a4bd326b2/Lucozade%2bbottle.zip

lardbros
02-17-2014, 05:31 AM
Has the package been linked to the image? I'll have a look tonight if I can download it. :D

mike_stening
02-17-2014, 05:36 AM
Has the package been linked to the image? I'll have a look tonight if I can download it. :D

sorry had issues attaching the compressed file so have put up a link from our ftp with it in , but cant have it on there too long, be great any help i can get as this has me stumped.
like i say looks great on the solid areas of the label just crap on the transparent areas

lardbros
02-17-2014, 06:27 AM
Just a quick thought... Are you using GI? And if so, do you have 'use transparency' checked in the GI settings bit?

Not sure if it could cause that... but thought I'd ask. :D

mike_stening
02-17-2014, 07:31 AM
Just a quick thought... Are you using GI? And if so, do you have 'use transparency' checked in the GI settings bit?

Not sure if it could cause that... but thought I'd ask. :D

use transparency is not checked as as i understood it that forced GI to use a legacy format and was not as accurate as having it turned off. I'll check that though now you mention it.

lardbros
02-17-2014, 08:09 AM
I thought it meant that GI rays would be traced through transparent objects instead of treating them as solid... (alot slower though)
would be good to get an experts opinion on this though, as I probably presumed this.

mike_stening
02-17-2014, 08:17 AM
it may be just the manual is unclear, it states:
"The purpose of Use Transparency is to disable transparency calculation, including clip maps, for backdrop radiosity rays. Backdrop radiosity is supposed to be a fast and simple form of global illumination. Having it do a lot of transparency calculations makes the image slightly higher quality but if you want higher quality you will probably be using one of the other modes anyway. This option is only available for Backdrop type radiosity. When Use Transparency is disabled for Final Gather, all transparent surfaces are ignored by rays. This acts the same as putting all the transparent surfaces into an object and setting it to be unseen by radiosity."

so to me that says if you tick it you turn of GI for transparency, though i could have completely missinterpreted this.

i'm doing a test now with it checked and it is god awful slow so maybe i was wrong, we'll see how those dark drop come out.

lardbros
02-17-2014, 08:23 AM
Hmmm, sounds very contradictory to me! Only available for backdrop radiosity? It isn't here.
Anyway... will be interesting to see what happens, if anything! I'll still have a peek when I get home though! :D

mike_stening
02-17-2014, 09:07 AM
cheers fella, the test i'm doing at the moment in a limited region doesn't look much better so far but is really really slow compared to how i had it before which wasn't particularly fast in the first place :rolleyes:

dee
02-17-2014, 09:14 AM
Did some quick changes:
- turned on sRGB
- changed "clear plastic bottle" to dielectric (pure white, absorbtion 1, refraction index 1.575)
- removed Fast Fresnel

I would say you need to rework your materials and lighting.

First is your scene, second is my changes, third is my changes and another HDRI.

mike_stening
02-17-2014, 09:37 AM
Did some quick changes:
- turned on sRGB
- changed "clear plastic bottle" to dielectric (pure white, absorbtion 1, refraction index 1.575)
- removed Fast Fresnel

I would say you need to rework your materials and lighting.

First is your scene, second is my changes, third is my changes and another HDRI.


cheers dee, tbh ii was quite happy with the bottle and label and i felt i had more control over the surface rather than use a material, what changes did you make to the water drops as that was the only area that had a real problem.

mike_stening
02-17-2014, 10:42 AM
changing to sRGB has made the image overall a bit washed out and lost punch
i changed the bottle to dielectric and the dark spots have gone now to be replaced by blue,
its almost as if the water drops are becoming 100% reflective, as you can see from the long blue drop at the top on the clear plastic area and just below that.
120232
use transparency was on for this but directional rays was off as that was killing it.
i didn't change the settings on the water drops at all but this doesn't explain why the black areas have gone from the drops,
the bottle label is still a standard surface so the only surface to have changed is the bottle to dielectric, doesn't make sense to me.

dee
02-17-2014, 10:57 AM
I didn't change the drops. Turning on sRGB did a lot.

mike_stening
02-17-2014, 11:03 AM
I didn't change the drops. Turning on sRGB did a lot.

