Memory use?

Notice how my render time blew out when I focused in on the glass sphere.

Go back to how a ray-tracer works... for each pixel you "render" you run a "ray" back into the scene until it hits something. If that something has a purely diffuse/specular material you simply trace back from that spot to each light (within range) in the scene and you're done. If it's transparent, you need to trace another ray back through the object until you hit something else. If it's reflective you need to trace back along the reflected ray. Glass is both, so you need to do both. Now, this process is recursive ... so when you hit glass you need to start two new rays -- one reflected and one refracted. The refracted ray will almost certainly hit another glass surface (on the way out) and that splits into two as well.

Blurred reflection and transparency are even worse. Each time you hit a blurred surface you need to fire off a bunch of new rays.

Maximum ray depth determines how many times you're willing to recurse before giving up.

So yes, rendering will slow down in specific areas where there are transparent or reflective objects.
 
Just another thought. Wouldn't a bounce feature like Modo and other programs enable you to use say fewer area lights therefore a quicker render time? It seems to me i need to "fill" in so many areas with additional lights it kills the speed.
 
Generally getting stuff to render nicely and quickly involves cheating (e.g. only using area lights or rendering shadows when you have to).
 
Just another thought. Wouldn't a bounce feature like Modo and other programs enable you to use say fewer area lights therefore a quicker render time? It seems to me i need to "fill" in so many areas with additional lights it kills the speed.
You can already control how many bounces you want for the GI solution as well as for the raytrace depth. :)

The thing i really would like to see appear is a "threshold" value, like the one you have in modo and now also in LightWave, which control the value at which a ray doesn't contribute enough to the scene and "die", even if the maximum depth setup hasn't been reached. :cool:

It's the kind of feature that can save a lot of time when you deal with scenes aith high recursion levels needed, like scenes with glasses or lots of mirror surfaces...


Cheers,
Laurent aka Tartiflette :)
 
If you mean increase the radiosity value for the bounce in my experience that does not give the same results as Modo's bounce feature. I'm talking in the way it works and the quality.
No no, i'm talking about REAL "bounce" feature in the radiosity setting ! :cool:

I have attached a screenshot showing what i'm talking about.
In this screenshot, you can see that there are even 2 settings for Cheetah when most of render engines only feature one !
  • The first is the equivalent to modo's bounce feature (and almost all the other render engines...) and allow you to setup the diffuse bounce value.
  • The second one takes into account one of the strength of Cheetah which is the ability to do caustics via the GI engine, leading to some very cool effects. :icon_thumbup:

I hope it answers your question ! :)


Cheers,
Laurent aka Tartiflette :)
 

Attachments

  • bounces.jpg
    bounces.jpg
    42.4 KB · Views: 209
Last edited:
No no, i'm talking about REAL "bounce" feature in the radiosity setting ! :cool:

I have attached a screenshot showing what i'm talking about.
In this screenshot, you can see that there are even 2 settings for Cheetah when most of render engines only feature one !
  • The first is the equivalent to modo's bounce feature (and almost all the other render engines...) and allow you to setup the diffuse bounce value.
  • The second one takes into account one of the strength of Cheetah which is the ability to do caustics via the GI engine, leading to some very cool effects. :icon_thumbup:

I hope it answers your question ! :)


Cheers,
Laurent aka Tartiflette :)

I have tried those and I don't think they are the same? I'm not a programmer but there seems to be a big difference in the way they work.
 
I have tried those and I don't think they are the same? I'm not a programmer but there seems to be a big difference in the way they work.
It works exactly as i wrote it in my previous post, and YES, the "diffuse scattering" setting is exactly the same as bounces in modo or any other render engine featuring this kind of setting. :cool:


Cheers,
Laurent aka Tartiflette :)
 
It works exactly as i wrote it in my previous post, and YES, the "diffuse scattering" setting is exactly the same as bounces in modo or any other render engine featuring this kind of setting. :cool:


Cheers,
Laurent aka Tartiflette :)

Well, you may say that but in practice I have to disagree. I used the Modo demo for a month straight and they don't have the same type of effect on the scene. The Modo bounce function just works the way you would expect it to.
 
Well, you may say that but in practice I have to disagree. I used the Modo demo for a month straight and they don't have the same type of effect on the scene. The Modo bounce function just works the way you would expect it to.
Could you just show something -try to keep it simple- modo does with this "bounce feature" that isn't possible in Cheetah ? :confused:
I'll be really curious to see what you're exactly talking about because i can assure you that technically there isn't any difference. :)
Heck, they even use the "irradiance caching" method to accelerate the GI calculation ! :cool:


Cheers,
Laurent aka Tartiflette :)
 
Could you just show something -try to keep it simple- modo does with this "bounce feature" that isn't possible in Cheetah ? :confused:
I'll be really curious to see what you're exactly talking about because i can assure you that technically there isn't any difference. :)
Heck, they even use the "irradiance caching" method to accelerate the GI calculation ! :cool:


Cheers,
Laurent aka Tartiflette :)

My demo expired over a month ago. You can download a demo from their site if you want to play with it.
 
