Bryce5.com

Registered users

» Forgot password
» Registration

Random image

Aircraft
Comments: 2
congaman2

16.04.2024, 15:17

Home Page

Google ads below

Indirect lighting in bryce MKII+

Indirect lighting in bryce MKII+
Description:	Thinking cap's been on again. I reckon, I've got to grips with the mapping accuracy issue now, however, to get perfect mapping the filters for the lamps need to be rendered at much higher resolution. Which makes the production of the filters rather more time consuming. Again the final render took about two minutes. Which is nice and fast. The filters here took about 7 minutes to render each and that again to "process" in psp8. Seems the taller and thinnner the panoramic the closer to being 360x180 degrees it becomes. A proper 360x180 render is really what Bryce could do with. Then it could make it's own HDR's easily. That's by the by. I know I can convert a cross to a sperical... well, I know Horo can. But that would be six renders for each filter and then the whole idea is getting way too fiddly to be practical. So this is still a bit of a cludge, but an improvement in terms of image registration accuracy I think you will agree.
Added by:	davidbrinnen
Keywords:	davidbrinnen, bryce6.1, indirect, lighting, experiment
Date:	05.14.2008 17:38
Hits:	3894
Downloads:	69
Rating:	0.00 (0 Vote(s))
File size:	49.8 KB

Previous image: Paint Job Test

Next image: Indirect lighting in bryce MKII

Author:

Comment:

Horo
Admin

Join Date: 05.26.2004
Comments: 4721

I just can't put up with your speed at the moment. Thanks for the source. Looks as if I had a busy weekend to look forward to.

05.14.2008 18:12

rashadcarter1
Admin

Join Date: 06.04.2006
Comments: 2610

Oh yeah, that thinking cap is on tight. After carefully reviewing the source files I find that you are no less than brilliant. First, it is a lightning fast render. Once you get your projections rendered. the mapping you have solved so very well especially above.

I think they use this method with game simulations to speed up renders where they bake the light and shadows into the sacene. It allows render from many angles but likely there are multiple maps to choose from.

Great potential with this technique.

05.14.2008 18:34

davidbrinnen
Admin

Join Date: 01.03.2004
Comments: 2224

Thanks yes, that's it, "baking". I read about that somewhere, possibly in connection with games, since I am a big fan of HL2's lighting, which manages to look quite natural along with the distressed textures. With bake method, as I understand it the shadows and light is applied to the texture. Of course, it would be very difficult to achieve this in bryce without modifying the software, but there are strong simularities I agree. The main bug bear I have so far is that the geometry is very precise and the mapping of the panorama not quite as precise so weird little halos of light are prone to appear around object shadows where there is overspill from object light. I also think there could be an advantage in placing a light (with appropriate filter) where the viewing camera is, this will give ambient light for surfaces that are mostly visible to the camera, the drawback I have found with this, since I've regularly used a camera light in many renders is that the edges of models where they are tangential to the viewing angle tend to look a bit grubby. However if the light at the camera is not too strong the overall effect is benificial. Without breaking into the software and rewriting the renderer the best we can hope for is an aproximation, but it seems to me from my few experiments that inperfection itself can add some character to a render and that perfection in a lot of situations is not all that desireable if realism is the aim.

05.14.2008 19:11

Horo
Admin

Join Date: 05.26.2004
Comments: 4721

I had time to look at those ideas of yours and Rashad's. What you did here and what is obviously called "baking" is somehow the nearest you come to an inverse HDRI - meaning an almost infinitesimal sphere with a full HDRI mapped around it that lights an indoor scene from within.

What I have some problem with is: how do you start (hen-egg-dilema)? In order to get that gel, you need to render a room which is lit (without the bouncing) and then you use the gel on a radial to fake bouncing.

What would be needed is a low resolution specular convolved mask (or, since the gel is no HDRI, low pass filtered - strong Gaussian blurred).

05.16.2008 17:26

davidbrinnen
Admin

Join Date: 01.03.2004
Comments: 2224

Light baking as opposed to texture baking. Aye, it's sort of an answer(ish) at the moment - but bit by bit I'm solving the associated problems. The answer to your chicken and egg issue, the light is at the centre of the scene, the scene's materials are modified to be 100% metalic, 100% reflective and specular halo between 250 and 254 full white - not finally decided on that figure yet. Render depth need only be 1 and no shadows required. Render at 1000x1000 takes about 7 minutes. A second render at 100x1000 takes another 3 minutes and these combined and filtered provide the information needed to accurately map that light source for a room(ish) I'm working on a still life at the moment and accuracy with narrow FOV's is definately an issue. But not insumoutable, just slightly different filtering is required. So far I'm not holding with excessive blurring - my experiments so far with blurred filters have been unsatisfactory. The lack of HDRI info is likewise less of an issue since it is more about colour information than light, since the light is being used in the final scene anyway, though it does add another level on top of that by suppressing light in parts and not others. I've not decided yet the type of fall off needed for best results, ball park guesswork as would be used in any render seems to fit the bill so far. Still plenty of interesting experiments to be performed.

05.16.2008 18:45

Previous image:
Paint Job Test

Next image:
Indirect lighting in bryce MKII

[Discord Server]