yeah i can see that but it doesn't explain anything, why would changing the colour space effect how the transparency and refraction is calculated??
also changing the bottle plastic to dielectric sucks alot of the light out of the liquid itself along with the sRGB change.

and things still aren't behaving how you would expect as they still different from being on an opaque surface to a transparent surface (water drops) even with these changes

dee
02-17-2014, 11:07 AM
changing to sRGB has made the image overall a bit washed out and lost punch

You can easily change that in PS.

dee
02-17-2014, 11:09 AM
yeah i can see that but it doesn't explain anything, why would changing the colour space effect how the transparency and refraction is calculated??
also changing the bottle plastic to dielectric sucks alot of the light out of the liquid itself along with the sRGB change.

and things still aren't behaving how you would expect as they still different from being on an opaque surface to a transparent surface (water drops) even with these changes

That's why I said you need to rework the surfaces, going from linear to sRGB has a huge impact.

lardbros
02-17-2014, 12:08 PM
The way colour space affects an image is quite profound. Unfortunately reflections are one of the things that look really quite different when using LCS (linear colour space). Motion blur, DoF and plenty of other things can change dramatically under LCS workflow... but it will be more accurate.

A big issue with the way LW's dielectric shader works is that the lights aren't realworld. So, the brightness of the lights aren't correct in specular highlights etc. There are ways to boost this with Nodes from DB&W I think... but as dee says, you could sort it in post.

XswampyX
02-17-2014, 01:24 PM
I think you have a problem with you transparent label wrap?

Here it is with the inner surface deleted and a polygon side node used to get the 'inside' of the wrap. Ray recursion limit is 'only' set to 12 and I can't see any black bits.

http://i465.photobucket.com/albums/rr16/xXswampyXx/Lucozade_Bottle_tonemapped_zps423b62ef.jpg (http://s465.photobucket.com/user/xXswampyXx/media/Lucozade_Bottle_tonemapped_zps423b62ef.jpg.html)

Nice model BTW.
Hope it helps.

Cheers!

dwburman
02-17-2014, 10:35 PM
Another possible approach could be to render the parts (and/or render buffers) separately and combine them in post. I'm not exactly sure how or if that would work in this case, though except perhaps for the initial problem that got cleaned up with the reinstall.

mike_stening
02-18-2014, 01:49 AM
That's why I said you need to rework the surfaces, going from linear to sRGB has a huge impact.

thats what i hoped you meant not that they were just crap in general.

mike_stening
02-18-2014, 01:52 AM
The way colour space affects an image is quite profound. Unfortunately reflections are one of the things that look really quite different when using LCS (linear colour space). Motion blur, DoF and plenty of other things can change dramatically under LCS workflow... but it will be more accurate.

A big issue with the way LW's dielectric shader works is that the lights aren't realworld. So, the brightness of the lights aren't correct in specular highlights etc. There are ways to boost this with Nodes from DB&W I think... but as dee says, you could sort it in post.

i thought that was the point of LCS being introduced and hence why i do all my work in LCS, am i wrong in doing that?
every day is a school day.

mike_stening
02-18-2014, 01:59 AM
I think you have a problem with you transparent label wrap?

Here it is with the inner surface deleted and a polygon side node used to get the 'inside' of the wrap. Ray recursion limit is 'only' set to 12 and I can't see any black bits.

http://i465.photobucket.com/albums/rr16/xXswampyXx/Lucozade_Bottle_tonemapped_zps423b62ef.jpg (http://s465.photobucket.com/user/xXswampyXx/media/Lucozade_Bottle_tonemapped_zps423b62ef.jpg.html)

Nice model BTW.
Hope it helps.

Cheers!

cheers swampy, so how do you do that with the nodes then, i'm not up to speed with nodes still.
the only problem i see with your fix is the whole label looks a bit transparent. and the liquid needs to be slightly cloudy as well.
did you not change the bottle plastic surface then?
but overall your render looks sweet (why do mine never look like that?)
a bit more info would be good

spherical
02-18-2014, 02:35 AM
I've been working with your scene, trying to understand the how and why.

First question is: Why is the bottle (and everything) so huge?
Second question is: Why are the models not frozen in Modeler, thus requiring Layout to freeze them at each render?
Third question is: Can 50% GI Multiplier be used to cut down the computation time, without compromising the quality?

mike_stening
02-18-2014, 02:53 AM
I've been working with your scene, trying to understand the how and why.