My demo expired over a month ago. You can download a demo from their site if you want to play with it.
No thanks i already know modo ! :tongue:
What i would like to know is what you see in modo with the bounces we're talking about that is so different from what we have in Cheetah with diffuse scattering. :confused:
You can ask Luxology to extend the demo if you want to play with it a litle bit more ! :icon_thumbup:


Cheers,
Laurent aka Tartiflette :)
 
No thanks i already know modo ! :tongue:
What i would like to know is what you see in modo with the bounces we're talking about that is so different from what we have in Cheetah with diffuse scattering. :confused:
You can ask Luxology to extend the demo if you want to play with it a litle bit more ! :icon_thumbup:


Cheers,
Laurent aka Tartiflette :)

I no longer have the demo on my Mac.

The light level in the entire scene fills in as you would expect in Modo where in Cheetah it does not using the diffuse scatterings.

As i said I'm not a programmer nor am I stupid.:tongue: They don't work the same in practice.

/end topic. :icon_thumbup:
 
Last edited:
Yipee!

I no longer have the demo on my Mac.
/end topic. :icon_thumbup:
Not so fast! :)

I have good news. I changed a bunch of settings, mostly related to the lights, and I got a pretty decent render at 348 secs! I think pod's comment that you have to cheat was essential. Thanks pod! No shadows for any lights, only the spot key light. I think I'm starting to learn that 3D ≠ reality. ;)

@ Francois: Sorry if I aggravated you, I was really frustrated about the hight render times, and I did not understand the importance of proper lighting settings. Thanks for trying to help. Let us bury this ax. Cheetah forevah!

PS: One thing I don't get is why I'm still getting transparency, with only Radiosity checked for the key light. Oh well, I'll worry about that some other day.
 

Attachments

  • fridge-5.jpg
    fridge-5.jpg
    46.2 KB · Views: 188
Last edited:
The light level in the entire scene fills in as you would expect in Modo where in Cheetah it does not using the diffuse scatterings.
That is the part i don't understand or at least i don't want to understand the way you describe it ! :tongue:
I have setup a very small and simple scene describing what effect the scattering setting has for Cheetah and as you can see it has the same effect as the bounce setting in modo or any other render engine... :confused:
I have attached the file so that you can play with it if you want ! :)
As i said I'm not a programmer nor am I stupid.:tongue: They don't work the same in practice.
Sorry if my question sounded like i had the intention to imply that, it was not intended at all. :redface:
The fact is we disagree on that particular thing but that's the purpose of the discussion : to see what's causing our diverging point of vue on that feature/setting ! :smile:
/end topic. :icon_thumbup:
Well i hope you will at least take a look at that file and judge for yourself if the scattering in Cheetah doesn't do the same as bounce setting in modo. :icon_thumbup:

Cheers,
Laurent aka Tartiflette :)
 

Attachments

  • test_scattering.jas.zip
    8 KB · Views: 242
  • no_gi.jpg
    no_gi.jpg
    9 KB · Views: 193
  • scattering_1.jpg
    scattering_1.jpg
    19 KB · Views: 188
  • scattering_2.jpg
    scattering_2.jpg
    26.1 KB · Views: 186
  • scattering_3.jpg
    scattering_3.jpg
    25.3 KB · Views: 186
I post a second message in a row to continue with the attachments of different settings for the scene, as you can "only" have 5 attached files per post. :redface:

So the 4 pictures in the previous post then the 2 in this one show the effect of :
  • No GI at all, only direct light (0,53sec)
  • Diffuse Scattering = 1 (2,71sec)
  • Diffuse Scattering = 2 (4,30sec)
  • Diffuse Scattering = 3 (5,79sec)
  • Diffuse Scattering = 4 (7,33sec)
  • Diffuse Scattering = 5 (8,69sec)

All other setting in the GI tab stay the same for all the renders. :smile:
Sorry for the bad quality, but it was only intended as a quick showcase of the scattering setting in Cheetah3D. :redface:

Hope you'll find that instructive. :icon_thumbup:


Cheers,
Laurent aka Tartiflette :)
 

Attachments

  • scattering_4.jpg
    scattering_4.jpg
    29.1 KB · Views: 182
  • scattering_5.jpg
    scattering_5.jpg
    31 KB · Views: 179
Mr. T,

I've done that test myself. Although I can't test Modo anymore my recollection is that the bounce's are far more pronounced. More like the difference between 1 and 3 for a single bounce. I wish I still had it...oh well. Thanks for your effort.

Here is my "real world" test. This is a WIP the first render has 3 and the second 6 diffuse scatterings. I can barely see a difference.
 

Attachments

  • diffuse3.jpg
    diffuse3.jpg
    58.5 KB · Views: 185
  • diffuse6.jpg
    diffuse6.jpg
    58.3 KB · Views: 187
Last edited:
Well there is something else in your image that makes that image look "flat", that's why you don't notice the difference... :frown:
Do you have something like emissive "ON" in your textures or an ambient set too high ? :confused:

Anyway there is something else, because as you can see from my example the scattering setting has a real effect on the lighting of the scene. :)

The other thing to notice is that there will be far less difference between let's say 4 and 6 "scattering" than there will be between 1 and 3 ! ;)
That's the nature of the algorithm to lesser the effect of each recursing bounce/scatter. :)

As for modo enhancing the effect, you may be right as it already does that kind of things when you have refractive objects : it has a tendancy to "over-play" the effect of "total internal reflection" and add a lot of detail that really don't exist ! :tongue:


Cheers,
Laurent aka Tartiflette :)
 
Back
Top