Bufshadow Problem


(Umbroth) #1

Hi all,

i have a problem with Bufshadows. Bias and white edge!! :frowning:

A solution? Thanks a lot

Umbroth

Attachments



(Fligh) #2

Your main prob is the low Bias setting, but your ClipSta and ClipEnd could get closer to the objects.

%<


(traitor) #3

Wrong forum too, this should be in Blender General


(AndyD) #4

This is obviously just a test but you have a very wide angle spot for such a small area. The samples are relative to the area of the lamp so by making the angle wider, you effectively lower the sampling.

The ClipSta and ClipEnd are frustrating little buggers and sometimes it seems impossible to get them to work consistently across an entire model (like a city scene with long, tall building and short, stumpy cars) and you can end up with a gap between your object and its shadow. So you have to choose the important areas and light those well - or switch to raytracing :frowning:

It would be nice if someone with good lighting experience would explain the options for lighting big, complex scenes without raytracing. :smiley:


(Umbroth) #5

Thanks for you replies

Traitor : sorry for the wrong forum but in the Blender general I can attach a image at my post.

I try all the thinks : low angle, closer to the objetct, high bias, much shadow resolution, low samples, … it worse

I try in 2.37a, it the same. I think it’s a bug.

Umbroth

Sorry for my bad english

Attachments



(Fligh) #6

I try all the thinks : low angle, closer to the objetct, high bias, much shadow resolution, low samples, … it worse

I try in 2.37a, it the same. I think it’s a bug.

The Bias setting is exactly the same and so are the Clip settings.

%<


(SoloCreator) #7

I’m not a pro but I have been working with zbuffer lamps for quite some time and I rarely use raytraced lamps. Here are some examples of my work using zbuffered lamps.

http://img.photobucket.com/albums/v174/Nitro3d/HaloTest2.jpg

http://img.photobucket.com/albums/v174/Nitro3d/HaloTest.jpg

What helps is to follow these simmple guidelines:

[>] Always ensure that your lamps are placed symmetrically. By this I mean having one set of lamps on one side of your scene and then basically mirroring their positions so you have all four sides of your scene using the exact same layout. This will deliver a lot more consistency when it comes to “clip end”, brightness etc…

[>] Always try and use simple angles for your lamps. 90 digrees, 45 digrees etc…agains this gives a nice consistency in your scene.

Final note:
Just keep things as simple as possible. Although raytraced lamps are easier, with patience the zbuffered lamps can achieve some truly amazing results. Remember you can use the “Only Shadow” feature to make sure you have extra shadow calculation where you need it.

Master zbuffered lamps and you will have full control over the light and shadow of every aspect in your scene.

Hope this helps


(Metsys) #8

SoloCreator, I’m glad you posted that. I’ll be sure to use you as a reference whenever I need to prove my point of “cheating like hell.”

Just to increase my arsenal, how long did it take you to render that? (yes, I’ll probably quote you on that :))


(Ongnissim) #9

I think these render “problems” are caused by the settings on the mesh, not the spotlight. Your mesh has either sub-surf turned on, with the edges creased, or, you have set-smooth turned on when it doesn’t need to be. I’ve gotten similar results with both.

Hold on Tight,
Ongnissim

EDIT: Tried them both…


(SoloCreator) #10

Metsys, you think I cheat?

I don’t quite understand - Please explain!


(Metsys) #11

It’s very much a compliment. I’m not saying you are cheating in the same way that posting a photo of a glass of soda and saying you did it in YafRay. I mean cheating in the way that 3D artists cheat; getting the results you want without resorting to easy-to-set-up but long rendering solutions. Scene compositing, instead of rendering the entire scene all at once is cheating. Using a cluster of spotlights instead of ambient occlusion is cheating. Using environment maps instead of raytraced reflections is cheating. Are those all good things? Absolutely. And kudos to those of you that actually do that.

Instead of using GI, you used a ton of zbuffered spotlights; less physically correct but still artistically right. I don’t know if you used radiosity or not, but if you didn’t, then it’s cheating (actually even radiosity counts as cheating compared to GI, but still similar results). So, I’m complimenting you on your ability to cheat, get the results you want, without resorting to push-button solutions (radiosity is also a push-button solution). You have full control over the lighting of the scene, instead of having some algorithm figure it all out for you.

Teaching people to “cheat like hell” has been one of my personal crusades. I made a point of that in the article I wrote, To Those Learning 3D, that I posted here a while ago. And that’s why I wanted to know how long it took you to render that, because I have a good idea on how long it would have taken you to render that, even a grainy image with low samples. I just wanted to prove my point, and you did a good job.


(Khnum) #12

Perhaps change the word “cheat” then, because that has a negative conotation. Instead, why not simply say “alternative”. :slight_smile: I thought you were hating on him too.


(SoloCreator) #13

It’s kind of a strange way to put it though. I remember in maths class I always use to get in trouble for using the calculator and not my head.

I personally consider letting blender do all the calcualtions for me more as cheating. It took me a whole week to make that light setup in that scene. It would have been a lot easier and faster to use a GI dome etc…

…but then again would you really recommend using the GI dome for an animation. This scene is after all intended for an animatic so I wouldn’t recommend using any of these pre-calcualted methods.

Ironically, I chose Zbuffered lamps because they are much faster than raytraced ones. The reason? Well to get the raytraced shadows looking soft and smooth there needs to be a lot of Anti-aliasing and this takes a lot of process time.

ZBuffered lamps however, produce such soft shadows more naturally which makes them better suited.

Render time was 1 mins 11 secs (Would have been a lot more if I had used AO and/or raytraced shadows to achieve the same effect.

I don’t see it as cheating, it’s just an “alternative” towards much faster rendering times and a better enviroment for animation…cos at the end of the day, if it looks good, then who cares :expressionless:

Sorry I took it the wrong way :wink:


(mzungu) #14

I think “faking it” is a better way to describe it than “cheating”… discussed ad nauseum in this thread. Whatever you call it, if you can master it (as SoloCreator is trying to) you’ll produce great things.


(Metsys) #15

Yeah, “faking it” doesn’t sound as bad as “cheating”, but “cheat like hell” is a lot more fun than saying “fake it like hell.” Maybe I should find a new catchy phrase :).