 |
 |
 |
|
|
Topic - Salt and Pepper...
|
|
|
|
|
|
|
|
|
Plymouth, UK
667 posts
|
Hi Jeff,
It's looking pretty good. I tend to force myself to lower the reflection and specular for glass, much more than I would expect it to be, and then when everything else is in place, gently increase them until it feels 'just right'.
It would be interesting to see the scene with reflection at around 15% and specular at 6%. If it's already that low, then I'm not too sure ;o)
I'd also have (and again, this is just my own experiments) fresnel around 90%, and transparency around 95%.
Best bet, just keep tweaking until you're happy, but don't worry too much until you have your final scene geometry, texturing and lighting in place, as it will change the feel again.
HTH,
Michael.
|
|
8/24/2006 - 10:56 PM
|
|
|
San Diego, CA USA
2150 posts
|
Thanks Michael!
I messed around with all these settings and did not get much difference in the areas where I was most concerned (those mirror-looking strips on both jars). No matter what I do, I still get those, and maybe they are supposed to be there, but they look very strange to me?
I thought the refraction depth would be the place to fix this, but all that does is decrease the darker areas to make them more transparent or light. I stop getting noticeable improvements there after about 12.
Anything else I can play with that would change the way those mirror strips appear?
If anyone would like to try with my object file, here it is:
http://jeffrutan.com/modo/SaltPepper.lxo.zip
Thanks,
-Jeff
I changed the pepper lid to have only 3 holes instead of 7 like the salt.
The geometry for these pieces is pretty much what I want for final.
The lighting is also close. I want the counter and such to be neutral and attention focused on the lids.
san diego modo user group: www.sdmodo.org
product design+development: www.jeffrutan.com
LEGO fun: www.aerospacebricker.com
business fun: www.timeitnow.com
Message edited by Jeff Rutan on 8/25/2006 - 10:59 AM
|
|
8/25/2006 - 10:45 AM
|
|
|
Plymouth, UK
667 posts
|
Hey Jeff.
Well, you've certainly kept me busy playing this evening (that's a good thing) ;o)
I finally achieved a new version of your file, that is quite different, but I really couldn't say why, except that it loses the inner reflections, but also seems to be a 'different' glass. Allow me to explain.
I basically selected all of the 'inner' polygons of the bottle, and assigned them a different material to the 'outer' polygons. Now, the two materials are identical *except* for the Index of Refraction. The 'inner' glass has 1.5, but the 'outer' has 1.49.
Now, I swear Modo seemed to decide for itself sometimes, just how these would render, and I'm wondering if there's a glitch in there somewhere, but then I could have been getting confused.
Either way, if the 'outer' glass is raised from 1.49 to 1.5, the reflections come back. Strange but true. There's certainly something a bit weird, as I wouldn't expect a 0.01 change in IoR to make that much difference, but I'm sure somebody with better physics than me (or a Lux renderer expert) can maybe shed some light on it.
Anyway, the revised scene is here (The lightprobe image will be in a different location, as I'm on a PC) and the updated render is below:
I hope this helps in some way.
Michael.
|
|
8/25/2006 - 2:07 PM
|
|
|
1883 posts
|
Jeff,
The "chrome shoulder pads" you are seeing is actually the edges/ends of the table top. There is nothing wrong with modo. It actually has to do with your model. You made it very think in the center. If a real salt and pepper was as thick as you model them, you would end up with a refraction that looks like your render. I guess you will have to model it so the inner glass is thinner to mimic a real one.
Michael solution helps as a workaround if you don't feel like modeling it again :)
A few other things I noticed that can speed/improve your render:
1) IV rate is very high. With 202 you can get good results with IV 1. I don't think I ever went more than 2 or 3 in 202.
2) Lower ray threshold to %0.01 to get cleaner glass.
3) Set diffuse to about 0-5% for the glass material.
Hope this help.
|
|
8/25/2006 - 2:55 PM
|
|
|
1883 posts
|
I did a quick test to confirm that it is actually the model... one on the right is the original model, the one on the left is with a thinner glass using the Thicken tool.
|
|
8/25/2006 - 3:18 PM
|
|
|
|
|
Netherlands
3626 posts
|
they look very nice ..Jeff
Greetings!
