New render engine RayTrophi

Some of the cpu-based render engine test scene models that I started writing with C++ are free ready-made models. I started my render project by laying the foundations about 2 months ago. The interface is not available, I published the first code samples here https://github.com/maxkemal/raytracing_Proje_Moduler. It took 1 minute with 10 samples and 1920x1080 resolution. BVH rendering time is 4 seconds. It consists of a total of 400,000 triangles. My processor is r7 5700x. I continue to develop the render engine. I am eagerly waiting for your opinions and suggestions. The name of my render engine is Ray Trophi. No processing was applied to the render output.

5 Likes

Congratulations for making it to that point!

If I had something to ask: what type of renderer were you planning on making? have you chosen a direction to go in?

Are you going for full, unbiased optical realism over performance? Are you going for a more flexible production renderer like Cycles? Or maybe something more optimized / less realistic with faster rendering techniques rather than pure raytracing?

1 Like

Thank you for your kind words. Actually, I am at a crossroads. Should I make more realism or should I make optimeze an open source build that can be used for cpu rendering? I don’t know. Could it be an alternative render engine for blender? I would like to design more interesting and unmade materials. Maybe I can focus on that.

1 Like

If you are trying to make a renderer that’s meant to be used in Blender, it will have to compete with Cycles. That’s no easy task, as Cycles is a great renderer. This means you are unlikely to beat it on its strengths.

If you are making a CPU renderer, there is no way you could beat Cycles (or most modern GPU renderers) in pure raytracing speed, so you would benefit from offering something different.

Where Cycles can be beaten:

  • High detail, realistic caustics. Cycles has no real way of doing them reliably. Well, there is a way for refractive caustics, but nothing reliable for reflective caustics.

  • Complex lighting scenarios. Light coming through glass or through a small crack. Interior scenes lit indirectly from a small patch of sunlight on the floor. Path guiding has limited effectiveness and is very slow for now. Cycles also has no option to use a secondary solver or irradiance cache that could help those cases.

  • Cell shading, non-photorealistic rendering. Cycles has limited abilities for it.

1 Like

As you said, I don’t have the capacity to compete for Cycles. I don’t. Because I’m only one person. Even knowing the physics and mathematics is not enough. At the end of the day, it is necessary to convert this into code in an optimal way. Maybe if I can make an application that has a more independent, simple interface, can load objects, provide materials and scene layout and render. That would be great.

1 Like

All I can say is: “bring it on!” Another rendering option is always “a good thing.”

I would strongly encourage you to choose to make your brainchild “open source.” Share the Love.

To me, it’s not about “competition.” It’s all about “choice.” If you’ve just developed another “choice,” bring it on and thanks for sharing!

1 Like