When does it make sense to switch to Redshift?To achieve the same result, how much faster is a 2080 Ti GPU in Redshift compared to a 1950X CPU in a standard CPU renderer like V-Ray, Arnold or Mantra?
probably is nonsensical not to leverage both processors to their fullest potentialCPU & GPU rendering instead of CPU vs GPU rendering
>>656497With a 2080 Ti it makes sense to switch to Redshift no matter what CPU you have. Same for a 1080 Ti. CPU rendering is pretty much dead for the indie artist. Arnold and Renderman are admitting this and are coming out with GPU implementations.
>>656497I've heard threadripper works miracles for mantra and renderman. Not exactly an exhaustive test, it's in windows, it's an old 16 core chip, and it's Renderman 21(which is slower than 22). https://www.youtube.com/watch?v=CYfA1prozsE
>>656509What about a 2070?
GPUs are pretty much better due to dedicated parallel processing. It'll free the threads on your CPU, too. Faster rendertimes + free CPU vs slower(not too slow relative to the GPU, unless it's way faster) rendering + clogging threads(especially the main thread). If you have/can afford a good GPU, go for it. Using OC'd 980 Ti and an i7 6700, I like the results of GPU rendering.
>>656497 i don't know anything about 2080Ti or redshift but i've been using Blender for a few months so i'm pretty much a expert on all things 3d. I have a ryzen 2600 6core and a Vega 56, my GPU is exactly twice as fast as my CPU in cycles render engine Your 1950x has basically identical per core performance as my 2600 just with 10 more cores so a 1950x is a little faster then a Vega 56 and i'm assuming a 2080Ti is at leas twice as fast as a vega56 So... i guess a 2080Ti is twice as fast as a 1950x ? idk wtf im talking about If a train leave New York traveling 20 mph and you stop is 40 miles away how long will it take for you to arrive? Answer: depends on how many samples you use
>>656868>i don't know anything about 2080Ti or redshift but i've been using Blender for a few months so i'm pretty much a expert on all things 3d.
>>657139I hope so. But these days you can't really tell.
>redshift>cycles literally does the same thing, at totally free cost, without the autoderp...redshift is a certified meme
>>657189But there is no Cycles in Max, Maya and Houdini. Only in C4d and Blender. Redshift is faster, and has custom AOV's and you only pay once and get the plugin for all DCC's. Or you don't pay at all and let it run under the Black flag. No matter how you look at it. Redshift almost always wins.
>>657189Blendtards: The Post.
Is GTX 1070 still a good GPU for rendering?
>>657309Yeah I run redshift for houdini on a 1070, performance is a lot faster than mantra
>>657189You blender users are dumb as hell. You never know what the hell you're talking about, that's why nobody takes you seriously.
>>657346Drawing each pixel with pen tool size 1px in Photoshop is faster then Mantra.
>>656497I haven't done A/B tests against V-Ray in Houdini (and no experience with Arnold), but with 2 x 1070 cards, Redshift is orders of magnitude faster than Mantra on a 1950X. I would >>657426Lol, true true>>657189Obviously Cycles is capable of beautiful renders (any render engine is), but you really can't compare it to Redshift. There are a lot of Blender users begging for a port on the Redshift forums, meaning there are Blender users (of all people) who would rather pay $500 for a license of Redshift than use Cycles for free. One is a proven production renderer already used on many shows/commercials/films, while the other is not. It's all about speed and good support for common production workflows.