|
|
Old Glass Insulators
About Truespace Archives
These pages are a copy of the official truespace forums prior to their removal somewhere around 2011.
They are retained here for archive purposes only.
Old Glass Insulators // Work in Progress
Post by stan // Jun 16, 2006, 7:10pm
|
stan
Total Posts: 1240
|
I've been trying my hand at fresnel glass and other vray goodies..here's my first renders of some old glass insulators..there is too much blackness in the glass, but not a bad starting point..renders are taking about two hours..:D |
Post by Shike // Jun 17, 2006, 9:37am
|
Shike
Total Posts: 511
|
Looks great ! :)
Are you using all the VRay goodies here? HDRI and Caustics for example? |
Post by stan // Jun 17, 2006, 10:13am
|
stan
Total Posts: 1240
|
Shike..Thanks..yes caustics is on..GI and HDRI too ..the GI is set to "very low" ..once I get the glass looking right I will try tweaking the other setting some more....:D |
Post by MadMouse // Jun 17, 2006, 11:40am
|
MadMouse
Total Posts: 1069
|
Looking Great Stan. Keep up the good work.
Steve |
Post by Délé // Jun 17, 2006, 3:25pm
|
Délé
Total Posts: 1374
|
I think black outlines in glass material is something that Vray currently has trouble with.
Still looks good though Stan. Will be fun to see more. :) |
Post by W!ZARD // Jun 18, 2006, 2:58am
|
W!ZARD
Total Posts: 2603
|
Cool! This brings back LOTS of memories - I was a lineman for many years and have worked with lots of different line insulators. These look good but should be covered with a lot of lichen and bird guano.
An excellent subject for V-Rays rendering capabilities. Good work. |
Post by Shike // Jun 18, 2006, 3:23am
|
Shike
Total Posts: 511
|
Have you tested the "Internal Reflections" checkbox in the Freznel shader?
Parva started a nice thread that might be useful:Roundtable - Vray tips&tricks :)
(also with link to general VRay tips.) |
Post by stan // Jun 18, 2006, 3:47am
|
stan
Total Posts: 1240
|
Shike..in this render internal reflections are on....it brought the render time up over 12 hours. from 2 hours for the first ones..eek
Wizard..this one was on my house before I renovated..it was rather clean..some people collect them around here to put in the window..
Délé the blackness problem is nearly gone in this render..I think what is there now should be there..
and MadMouse..Thanks guys :D
what's bugging me now is the bit of baby blue in the one on the right near the bottom..not sure what that is??
:cool: Gord |
Post by trueBlue // Jun 18, 2006, 4:43am
|
trueBlue
Total Posts: 1761
|
Very nice there Stan! |
Post by parva // Jun 18, 2006, 4:48am
|
parva
Total Posts: 822
|
12 hours is far too much for this resolution. Now it's difficult to say from first watch what causes this high rendertime. Lights, GI calculation, Material or all of them. I'm sure if we could examine this scene we could halve the rendertime (or more).
The bottle models looks good. Think it would be a great scene together with some medical stuff on a desk or in a cabinet.
OK quick analyse.
Lights - 1 spotlight? + HDRI? or in a "real" 3D environment?
GI - looks like low or similar low GI setting (but guess GI isn't the problem here)
Material - you said Internal Reflect is on - how are the other values? |
Post by Shike // Jun 18, 2006, 4:51am
|
Shike
Total Posts: 511
|
Shike..in this render internal reflections are on....it brought the render time up over 12 hours. from 2 hours for the first ones..eek
Yikes ! I get the feeling that these insulators are a little bit to complex when experimenting with settings ;)
But it seems like you've got VRay under control now, it looks really nice! :D
Hm, the baby blue looks like a rough internal reflection of the windows... I think it should be there ? I like it, and wouldn't even thought about it unless you mentioned it. :) |
Post by stan // Jun 18, 2006, 5:21am
|
stan
Total Posts: 1240
|
Thanks trueBlue..
Parva..as I mentioned earlier render time escalated when I checked internal reflections..that and changed weight from 95 [too much blackness] to 55
GI is the same in all renders.."very low"
HDRI was tweaked but doubt thats what did it..
caustics was the same in all renders..
there is only one regular light in the scene ..
glass insulator model has 7568 faces..
but despite the claims Vray renders very slow..try making bubbles..your render will slow to a crawl..