Corien Klapwijk
|
|
8/28/2006 - 2:49 AM
|
|
|
Plymouth, UK
667 posts
|
Great results Jeff !
FOr the salt, I'd say you need less specular and maybe a bit of transparency (or SSS if you're feeling really keen), and you need to either go smaller for the grains, or larger towards the rock-salt granules.
You've got that glass sorted now though, and the render is looking really nice.
Michael.
|
|
8/28/2006 - 3:05 AM
|
|
|
3 posts
|
The glass still needs work IMO. Glass doesn't have these shiny, white, specular highlights, it makes it look plastic. The only highlights on realistic glass should come from reflected lightsources in the environment, such as bright window light, bright studio light etc.
|
|
8/28/2006 - 3:36 AM
|
|
|
London, UK
7321 posts
|
Glass looks better with these settings:
5% Specular/reflection
100% Fresnel
CE on
|
|
8/28/2006 - 5:45 AM
|
|
|
San Diego, CA USA
2150 posts
|
Thanks guys! I originally thought this would be an easy job, but it turned out to be quite a learning experience for me. My first real attempt at a piece of glass! ;-)
I will try those corrected glass settings soon, but maybe not today since tonight is the inaugural get-together for the San Diego modo user group!!!
-Jeff
mirko: I just noticed your helpful hint was one of your first 2 posts to this forum. Thanks for getting in the spirit right away and welcome to a great user community! There is a Mirko in the NewTek forum, joined in 2003, but no posts (I looked it up because I thought I recognized your user name). Is that you too? I hope you will join in here often and post some of your work too!
Rockmed: I was confused by one of your suggestions... For "IV rate" did you mean the Irradiance Rate should be 1-3 or the Interpolation Values should be 1-3? There are so many of these settings that I just don't understand, but it seems you really know your stuff. Where are you located? I am wondering if you might be interested in presenting to our new modo user group in San Diego some time, even if via web conference? Please send me an email if so (jeffrutan at mac dot com).
san diego modo user group: www.sdmodo.org
product design+development: www.jeffrutan.com
LEGO fun: www.aerospacebricker.com
business fun: www.timeitnow.com
Message edited by Jeff Rutan on 8/28/2006 - 10:12 AM
|
|
8/28/2006 - 7:48 AM
|
|
|
London, UK
7321 posts
|
The Irradiance rate indicates the spacing between the irradiance caching samples. The value you enter is the minimum value, ie the closest range (in pixels) between two samples. The max value is something like five times higher, though the samples will never be further apart than the dots in the first of the irradiance caching prepasses.
|
|
8/28/2006 - 9:34 AM
|
|
|
San Diego, CA USA
2150 posts
|
Well, I tried Captain's settings, and it made the glass all blown out white and murky. I tuned off conserve energy and it cleared up some. I still think the version with the crisp highlight from the directional light looks much better, no?
Here is the result with 5% specular 5% reflection 100% Fresnel and no CE:
-Jeff
san diego modo user group: www.sdmodo.org
product design+development: www.jeffrutan.com
LEGO fun: www.aerospacebricker.com
business fun: www.timeitnow.com
|
|
8/28/2006 - 1:40 PM
|
|
|
London, UK
7321 posts
|
What happens when you turn CE on? It should look about the same, except the transparency will decrease as it goes towards the edge.
What's the roughness setting now? Glass usually looks good at about 5-10%, since it's generally very smooth.
|
|
8/28/2006 - 2:04 PM
|
|
|
San Diego, CA USA
2150 posts
|
Captain,
When I turned on CE, the glass turned all cloudy white and the bright spot got much broader and longer and blown out. I didn't let it complete rendering since it takes 30 minutes even with my G5 Power Mac and my P4 working on the same image at the same time.
My roughness setting for the glass is currently 25%.
I will play some more with this later, but we have our first San Diego modo user group meeting tonight, so no rendering work will get done until tomorrow. I also need to move on to the next job since the client was satisfied with the last output and he has something else for me to get started on. Note that there are also elements in the scene for the client that I have not shown here.
Captain, if you want to play with this scene, you can download the SaltPepper.lwo.zip file (under 200K) using the link in one of my posts above.
