Progressive Rendering


(mpan3) #1

Reading some of the posts here on progressive rendering (not sure what the proper term is…step refinement?) I believe this feature will benefit the artist a lot.

If you are not sure what I meant, the standalone Maxwell renderer is an example. Basically as oppose to starting at a fixed coordinates and work its way up/down, Maxwell renders a low res image first and progressivly adds the detail.

I am pretty sure Blender can implement this easily, so for example, it will render every 100th pixels (0,100; 0, 200; 0, 300; 0, 400, 100, 100; 100; 200…etc) and interpolate the rest, and then it goes back and renders say every 50 pixels: (0,50; 0, 150, etc…) Slowly, a blurry image will magically appear :wink:

Advantages:
Artist able to see composition/lighting/color before the rendering is finished.
You can go back weeks later to a previous render and do a semi-rerender in which pixel data will be added in for a higher res. image.
Looks cooler 8)

I realized that someone had a script for this, but I was hoping that it will be implemented in the main BF tree for the next release 2.40/1?


(SoftWork) #2

Here’s the link to the script: http://www.elysiun.com/forum/viewtopic.php?t=48946&highlight=progressive&sid=016bf975ddd7bd12cb5b1c60ec632180

In the render panel the 100%-75%-50%-25% are used for this.


(mpan3) #3

Unless i am wrong, but it seems rendering time actually increased. It looks like that it’s Blender is rendering a 25% image, and then a 50% (with total disregard to the previous rendered data), then 75%, then the full image. therefore with this script,

the rendering time = rendering time of the 25% image + … rendering time of the 100% image


(SoftWork) #4

Rendertime indeed increases (not only for this script but for all progressive renderings), but normally you don’t use progressive render for a final rendering but only in the testing stage. Like you already said it is useful to quickly see if a scene would look right after you changed or added something.


(blazer003) #5

I haven’t used it, so maybe I don’t know what I’m talking about, but if this script just renders the image 4 times at diferent sizes, why not just do it yourself at 25% to see what it looks like, and 100% for the final?


(mpan3) #6

Unfortunatly i think that’s what the script does, and that’s exactly why I am looking forward to a better approach to this.