Is anyone familiar with the POV ray renderer? I'm having SERIOUS shadow problems!

I wanted to render my scenes using POV ray, but there are triangles near the shadows. Do you know what the problem could be?


Also, on a side note, how would I make the color space of the POV renders gamma instead of linear??
It looks like linear is Blender’s default:
https://docs.unity3d.com/uploads/Main/LinearRendering-LightingSphereLinearGamma.png

Self shadowing. Also known as the shadow terminator problem. Use more lights or more geometry.

Not sure what you mean there. Gamma and linear are not colour spaces.

“Self shadowing. Also known as the shadow terminator problem. Use more lights or more geometry”

There aren’t any of those triangles/self shadowing problem with Blenders default renderer. And I downloaded an old .obj converter for POV ray I don’t seem to get those problems. I would like to use Blender to import models/uv coordinates/animations in to POV ray. What should I do??

Also, by more geometry, do you mean a more complex mesh or more meshes in the scene?

"Not sure what you mean there. Gamma and linear are not colour spaces.
Sorry, I’m new to blender and I’m more familiar with Unity Engine. I would like for my renders to look close to the gamma space picture I posted earlier, in case you’re wondering. How would I go about this?

Povray is a ray tracer, Blender internal is a scanline. The terminator problem is related to the way ray tracers work. An object with more/smaller polygons is less affected. Shadows with some lighting are less affected. Maybe there are shadow bias settings for Povray?
I don’t know what an old .obj converter is, other than that it must convert something to, or from .obj.

‘Linear space’ and ‘gamma space’ are fairly meaningless terms. If you want to save renders as non linear you can use anything 8bit, such as .jpg or .png. I can’t tell you what to change to make this render look like that render just by looking at a picture. Maybe use inverse square fall-off on the lights?

1 Like

“I don’t know what an old .obj converter is, other than that it must convert something to, or from .obj.”

It can convert .obj files to POV files. But of course, they have no animation.

The reason I wanted to use POV Ray for my renders is because I’m currently in the process of making a game that’s based off of early 1990s point and click adventure games (like Myst), and POV ray has been around that long and the renders look very similar.

Myst (1993):

“Povray is a ray tracer, Blender internal is a scanline.”

Ah, Ok. Sometimes I have the same triangle artifacts with Mental Ray as well, but I never found a solution for it.

“Maybe there are shadow bias settings for Povray?”
Thanks, I’ll look into it.

1 Like

Any render engine can do that. Use point lights, procedural textures and disable global illumination.

“Any render engine can do that. Use point lights, procedural textures and disable global illumination.”

Render engines do look different from one another. I can be pretty easy (for me at least) to tell when something was rendered in the 1990s or looks very dated looking. I’m definitely striving for authenticity.

Old POV render:
PAWN3

I even spent some time learning Autodesk 3D Studio (the old DOS one) to kind of understand the limitations of early 3D programs- they only had planar,spherical, and cylindrical mapping. There was no UV unwrapping at all (up until the mid-late 1990s) Skeletal deformation was also uncommon, and characters mostly consisted of multiple meshes parented to one another. But 3D studio only had raytracing for shadows, so POV ray is a step up.

There is no good gamma space in POV-Ray.

You have two options:

  1. Linear space rendering with light source fade_power 2. Render 16-bit per channel images and adjust it later.

  2. Faked high dynamic range. Set light sources fade_power to 1, and calculate light source intensities using formula: (intensity*exposure)^(1/2.47393118833241)

  3. Dynamic range with gamma 2.2, you need to use (intensity*exposure)^(1/2.2) and set fade_power 1.00907

And that “exposure” is tuning variable. To adjust exposure in image, you need to re-render using different variable.

The main idea in POV-Ray is to do much as possible modelling without meshes, using isosurfaces and CSG you can define it POV-Ray SDL language. That is the cool thing in that renderer when you can make perfect shapes without any polygons and raytrace images without noise.

So in POV-Ray way, you don’t have mesh to sphere, you use sphere. Meshes are used when you can’t use pure shapes.

2 Likes

1990s games used a lot 3D studio and limitations were something like "max 8 light sourcea, only one light source make shadows (one shadow map textures), big textures 256x256 and tileable textures 64x64. Max vertices in mesh 65536, blinn-phong shading and rendering was linear.

POV-Ray was used some games and similar rendering was used little. Procedural textures and CSG modelling was the thing here. POV-Ray allows real reflections and glass, and anti-aliasing looks different in raytracing compared to scanline tracing. Lower resolution images (back then games were 640x480 or lower) that was visible.

"990s games used a lot 3D studio and limitations were something like "max 8 light sourcea, only one light source make shadows (one shadow map textures), big textures 256x256 and tileable textures 64x64. Max vertices in mesh 65536, blinn-phong shading and rendering was linear.

POV-Ray was used some games and similar rendering was used little. Procedural textures and CSG modelling was the thing here. POV-Ray allows real reflections and glass,"

Thanks. I was going to use Stratavision (since I head it was used to make Myst), but I couldn’t get it to work/render at all, and the files aren’t compatible with modern programs , unlike .3ds files or just using Blender + POV ray.

"The main idea in POV-Ray is to do much as possible modelling without meshes, using isosurfaces and CSG you can define it POV-Ray SDL language. That is the cool thing in that renderer when you can make perfect shapes without any polygons and raytrace images without noise.

So in POV-Ray way, you don’t have mesh to sphere, you use sphere. Meshes are used when you can’t use pure shapes."

So, are you basically saying there is no way to get rid of the shadow artifacts in POV ray??

I didn’t remember having any shadow artifacts when I used POV-Ray. I did modelling as much as possible with POV-Ray SDL language and CSG, and when I imported meshes I used smooth triangles. But if you must use meshes, just make them high resolution and use smooth triangles. Raising ambient and using bumpmapping of course can make them disappear.

Maybe you can try to mimic Stratavision with Cycles. Just set light paths “direct light”, use only diffuse BSDF or Glossy BSDF with setting “sharp”, render using branched path tracing with 81 AA and procedural textures, light threshold to 0. That should give that nice noiseless look. Set colour management to “raw” and when you need to add light to shadow area, use ambient occlusion with distance 0. Instead of exposure, adjust light sources.

To get image more authentic, convert them to 8-bit palette.

1 Like

Thank you.

So as not to mislead new users about POV,
It is much more powerfull AND MODERN than what common opininon propagates without ever trying its latest versions. It has a progressive stochastic rendering mode just as Yafaray like renderers. Just read its documentation before using. Also To still make it look more raw, you could simply use its +Q command line launcher (there is a field to input that in Blender’s exporter) at lower values. Just read POV documentation