[dilbert]

Hi all. I've been gone overseas for about a year, so I'm just coming back to work with Maxwell. When I got back, I was excited to see version 2 had been released, especially because of all the advertised improvements in speed and quality. I was stunned by the side-by-side comparisons that Next Limit put up to sell Version 2, and was equally excited at being able to increase my productivity with the advertised decreased render times to hit a specific Sample Level. So, I went ahead and made the purchase....

Deep breath...........Argghhhhhhh! I've been bamboozled!! I've opened up countless scenes from my portfolio, expecting to see a significant speed increase, but it's the exact OPPOSITE!! The scenes are taking longer to render, with some taking significantly longer to render!! I can't open a single scene and get a speed increase, to include a simple ball with a 50% grey plain material stuck on it. What is going on?? When Next Limit puts up a nice glossy advertisement selling their product as having certain features compared to V1.7, well I expect those features to exist. When you say it will render 2 to 5 times faster, I expect to open ANY scene and see a speed increase. I understand that the new version might be faster for certain materials (dialectrics etc.), but if you can't open a simple ball scene with a 50% grey material stuck on it and see a speed increase, then the advertising is blatant lying. Am I the only one who is pretty damn angry about this? I'll get off my soap box now...

[Bubbaloo]

Are you comparing time and noise? You should be. Benchmark and sampling level differences between 1.7 and 2.0 are irrelevant.

Here's a quick test from me:
1st one is V1.7.1 @ 1 minute, 2nd one is V2.0 @ 1 minute:



V1.7.1 is noisier at 1 minute.

1st one is V1.7.1 @ 5 minutes, 2nd one is V2.0 @ 5 minutes:



V2.0 is noise free, while V1.7.1 still has noise that will take quite a bit longer yet to clear to the level of the V2.0 render. I will keep rendering the V1.7.1 scene to see exactly how long before it reaches the same noiseless clarity and report the time. I'll also set up some more tests.

[Bubbaloo]

Here's the 1.7.1 version @ 15 minutes. Is it as clear as the 5 minute V2.0 render?

[dilbert]

Thanks for the response Bubbaloo. So if I'm understanding you correctly, Sampling Level has no correlation between the two versions? i.e Version 2's sampling level 12 is less noisy than Version 1.7's Sampling Level 12? So with Version 2, you only have to render to a lesser Sampling Level to get the same result, hence the "increased speed". If that is the case then I take back what I said. However, I always remember rendering scenes in V1.7, and finding that I had to render to a particular sampling level to get all the subtle highlights/caustics etc to appear. If that's the same in V2, then that would negate the explanation.

If your suggestion is correct, this is hard to subjectively test.

[Bubbaloo]

Yes, comparing V1 to V2 is subjective to a point. But some materials clear very very fast compared to v1 (multi-bsdf materials). It really is scene specific, but overall I have seen an increase in noise clearing speed in every scene I have rendered so far with V2. Also, noise that was very hard to clear in V1, in dark corners, for example, now clear much faster. You add these things to all of the multitude of other improvements, and honestly, I feel that V2 is far superior to any other render engine available today. That's why I am so passionate about the software!

[shen.de]

I feel that V2 is far superior to any other render engine available today.

that depends on the user... or do you talking about unbiased renderengines only?

I've seen alot of crappy maxwell images and a lot of cool vyray and modo renderings lately ... I'm a maxwell fan myself but I think maxwell alone can't do the job for you

[Bubbaloo]

Everyone's entitled to their opinion. That's mine.