First question is: Why is the bottle (and everything) so huge?
Second question is: Why are the models not frozen in Modeler, thus requiring Layout to freeze them at each render?
Third question is: Can 50% GI Multiplier be used to cut down the computation time, without compromising the quality?

1. the model is that size as the original was created in cinema 4d which i then reworked in LW to this version using the original as a guide. It also means you can effectively zoom
in to details further than with a real world scale model where you hit the zoom limit in modeler when working on small changes

also with the water drops being created in Realflow, it doesn't behave well with very small objects i.e. the particles cant create small drops/details, so to get the small details the objects used need to be over sized

2. I never freeze models in modeler which makes any amends to the model easier than dealing with a huge mesh, its just the way i work.

3. as far as i'm aware that will only change the GI computation time and not really effect the overall render time.
You loose a bit of GI accuracy but it shouldn't be that noticeable. I just always leave it at 100%

mike_stening
02-18-2014, 04:39 AM
just trying out a test with the model, where the label object inner surface extends into the bottle object slightly. This is recommended by cinema 4D when creating say a glass with fluid in it so that the fluid object extends slightly into the glass.
i'm currently running the render test now but it is comparably slow with having directional rays turned on. Interesting

lardbros
02-18-2014, 06:44 AM
I always keep my objects as sub-d's, means you can have a character as a proxy automatically... set the sub-d's to 0 for animating, and then 6 or whatever for rendering. Nothing wrong with not freezing meshes is there? It may add a bit of time to pre-process the render, but never bothered me too much either.

mike_stening
02-18-2014, 07:39 AM
I always keep my objects as sub-d's, means you can have a character as a proxy automatically... set the sub-d's to 0 for animating, and then 6 or whatever for rendering. Nothing wrong with not freezing meshes is there? It may add a bit of time to pre-process the render, but never bothered me too much either.

:agree:

current test in progress, things are looking better but the render time is about 10 times slower, this all by just intersecting the objects slightly.



could the original problem be caused by co-planar polygons from different parts (layers) of the object?

lardbros
02-18-2014, 07:56 AM
I've seen black parts of transparent/dielectric materials which have been caused by intersecting or co-planar geometry before... and usually if there aren't enough ray recursions, the renderer tends to return the colour black.

Maybe co-planar is the issue? Hmmmm... still haven't had a chance to download and test your model, if it's still up on the ftp I'll try and have a look tonight quickly.

Andy Webb
02-18-2014, 08:47 AM
Hi Mike,

Just looking at your scene file, why have you got both Fast Fresnel and gradient Incidence angle on your water spritz, surely they both are trying to do the same thing?

If I take all the gradients off of diffuse, spec and reflections but leave Fast Fresnel the black disappears, alternatively leave the gradients on and remove the Fast Fresnel.

Unless I'm not getting your problem correctly?

Cheers Andy

mike_stening
02-18-2014, 08:50 AM
yeah its still there, trying a few other modeling options as i have a feeling its there where the issue is in the raytrace calculations

mike_stening
02-18-2014, 09:05 AM
Hi Mike,

Just looking at your scene file, why have you got both Fast Fresnel and gradient Incidence angle on your water spritz, surely they both are trying to do the same thing?

If I take all the gradients off of diffuse, spec and reflections but leave Fast Fresnel the black disappears, alternatively leave the gradients on and remove the Fast Fresnel.

Unless I'm not getting your problem correctly?

Cheers Andy

cheers for looking andy, fast fresnel is there but not turned on so shouldn't make any difference anyway

though still doesn't explain why the spritz on the opaque areas of the label are just fine

Andy Webb
02-18-2014, 09:17 AM
Well Fast Fresnel is turned on on the bottle and the spritz here!

