Coca Cola Bottle

[R Reynolds]

Your basic small, glass, Coke bottle.

 

I found that I could noticeably reduce render times by separating the bottle surface into two groups, external and internal. I applied the same material to both groups but I modified the internal material instance by turning off its reflections. Even with a two reflection bounce limit it still made a difference.

[Eric2575]

Boy, that Coke bottle shure looks good. Too bad you didn't leave any soda for me

 

That's a great tip too. Anything that reduces render time is worth Brownie points. Now the question is: "Got Milk?"

[Rodney]

Yeah right... like we are gonna believe that is a CG Coke bottle.

 

Outstanding work Rodger! Wow.

 

Added: The shadow does give away the trick I think... there are clearly two shadow shapes there. I'm not sure what could be done to blend/soften that shadow unless it would be to render the shadow in a second pass and add a Blur Post Effect.

 

I assume that the shadow would render correctly if the bottle was unibody in type. Hmm... could it be possible that flipping normals on the inner bottle might help.

[ddustin]

Roger,

It looks very good..

 

The texture on the landscape behind is more of CG stand out than the models shadow.

 

I've found that Bitmap-plus using a tileable image yields the best ground, background results.

David

[Kamikaze]

Oooo, my favorite soft drink goes AM CG....good job and useful tip also..what was your render time BTW...?

 

Michael

[R Reynolds]

Thanks for the kind words.

 

like we are gonna believe that is a CG Coke bottle

I have found that the key to texturing hardware is adding small bumps and defects to the surface. Perfection is not only over-rated but it's also unrealistic.

 

I assume that the shadow would render correctly if the bottle was unibody in type

It is unibody (see attached image) but the model is divided into two groups of patches with a tweaked material on the internal group. I assume its shadow would be more realistic if I turned on caustics but even I'm not that patient when it comes to render times (see below).

 

The texture on the landscape behind is more of CG stand out than the models shadow

I couldn't agree more; that's why I didn't post to Showcase.

 

I've found that Bitmap-plus using a tileable image yields the best ground, background results

I have yet to see an image that when used this way doesn't look like a repeating tile. Would you be able to point me in the direction of suitable images? Ultimately my background landscapes will have meadow-like grasses but I didn't want to take the render hit in this case (see below).

 

..what was your render time BTW...

10+ hours on an Athlon running at 1.5 GHz; 1024 X 683; 5X multi-pass; no soft reflections; no caustics; the sun is a klieg casting 2 shadows; there are 58 bulb skylights casting no shadows

 

I tried turning on caustics and reduced most of its default values by a factor of 10 but the first pass wasn't done after 5 hrs so I gave up. I'm assuming for my purposes grass and radiosity won't be practical until CPUs all have four embedded processors and A:M code is written to use them all (I'm guessing by 2013 when I turn 60).

 

Although I must admit the chances of a Coke bottle taking centre stage on my CG railroad are pretty slim; thank the gods locomotives are opaque. I pity the person who wants to render Wonder Woman's glass jet.

[Rodney]

Thanks for the response to all the queries thus far Rodger.

Thanks also for the wireframe. That helps us understand.

 

Insights like yours really help us determine what is worth setting up and what isn't in our projects.

 

Rodney

[DanCBradbury]

there are 58 bulb skylights casting no shadows

I recomend you turn all your shadows on to 100% darkness... this will give you a simulated radiosity without having radiosity in the shot. Right now your bottle looks too cg... and i think giving the sky the ability to cast shadows should give it a little more believability.

 

make sure, if you're not already, that you are using one of yves skylight rigs. Those rigs are crazy... he used some sort of formula or something to create each rigs hemisphere. The 25 sky light rig is the best for render time and quality ratios. We're currently working on skylight lighting over in theRadiosity forum.

[ddustin]

Thanks for the kind words.

 

like we are gonna believe that is a CG Coke bottle

I have found that the key to texturing hardware is adding small bumps and defects to the surface. Perfection is not only over-rated but it's also unrealistic.

 

I assume that the shadow would render correctly if the bottle was unibody in type

It is unibody (see attached image) but the model is divided into two groups of patches with a tweaked material on the internal group. I assume its shadow would be more realistic if I turned on caustics but even I'm not that patient when it comes to render times (see below).

 

The texture on the landscape behind is more of CG stand out than the models shadow

I couldn't agree more; that's why I didn't post to Showcase.

 

I've found that Bitmap-plus using a tileable image yields the best ground, background results

I have yet to see an image that when used this way doesn't look like a repeating tile. Would you be able to point me in the direction of suitable images? Ultimately my background landscapes will have meadow-like grasses but I didn't want to take the render hit in this case (see below).

 

..what was your render time BTW...

