|
Post by chrisg on May 4, 2020 18:42:55 GMT
I've been trying to take my ray tracer to the next level, and started to implement monte carlo path integration as well as some physically based materials.
I still have a lot of improvements to make - especially making it converge faster...and probably a few dozen bugs to fix, but wanted to show off some progress!
Notes on the cornell box: - I had disabled adding the emission from the lights at the top directly... They're still emitting light, you can see the square box of light on the floor - Also being showed off in the cornell box is a directional light. It's sort of like a point light, but all the rays are parallel heading in the same direction, rather than emanating outward from a single point
Attachments:

|
|
fremag
Junior Member

Posts: 73
|
Post by fremag on May 5, 2020 11:11:32 GMT
Very nice !
Do you have some links with details about Monte Carlo path integration to share ?
|
|
|
Post by chrisg on May 6, 2020 5:54:23 GMT
|
|
|
Post by Jamis on May 7, 2020 15:45:55 GMT
This is really cool! I've wanted to spend more time with path integration and some of the more recent research in that area, but things have been crazy for about the last year. I love that you've made progress on this, though---it looks really great! Please keep us all updated on where this takes you.
|
|
|
Post by bezdomniy on Feb 18, 2021 13:05:06 GMT
Those images look really cool. I've just started my journey on adding path tracing to my implementation. Early days yet, I have only implemented lambertian materials so far. Going to a path tracer really forces you to optimize a lot more... I went through and got rid of all dynamic memory allocation except for some buffers at initialization, I'm surprised at how much that reduces render time!
Some first path traced images attached.
|
|