I left the Fresnel shader and turned off the gradients.

mike_stening
02-18-2014, 09:25 AM
maybe i have turned it off since uploading it, sorry
kind of got rid of the black drops issue but now they look super reflective in places and not transparent

Andy Webb
02-18-2014, 09:31 AM
You can't win :D

mike_stening
02-18-2014, 10:05 AM
You can't win :D

it certainly feels that way, feel like i haven't got a clue at times like this

mike_stening
02-18-2014, 11:08 AM
ok so i unchecked the double sided option for the spritz and now the spritz over the transparent areas looks ok, but
the spritz over the opaque areas is now dark @$%%&*&@

120250

XswampyX
02-18-2014, 05:00 PM
Here's the node set up for the dual sided surface. It looks complicated because you also have a mask in there.

http://i465.photobucket.com/albums/rr16/xXswampyXx/Lzade_Nodes_zps241097a5.jpg (http://s465.photobucket.com/user/xXswampyXx/media/Lzade_Nodes_zps241097a5.jpg.html)

And my scene..... The setting for the renderer and camera have been lowered for the sake of speed. I get around 5 mins. :)

XswampyX
02-18-2014, 07:45 PM
Just for fun. :D

http://i465.photobucket.com/albums/rr16/xXswampyXx/Lucizade_Octane_zps841bd06d.jpg (http://s465.photobucket.com/user/xXswampyXx/media/Lucizade_Octane_zps841bd06d.jpg.html)

In Lightwave with the octane plug in.

mike_stening
02-19-2014, 02:02 AM
cheers swampy i'll have a look at that now, god knows whats going on with my scene now

Andy Webb
02-19-2014, 03:57 AM
You've reached that point where what ever you do nothing looks right...

mike_stening
02-19-2014, 04:06 AM
yeah feel like that Andy, almost back to the drawing board
I've taken some leads from the generous swampy, got some other changes i need to do then I'll post where I'm at

spherical
02-19-2014, 05:26 AM
things are looking better but the render time is about 10 times slower, this all by just intersecting the objects slightly.

could the original problem be caused by co-planar polygons from different parts (layers) of the object?

That was going to be my conclusion, yes. May work in C4D to intersect but it's been my experience that LW doesn't like intersecting geometry; especially when it's transparent. Having polys co-planar works when using materials as part of the Volume Stack. Unsure of how they do with the "old style" of obtaining correct refraction in transparents in the current version. Exclude From Volume Stack controls how these react.

Thanks for the explanation of your workflow. Always something to learn. I found the render time to be problematic, however. Do an F9 and it goes through all of the freezing of multiple objects, then computes the GI. Quit the F9 and it goes back (says "freezing" but it's unfreezing). Takes a lot of time sitting there just watching the throbber spin before you can run the next test. Then, it does it all over again.

Original: @ 1754x2480
GI: 6m 42s
Total Render: 4h 32m 33s
RAM: 1,649.6 MB
120266

Using my convention, your surfaces: @ Same Resolution
GI: 2m 28s
Total Render: 55m 33s
RAM: 903.2 MB
120267

Using my convention, Sigma2 for the cap and liquid, Dielectric for bottle and spritz: @ Same Resolution
Preprocess: 34.7s
GI: 18.2s
Total Render: 20m 29s
RAM: 903.2 MB
120268

Like you say, just a different way of working. To each his/her own. I know that there are some benefits to leaving objects unfrozen and let Layout freeze them; just have to figure out where and when to apply the method and accept the tradeoffs.

One thing to consider when working with transparents/translucents, especially when using materials, is that having a bottle that is 20' tall also makes the walls proportionally that much thicker. This will alter the absorption and transmittance and adjustments will have to be made to deal with something that is that thick.

mike_stening
02-19-2014, 05:37 AM
some nice changes there spherical. like the use of the sigma shader, though i've never had much luck with the SSS shaders.
Yeah totally agree with trying to do things to actual size but there are some instances like this where its more practical not to.
Realflow being the main key here in that getting the right resolution in there is very object size dependent, trade off between particle resolution smaller object or max particles with a larger object. the latter seems to work better and not slow down too much.
also the larger object size allows you to get a finer spritz without the poly size going stupidly low. I've found getting small individual drops difficult at a 1:1 scale, going large has the same "size" drops just on a bigger model hence they look smaller.

Yeah the freezing unfreezing is a little annoying but i'd rather have a model i can retopo fairly easily than wade through a heavy mesh, let the renderer freeze it to a higher density for rendering.

this has turned into a nice study of how transparency interacts in LW, certainly not as straight forward as i would have liked but then i've never had so many transparent objects overlaying each other.

spherical
02-19-2014, 05:57 AM
My scene is all scaled down to real-world size.

For retopo, etc., I keep an unfrozen source.

mike_stening
02-19-2014, 07:44 AM
My scene is all scaled down to real-world size.