-Jeff
san diego modo user group: www.sdmodo.org
product design+development: www.jeffrutan.com
LEGO fun: www.aerospacebricker.com
business fun: www.timeitnow.com
|
|
8/28/2006 - 2:42 PM
|
|
|
Livonia, Michigan - USA
1603 posts
|
Nice job, Jeff!
|
|
8/28/2006 - 2:46 PM
|
|
|
London, UK
7321 posts
|
Quote from Jeff Rutan :
Captain, if you want to play with this scene, you can download the SaltPepper.lwo.zip file (under 200K) using the link in one of my posts above.
I would, but I just don't have the patience to play with render settings on a 1.33 GHz G4. ;)
|
|
8/28/2006 - 2:48 PM
|
|
|
1883 posts
|
Looks really nice Jeff... thanks for the invite but I'm in the UK at the moment... I'm also a bit camera shy :)
As for IC settings, I was referring to Interpolation Value. 1 is fine for this scene.
I added some micro bump to give the glass a little imprefection... here is the scene: http://idisk.mac.com/aalireza-Public/saltpep.lxo
|
|
8/28/2006 - 3:58 PM
|
|
|
San Diego, CA USA
2150 posts
|
Very cool Rockmed! You are awesome! I will check out your settings tomorrow (modo user group meeting tonight). Impressive render time too! Is that a G5 Quad or a Mac Pro? Maybe your settings will help me improve the render speed, but I am really craving one of those new quad processor machines!
The way we do web conferences in our LightWave user group is with one program to show the presenter's computer desktop display on a projection screen and a second program (Skype) on speakerphone for 2-way voice. It works pretty well with fairly decent frame rates and good voice quality, and the presenter can hear questions from the group as long as everyone speaks loud and clear. You would not have to show your face at all! More likely the time difference would cause a problem since it would be 2:30 AM there when our 6:30 PM meetings start! ;-)
Thanks Cszetela!
I hear ya Captain! I don't have the patience to render much on my 1.67 GHz G4 PowerBook either! I thought you were going to get a new Intel computer Mac or PC?
I was getting tired of waiting for my dual 2.5GHz G5 on this scene, so I set up a network render with my 3.2GHz P4/HT WinXP PC using the hacks described by Alan Hastings and others elsewhere in this forum. That worked pretty good almost cutting the render time in half, but it also really heated up the room on hot summer day yesterday with dozens of renders! The workflow was not too bad. My Power Mac and my PC (both over 2 years old) seem to render at about the same speed. I did the modeling on my PowerBook and sent the files to my Power Mac and PC, then kept them rendering continuously all day.
I read that most people were using random buckets for network rendering, but in watching this with my setup, I frequently saw both machines working on the same bucket at the same time (at least half a dozen buckets rendered twice in each 800x600 image). I switched this to Hilbert on both machines with one set to Reverse Order, and that worked best (maximum of 2 buckets rendered twice- since there are 2 threads running on each machine).
I think the problem is that the buckets are only written to disk AFTER they complete rendering, so both machines can start on the same bucket since their initial check shows that bucket is not yet done. Maybe a better implementation would have each machine open the file or record for the bucket at the beginning to keep other machines from selecting that bucket (effectively locking the file/record until it is finished rendering). I have written this type of concurrent code before, so know there are lots of complications, but this is a hack, and I expect Luxology will give us a better solution before too long. Meanwhile, it is still really cool that we can build our own modo super computer! ;-)
-Jeff
san diego modo user group: www.sdmodo.org
product design+development: www.jeffrutan.com
LEGO fun: www.aerospacebricker.com
business fun: www.timeitnow.com
|
|
8/28/2006 - 4:38 PM
|
|
|
London, UK
7321 posts
|
I hear ya Captain! I don't have the patience to render much on my 1.67 GHz G4 PowerBook either! I thought you were going to get a new Intel computer Mac or PC?
I ordered a new machine (home built), but the motherboard I wanted wasn't in stock. I emailed the store a few days ago, and they said they had no idea when to expect them. I just cancelled my order. As soon as I get my refund, I'll order it from a store that actually has it! :P
|
|
8/28/2006 - 7:13 PM
|
|
|
just westa Chicago
2402 posts
|
That lookin' pretty good Jeff. )
deg
http://deg3D.com
|
|
8/28/2006 - 7:41 PM
|
|
|
|
|
|
|
|
|
|
|
|
|
 |
 |
 |
 |
|
|
