2
$\begingroup$

I'm using CPU rendering, I've heard somewhere that progressive rendering is faster than non progressive, so I checked it in the render panel, I also increased the Tiles size to 256 for both, my question is, is it really faster, because it looks slower than non progressive with 32 for Tiles size.
Also one of the advantages I heard about is that if you feel that the render is clear enough, you can just cancel and save the image "talking about rendering stills". But what if I'm using Composite? I'm mixing a white color with Mist, and mixing that with the rendered image, since this will be calculated after the render is done, will composite still happen if I cancel the render?
I know this looks like many questions, but it's not, the main question is: Why and when to use progressive rendering?

$\endgroup$
  • 3
    $\begingroup$ The compositor can use any image or image sequence from the hard disk or facsimile thereof. If you save the image manually you can specify in the compositor as an input. This is a comment in regard to stopping the render. $\endgroup$ – atomicbezierslinger May 3 '16 at 19:00
  • 2
    $\begingroup$ I don't think that progressive rendering is any faster, the only advantage is that you can stop it at the point where you find the noise level acceptable. In other words if you set your samples at 300 and decide to stop at 200, yes it might be a shorter render, but rendering 300 samples either way should be the same time. I don't know if progressive sampling might stress your CPU more than spliting the load and using different threads for different smaller tiles. $\endgroup$ – cegaton May 3 '16 at 19:00
  • 3
    $\begingroup$ As a matter of fact, progressive rendering is often slower than tiled rendering, assuming you have an optimal tile size for your scene/hardware. The only advantage is as you say; you can let it render until you deem it noiseless enough. $\endgroup$ – gandalf3 May 3 '16 at 19:03
  • $\begingroup$ Thank you, then I guess they meant by faster, that you can stop the render process and save the image. Is this the only advantage? $\endgroup$ – Georges May 3 '16 at 19:04
7
$\begingroup$

Progressive render is always slower than a properly configured non-progressive render.

You use progressive rendering if you don't know how many samples to use. For example, you might use progressive on your initial render to get a feel for how many samples will look good, or to extrapolate how long a given number of samples might take. You might also use progressive if you have a specific amount of time to render, but don't know how many samples you can get done in that amount of time.

You might also use progressive if you don't know how much time you'll have to render the image. For example, if you have a SUPER complex scene and you know you're going to be out of town for a few days, you might start the progressive render, set it for umpteen-billion samples, and just let it go until you get back... whenever that is. That way, you're rendering the WHOLE time (it doesn't stop too soon with too few samples), but you also have a FULL image whenever you decide to stop (without having to wait for the last tile... or two... or more, to finish).

http://www.blenderguru.com/articles/4-easy-ways-to-speed-up-cycles/

look to 3. point

With CPU you should use a small tile size it will be faster.

With GPU you should use a big tile size.

enter image description here Progressive rendering will make a tile as big as your image -> render time will be long

Composite could be done if you simply take a image (use openexr format if you use passes) as input

$\endgroup$
  • 4
    $\begingroup$ I find that chart to be misleading, Those numbers might be truth for the particular scene and the particular computer used. But in my opinion should only be thought as a starting point for others to make their own tests. Different CPUs, GPUs,amount of vRAM and RAM plus the complexity of a scene will make a difference. Test and see what works for you. $\endgroup$ – cegaton May 3 '16 at 22:02
  • $\begingroup$ I take it as a given that the chart is a single sample. It also shows the general trend of various tile sizes for CPU vs GPU, which is helpful. Plus the given link leads to the following caveat: "I conducted some studies using this scene, and came to these results..." $\endgroup$ – jdk1.0 Apr 4 '18 at 22:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.