For retopo, etc., I keep an unfrozen source.

my point was matching the drop mesh up to the bottle hence the large size, i suppose you could scale down but i've been making changes on the drops so scaling down would be a pain to match up.

mike_stening
02-19-2014, 08:56 AM
we're making progress, adjusted textures and lighting slightly and now i have this
120269

which is much better, though on certain areas like the yellow and red the drops almost disappear so i'm working on getting a bit of detail there

lardbros
02-19-2014, 09:33 AM
Looking lovely there though Mike! Way more realistic in my opinion... and very nice indeed!

mike_stening
02-19-2014, 10:00 AM
cheers lardbros, very grateful of the help i had and nice re-introduction to nodes, i really should use them more.
the vertical streaking i don't like but that's the reflection from the HDRI, a bit of surface reflection blur should sort that but its bumped the render time way up

like i say I'd like to get the drops showing up over the yellow lettering and red of the label, playing with diffuse values of the standard node and hopefully that works.
any suggestions on that would be great fully received :rock:

spherical
02-19-2014, 09:20 PM
Unless the HDR is supposed to be in the final shot, I would strongly suggest blurring it, instead. Will cut down significantly on the AA/AS passes trying to clean up that image.

Things to save a lot of time, apart from freezing the models that is:

With the use of Material nodes, you can drop the RRL to 13 from the 32 you had.
50% on the GI Multiplier. You're using MC Interpolated anyway. Saw no difference.
Use DB+W Material Booster node from LightWolf to pump up surface effects in materials that you think need it. (I got the orange liquid a bit too bright.)
Be sure to turn off Double Sided in any surfaces that you use node Materials on.
Disable/remove Fast Fresnel on them, too.
Use the power of Unified Sampling. Set Shading Samples to 1 from the 4 you had. Set Light samples to 8 from the 4 you had. This eliminates a lot of the noise that the AS had to clean up.
I killed the AA Max Samples Limit to 16 and set the Threshold up to 0.1. At this scale, 32 and 0.05 just seemed overkill and a waste of time. Even though the 0.05 would likely stop its computation before reaching 32, nothing really stuck out as needing that level of polishing.

mike_stening
02-20-2014, 02:27 AM
Unless the HDR is supposed to be in the final shot, I would strongly suggest blurring it, instead. Will cut down significantly on the AA/AS passes trying to clean up that image.

Things to save a lot of time, apart from freezing the models that is:

With the use of Material nodes, you can drop the RRL to 13 from the 32 you had.
50% on the GI Multiplier. You're using MC Interpolated anyway. Saw no difference.
Use DB+W Material Booster node from LightWolf to pump up surface effects in materials that you think need it. (I got the orange liquid a bit too bright.)
Be sure to turn off Double Sided in any surfaces that you use node Materials on.
Disable/remove Fast Fresnel on them, too.
Use the power of Unified Sampling. Set Shading Samples to 1 from the 4 you had. Set Light samples to 8 from the 4 you had. This eliminates a lot of the noise that the AS had to clean up.
I killed the AA Max Samples Limit to 16 and set the Threshold up to 0.1. At this scale, 32 and 0.05 just seemed overkill and a waste of time. Even though the 0.05 would likely stop its computation before reaching 32, nothing really stuck out as needing that level of polishing.


if i knew how to open and blur a HDRI without loosing the extra light info i would, reflection blur will have to suffice unless i can do it. ;) ignore that stupid comment photoshop does it just fine

1: ive dropped the RRL down to 18 (in that last render) but will probably drop again
2: I'll try that as any saved time would be good, especially as i have to render this pretty large.
3: I'm not sure exactly what you mena by this could you explain further.
4: It's off :D
5: Same again
6: I'll gie this a try and see how much time it saves
7: I have lowered the AA AS down already anyway

thanks for the pointers i'll see how they go.

spherical
02-20-2014, 03:36 AM
3: I'm not sure exactly what you mena by this could you explain further.