Shike..Thanks ..the baby blue just seemed odd ..there is no bottom on the insulator, so I thought it might be bad caustics..not really sure.. |
Post by noko // Jun 18, 2006, 6:14am
|
noko
Total Posts: 684
|
High stan,
Looks really good, with HDRI, GI may not even be needed since HDRI will light whole scene. You might want to try without GI, well I am very much curious to tell the truth :). Set sample rate up a little for HDRI and make sure saturation is set at zero, sometimes TS7.1 defaults with saturation of 1 which will oversaturate objects in the scene which makes me wonder if the blue is coming from that with internal reflections. Rendering times do seem excessive. |
Post by stan // Jun 18, 2006, 8:16am
|
stan
Total Posts: 1240
|
noko..Thanks for the suggestions..I had read your post about saturation..it was set to saturation -0.7, samples 10..
as seen from my test image it is HDRI related thought..[so it seems is some of the long render time..}.
the tests are done without GI
changing the saturation didn't work or changing samples..
I was beginning to think it was my bump map..but an area render test showed me it was worse without the bump map..
:cool: Gord |
Post by noko // Jun 18, 2006, 10:59am
|
noko
Total Posts: 684
|
I wished I had some of those niffty insulators laying around on hand. Then I could see if they do that in real life. As a note I've found 50 samples and up for HDRI with VRay to smooth out the image from noise. Still that would increase rendering time even more. I have to check into this more and see if I can duplicate. |
Post by stan // Jun 18, 2006, 11:21am
|
stan
Total Posts: 1240
|
I wished I had some of those niffty insulators laying around on hand. Then I could see if they do that in real life.
noko..the one I have doesn't do that..even in direct sunlight..:D it may just be how the HDRI image reacts on it..the kitchen one..with no bump map the area is still baby blue but it looks smooth so it might be from the window in the HDRI..
here are the HDRI settings |
Post by splinters // Jun 18, 2006, 11:24am
|
splinters
Total Posts: 4148
|
Just an idea Stan but try setting saturation to 0, I believe negative values have an adverse (see opposite) effect on colour saturation-I presume like a negative on a film....worth a try...:confused: |
Post by stan // Jun 18, 2006, 11:28am
|
stan
Total Posts: 1240
|
Thanks Splinters..if you look at my test image just above [post #12] you can see that I tried 0 saturation with the same resultsi ..:) |
Post by TomG // Jun 19, 2006, 3:22am
|
TomG
Total Posts: 3397
|
On Saturation, it works differently for Lightworks and for V-Ray.
For Lightworks, 0 is "normal" - that is, colored naturally from the HDRI image. Above 0 causes extra color saturation. Below 0 removes color, and at -1 there is no color in the lighting and it becomes greyscale.
For V-Ray, 1 is normal, 0 is greyscale, and -1 is reversed colors.
Note also "but despite the claims Vray renders very slow..try making bubbles..your render will slow to a crawl.."
Well, it is entirely possible to slow ANY renderer to a crawl with particular combinations of settings and geometry :) If you are asking a renderer to calculate caustics with bubbles in glass and internal reflections, you are asking it to do a lot - when you render a straightforrward Lightworks glass with bubbles in it, most of those calculations are NOT happening, so you are not comparing apples with apples as you are asking V-Ray to do a whole lot more.
Again note that with caustics, "fast" is a relative term. There is no such thing as "fast" caustics in the sense of seeing similar render times with them as without, they take a lot of math to get there and there is no two ways about it (were it possible to do it close to as quickly as without caustics, then all renderers would always have included them ;) ).
So do bear in mind it will always be possible to bring a renderer down to a crawl, especially with the addition of something computationally expensive like caustics. This will really always come down to raw processor speed, since for accurate effects, there will always need to be a lot of calculations, and (as far as I can tell) there will never be a way to avoid those calculations. We may see processors built that are dedicated to those particular calculations, which means we might see those happen faster, maybe thanks to GPUs on graphics cards - or we may just see ongoing increases in CPU power - either way, the only way caustics will become "fast" is through such hardware improvements.
That leaves us always having to balance render features, quality of the render, and render time, as best we can. Of course even with faster processors and faster render engines, we'll just add more features that we use, so that our renders don't ever seem to get faster - they are though, its just that rather than settle for the 5 minute render with the old features, we add HDRI, caustics, internal reflections, and more, which puts our render times back up ;)
I do think we can improve on that render time with some tweaking of settings though :) It will take experimentation to see what is vital to the look of the scene, and where time can be saved in rendering.
HTH!