10+ hours on an Athlon running at 1.5 GHz; 1024 X 683; 5X multi-pass; no soft reflections; no caustics; the sun is a klieg casting 2 shadows; there are 58 bulb skylights casting no shadows

 

I tried turning on caustics and reduced most of its default values by a factor of 10 but the first pass wasn't done after 5 hrs so I gave up. I'm assuming for my purposes grass and radiosity won't be practical until CPUs all have four embedded processors and A:M code is written to use them all (I'm guessing by 2013 when I turn 60).

 

Although I must admit the chances of a Coke bottle taking centre stage on my CG railroad are pretty slim; thank the gods locomotives are opaque. I pity the person who wants to render Wonder Woman's glass jet.

 

Roger,

I've attached the tileable image as well as the bitmap plus plugin.

 

The image looks best if you use the plugin.

 

Let me know if it works for you, even if you do a quick render without the coke bottle.

 

David

lawn512_tileable.zip

[DanCBradbury]

Are you sure you need 58 lights? the 25 light rig is pretty good, and it would greatly bring donw your render times. you should still turn on shadows for all your lights though.

[Korken]

How have you set the Index of Refrection on the inner part of the bottle?

 

//Korken

[R Reynolds]

I've attached the tileable image as well as the bitmap plus plugin. The image looks best if you use the plugin.

Thanks David, I'll give it a try once I'm finished with the bottle cap.

 

...make sure, if you're not already, that you are using one of yves skylight rigs...Are you sure you need 58 lights? the 25 light rig is pretty good, and it would greatly bring down your render times. you should still turn on shadows for all your lights though...

Well it started as one of Yves' skylight rigs. I originally downloaded it in v8 but it didn't import correctly in v10. So out of frustration with this lack of reuseability I individually positioned each light in the chor. (using Yves' x,y,z coords) at which time I also tweaked it to get more even illumination over roughly a quarter mile of landscape. At the time I felt the shadows from 25 lights still had too much granularity but that was before multi-pass capabilties. I'll probably do some more experimentation at some future date. I didn't turn on the skylight shadows since this was just a WIP. When I render for prime time, they'll be on.

 

How have you set the Index of Refrection on the inner part of the bottle?

The same as for the outer group, i.e. the average value I got when I googled "index refraction soda lime glass" (your standard cheap window glass), 1.5.

[DanCBradbury]

lol. google rocks. Someone should tell martin that there are values of ior far grater than 2.

[Stuart Rogers]

I hate to be picky about such a nice model, but... The shadow looks wrong to me. As it stands, the bulk of the shadow is dark with a lighter outer halo. Shouldn't this be the other way around? Here's my reasoning: light rays passing through the middle of the bottle have to pass through two thicknesses of glass, but light rays passing through the periphery would pass through one rather long thickness of glass. The periphery would, therefore, absorb more light than the middle of the glass. That is, of course, an over-simplification - refraction would noticeably complicate the shadows further.

[R Reynolds]

I agree with your analysis Stuart. I assume that turning on caustics (radiosity) would solve this inaccuracy but don't have the patience to find out at this time.

 

And even if the shadows are still wrong we must remember that A:M's demographic priority is people who think character driven animated movies are cool, not those who are turned on by accurate simulations of reality.

[Korken]

Dude, do you know how I get rid of this reflection?

It's not in you bottle so please tell me how you got rid of it!

 

//Korken

[attachmentid=11773]

[Korken]

OMG! I've found what's wrong the reflection is up-side-down.

Does anyone know how to flip it?

 

//Korken

[R Reynolds]

...do you know how I get rid of this reflection...

I'm not convinced that there is anything wrong with the optics your glass, it appears to me to be a refracted image of the sky combined with the reflected image of the table. Try moving the camera up about 7.5 cm, re-centering the glass in the fov and do another render. Then move the camera down about 15 cm, re-aim and re-render. You may find the results instructive.

[DanCBradbury]

In refractive materials that are convex you will have an upside down image of the other side. In your glass you have the sky on the bell of the glass, and the checkard floor at the rim. What you are seeing isnt reflection at all, rather it is refraction.

 

 

that is of course if you have a refraction less than or equal to 2.5... after 2.5 or around there the light rays start getting crazy, and no image will be detactable.

 

 

some references to different angles in IRO.

[Korken]

In refractive materials that are convex you will have an upside down image of the other side.

 

Yes, but if you look threw a reel wine glass, you'll see that it isn't up-side-down.

That is why I get so confused because AM calculates the rays and gets it wrong (?!).

 

And by moving it up and down didn't help. But thx for the try.

 

//Korken

[aaver]

[...]That is why I get so confused because AM calculates the rays and gets it wrong (?!).

 

No, don't think A:M calculates the rays wrong. You would most certainly see that in many other cases if it was indeed a bug. I will try to help you in your own thread, though.

[oakchas]

Rodger,

 

Regardless of the crits... This is one awesome coke bottle!

 

Now, knowing you are a modeler, and "A:M's Demographic priority" You don't intend to animate this bottle... A shame really, I't might be really interesting to see what a Coke bottle might do, if it weren't so busy holding liquid!