PDA

View Full Version : Compare Bunkspeed renderers, again! - Nissan GTR



Aselert
07-26-2013, 09:55 PM
Hi,

Here we can see 4 kind of renderings quickly done with Pro RC 2014 (0% photoshop):

514

The first (0,1s) was created with the Preview mode.
The second (11s) was created with the Fast mode but with the bad quality mode on settings. Finally very close to the preview!
The third (37s) with the Fast mode, again, but with High quality setting
And the last (118s) with the classic Accurate ;)

:cool: I'm fan of all this renderer mode! Just a request: please stabilize the Fast!

PS: All was done with a simple old GTX480, plus a I7 860 (hybrid mode for the Accurate)

CadDood
07-27-2013, 05:12 AM
Thanks for taking the time to test and post the results! We are making progress !

Aselert
07-29-2013, 10:42 AM
Thanks CadDood!

And now, I asked a friend to reproduce the same image than me, but with his Keyhsot (he has the latest version, 4.1) to make a comparison with Bunkspeed in Fast mode.
The goal was to compare 2 biased renderer, to be fair. :D

515

516

In terms of configuration, the time corresponding to its Quadcore I7 920 (which is actually quite close to my I7 860) for Keyshot.
And a single GTX 480 for me, with Bunkspeed, in fast mode (no hybrid).

The result, in terms of quality, is the closest possible (even if the metallic paint isn't exactly the same).

The final time is clear: 3x faster for Bunkspeed! :cool:

andy
07-29-2013, 01:17 PM
I'm not really sure this is a great example. If you are just looking at a car like that you'll get faster results still just using AuxPecker (free inside of rhino) or if you want to pay for something, head toward something like showcase. Both of those are actually real-time and will produce results that look indistiguishable to those two renders. Game engines have shown that there isn't really much of a need to raytrace cars if you're clever with some reflection maps, unless you have a modeled scene.

Also, it's too difficult to compare those render times because you're doing it on two different machines. If you want to really compare, then you'd have to do it on several different budgets, and build those machines. If you have an extremely low budget, then keyshot will win just because you can completely eliminate the need for a video card. But that's unrealistically low. In the low mid range bunkspeed will probably win in speed, but if you have very complex scenes at all, then keyshot will blow it away again because you have extra memory. Spending a bit more and you have a similar problem with memory, but then the needle sways pretty far in bunkspeed's direction. Spend an insane amount of money, and bunkspeed probably has an edge there as well. Although, if you add in costs for render farms then keyshot is considerably cheaper because you can make dirt cheap machines for processing.

Aselert
07-29-2013, 02:43 PM
Yes but Andy ... The goal was just to compare two renderers close in terms of conception (I mean conception in their use), simplified, an attractive interface for designers and not the CG artists and allow for have a quality made ​​in a very short time. That's it!

In addition, I do not agree with you when you say that two PC configurations are not relatives. Although it isn't a last generation config for Keyshot (I7 Ivy Bridge, for example) but it is exactly the same for Bunkspeed... with my old GTX 480 which is just a radiator pc!;)

I haven't make a comparison between a dual core and 4 GTX 690 ..!
Because ultimately, our two config (my friend's and mine) are quite close in terms of price ($ 800 roughly when we bought it).

And about the memory it is a real thing in the moment but will be less in the future, with the instancing and the out of core technologies.

And I dare you to get the same result with AuxPecker ahaha! :cool:

andy
07-30-2013, 01:21 PM
I've fooled people with AuxPecker before, and I could do it again if I had do. If you give AuxPecker the same amount of time you do tweaking a render you'd probably get results as well. There's also Neon for Rhino, which is also free, but I'm not as happy with it because it's really not that fast. The advantage of AuxPecker is it's fast, and it's actually real-time. I threw that one out there simply because it's free, but Showcase is pretty darn easy, isn't free, but would give you very comparable results to what you're showing here.

My point in the systems was not saying your comparison wasn't valid. It's just not valid if someone is purchasing a new system because it's not applicable for all tiers. And things like instancing actually implemented could be 5 months off, or 5 years off. There's no way to know. Even if it's used, how long before it actually works without crashing?

BrianHillner
08-05-2013, 11:38 PM
Thank you all for starting this discussion. It's very good to see side by side examples of Bunkspeed and competitors, especially with the included render times.