Tom |
Post by Alien // Jun 19, 2006, 3:50am
|
Alien
Total Posts: 1231
|
We may see processors built that are dedicated to those particular calculations
You mean like this (http://www.theregister.co.uk/2006/04/21/drc_fpga_module/print.html)? :) |
Post by stan // Jun 19, 2006, 6:17am
|
stan
Total Posts: 1240
|
this render uses GI on "very low" and an image behind the camera to reflect instead of HDRI..hands down HDRI is faster and gives better reflections..I will try another HDRI render today after a few area render tests..:D
Thanks for your insights on HDRI saturation and render times Tom. :) ...the old method took much longer .. raycatcher would have taken 24+ hours, with results that don't come close to vray....the render times don't shock me, it's some of the others who think it should render faster..personally I think it's some of the best glass results I've ever had..worth the wait :D
:cool: Gord |
Post by noko // Jun 19, 2006, 6:22am
|
noko
Total Posts: 684
|
Stan have you tried another HDRI yet? Same result or similar? Also thanks to Tom for Saturation setting differences between LW and VRay, now I know why some of my VRay HDRI renders seemed to be off :confused:. Good stuff and I agree some of the best looking glass I've ever seen generated by a computer. |
Post by stan // Jun 19, 2006, 6:37am
|
stan
Total Posts: 1240
|
noko..I'm just doing a few area render tests with some new settings for HDRI now..the same kitchen image..:)
downloaded a few other HDRI's last night to try later..:banana:
the image I just posted took all night so I haven't had the time to do another HDRI yet.[ I just had to see the difference ..old reflection methods vs. new] .when I start the render it will probably only get posted tomorrow..even area render tests take awhile..:D
Thanks :cool: Gord |
Post by parva // Jun 19, 2006, 11:14pm
|
parva
Total Posts: 822
|
And I still can't understand that the calculation for this needs 2, 12 or even 24 hours. This isn't Maxwell :D
HDRI can't boost the calc time so much with sample rates between 10 and 30.
Caustics would need a really high photon count + a low distance.
Material looks normal and I can't see that there is a Diffuse value used.
Have you checked the Raytrace settings? |
Post by stan // Jun 21, 2006, 6:00am
|
stan
Total Posts: 1240
|
here's another render..noko there is no GI in this one :) ..the blue area looks a bit better..it is definitely part of the HDRI image..I tried other HDRI images but went back to the kitchen one..
there is a minor view angle change..minor tweaking of the positioning of the insulators..lowered the back. intensity to 0.15..with a medium preset..doubled the photons to 6000..removed almost all specular [0.01] ..when I read about that the first time it didn't sink in :D ..but it wasn't the issue..
I noticed prepasses increase even with samples at 10 when using preset medium.. which added to the rencer time..
:cool: Gord |
Post by noko // Jun 21, 2006, 7:07am
|
noko
Total Posts: 684
|
Does look better, I am beginning to believe this is HDRI noise being magnified by castics, ground plan looks to be noisy as well. Which I've found to be the case when using lower sample settings using HDRI with VRay. Stan you might want to try a higher sample setting for HDRI (that is if you have the time), small image test, maybe with just one insulator, to see if this clears up the blue area. |
Post by parva // Jun 21, 2006, 10:51am
|
parva
Total Posts: 822
|
the blotches have few/ nearly nothing to do with the caustics (in fact the caustics in this picture are less visible), they are caused by the HDR image. Espacially HDRI with many stored lightsource positions (indoor) need a very high sample rate.
One of the reasons why separating the GI and Reflect/Refract channel would solve this problem a bit.
Currently one thing to erase the blotches is A: use of as much samples as these blotches appear B: attach a gaussian blur filter at the HDRI and decrease the size of the image (Paintprogram which can read/write HDR/EXR is needed here) use this instead of the original one will blur the GI splotches and smaller map needs less time to load in.
Like noko said trying to render one insulator first and see what you can tweak so far. |
Post by stan // Jun 21, 2006, 11:05am
|
stan
Total Posts: 1240
|
yes I will try another..
Parva you said this..not sure what you mean..One of the reasons why separating the GI and Reflect/Refract channel would solve this problem a bit.
I will try samples at 100 ..set GI at high [unchecked] so it's just HDRI..caustics have been saved so that will shave a little time off :D
I don't have any paint program new enough..just photoshop 5.0 .. |
Post by parva // Jun 21, 2006, 10:45pm
|
parva
Total Posts: 822
|
"Separating the GI and Reflect/Refract" - is currently not possible in trueSpace but guess it will be integrated soon ;)
You would get more flexibility there you can use separate images for GI and Reflection/Refraction. Vray for 3DSMax handles in this way.
All the HDR Images which are taken indoor, I tried (except one or two where just a window is the brightest lightsource) need all many Samples or I get the same splotches like Stan. |
Post by stan // Jun 22, 2006, 9:10pm
|
stan
Total Posts: 1240
|
this render has samples at 35 and "very low" preset {with GI unchecked}..at this point it looks like I should lower the light intensity when I raise the samples for the next render..I will refine the inside of the insulator too..that should help ...
I tried 100 samples and "high" preset, would have taken way too long..:D
this one rendered in 17 hours..
:cool: Gord |
|