Go to db-w.com (http://db-w.com/products/dbwtools/docs). This set of nodes allow you to modify the otherwise set-in-stone Material nodes in order to attenuate or amplify attributes.

mike_stening
02-20-2014, 03:57 AM
Go to db-w.com (http://db-w.com/products/dbwtools/docs). This set of nodes allow you to modify the otherwise set-in-stone Material nodes in order to attenuate or amplify attributes.

ok i'll take a look, thanks

note:Material Tweaker isn't available for mac by the looks of it as its not in the node list

spherical
02-20-2014, 05:33 PM
Booster is what I used. Is it in there? Tweaker is a recent addition but, unless you are on an older OSX, there doesn't seem to be a difference between the contents for OS versions. Did you get v2.1 or v1.8? Seems that there needs to be an update on the website, as the links that are in the text (http://www.db-w.com/component/remository/LightWave-Plugins-%28free%29/) go to the latter, while the link to the former is in the Current Software sidebar and the Download tabs at the top.

mike_stening
02-21-2014, 02:18 AM
Booster is what I used. Is it in there? Tweaker is a recent addition but, unless you are on an older OSX, there doesn't seem to be a difference between the contents for OS versions. Did you get v2.1 or v1.8? Seems that there needs to be an update on the website, as the links that are in the text (http://www.db-w.com/component/remository/LightWave-Plugins-%28free%29/) go to the latter, while the link to the former is in the Current Software sidebar and the Download tabs at the top.

1.8 as i went through the text links, i'll have a look through the downloads
i've got it to a good level now so i dont want to and dont have the time to mess with it too much more.
doing a hi res A3 render at the moment so far taken 15 hours got 3 hours left :D

XswampyX
02-22-2014, 04:50 PM
Are you going to post a render Mike? Would love to see it.

mike_stening
02-24-2014, 02:04 AM
yeah i will do just doing some retouching on it unless you just want to see it raw

mike_stening
02-24-2014, 04:41 AM
here we go with a little bit of colour work
i added two different layers of spritz in seperate renders then blended the 2 together
i've had to blur out the promo part which ruins it a bit but i have to follow copyright at this point.:tsktsk:
i still feel the spritz could do with more work but its a massive fine tuning task in real flow to get it just right and then getting the surface settings right in LW

120387

interested to know your opinions (good or bad)

update, this isnt the chosen version as they want less spritz, i'll update that when it goes to print

erikals
06-28-2014, 09:14 AM
http://forums.newtek.com/images/misc/quote_icon.pngXSwampyX

122613


In Lightwave with the octane plug in.

Stunning...! http://erikalstad.com/backup/misc.php_files/king.gif

did you have to change the node setup when rendering with Octane ?

mike_stening
06-30-2014, 03:03 AM
just a quick update with my most recent version of this, i think i'm slowly getting there.
Spritz could do with a little more definition but its such a fine balancing act with lighting/reflections/refractions.
oh rendered just in LW
122647

Waves of light
07-03-2014, 03:47 AM
Coming to the party a bit late, but this has been an excellent thread. Good progress Mike!

And this is excellent information Spherical, never thought to blur HDRs to reduce render times!


Unless the HDR is supposed to be in the final shot, I would strongly suggest blurring it, instead. Will cut down significantly on the AA/AS passes trying to clean up that image.

Things to save a lot of time, apart from freezing the models that is:

With the use of Material nodes, you can drop the RRL to 13 from the 32 you had.
50% on the GI Multiplier. You're using MC Interpolated anyway. Saw no difference.
Use DB+W Material Booster node from LightWolf to pump up surface effects in materials that you think need it. (I got the orange liquid a bit too bright.)
Be sure to turn off Double Sided in any surfaces that you use node Materials on.
Disable/remove Fast Fresnel on them, too.
Use the power of Unified Sampling. Set Shading Samples to 1 from the 4 you had. Set Light samples to 8 from the 4 you had. This eliminates a lot of the noise that the AS had to clean up.
I killed the AA Max Samples Limit to 16 and set the Threshold up to 0.1. At this scale, 32 and 0.05 just seemed overkill and a waste of time. Even though the 0.05 would likely stop its computation before reaching 32, nothing really stuck out as needing that level of polishing.

mike_stening
07-03-2014, 05:26 AM
cheers Waves, found a little trick in Realflow to get the drops to sag a little bit and make more tear drop like shapes, and tweaked the lighting and surfaces a touch too.
client has come back saying the drops dont stand out enough so i've put a gradient on the colour channel of the dielectric material to darken the edges a little i'll post that in a bit.