OTOY Forums

HI all,
Firstly apologies if this has been answered or posted before, but I can't seem to find it anywhere. Why does the live viewer render much faster than the Picture viewer, I setup a scene and love the results eg. 30 seconds with 4 GPU's (2x Titans & 2x 780 TI)I get in the live viewer but then when I start to render my animation it takes approx. 3.mins approx 6 times longer to render. All the settings are the same in my settings window and my render settings.

Again any explanation or advice is greatly appreciated

hi, what a question? hard to say something that without know what are you doing? content, settings, export times, everything is a factor. do you have idea about exported scene structure can change with render settings? also so big differences on times, not logical.

Hi Aoktar,

Very new to your plugin and octane in general, great work btw.

Ok I realized that I am rendering at 1920 x 1080 32 bit linear workflow with a depth multipass, should of added that in the first post . This obviously affects the render times I presume?

attached is the file tested

Again thanks for your assistance, greatly appreciated.

Some explanation about render settings:
Faster updates, this makes same Octane scene structure as LV(interactive mode). Untagged objects(with octane objecttag) will be collected to a single mesh. Other all tagged objects will be exported as each a seperated meshes.
Faster rendering, this collects each meshno to a single mesh. I recommend mostly to use first method.

But can't see a problem about this settings. Maybe a wrong maxsamples value can be send to renderer while anim. rendering. Can you check it? Also try without multipass and floatbuffer to see different.

Hi Aoktar,

Thanks again, I have played around with a few settings I managed to reduce my render times per frame in PV by eliminating the alpha channel, and alpha shadows in pathtracing to a approx. a 1.40, which on LV render at 25 sec. I eliminated the HDR image which made no major difference except for a few seconds. I have tested it by saving the images as jpgs, no multipass or linear values and still got the 3.08 mins per frame. No matter which Kernel I select DL, PT or pmc and only setting samples set to 1000 render times are way longer. I did however max my priority settings to PRIORITY_HIGH which seemed to help my render time to 1.20/frame.

I'm still not getting close to the 25-30sec LV displays fastest render the best I achieved in Picture Viewer was 1.19 compared to the Whopping 25 seconds in LV.

Again thank you for your help I'm just trying desperately to get the best render times out of my hardware whichworks awesomely in LV, just not PV.

Hi again,

Any ideas anyone? This is so frustrating I have tried almost everything I can think of, to render the same time as Live Viewer as in Picture Viewer.

25 seconds it takes in LV and 1:40 in PV WTF!! file seems to be setup as should be.

Hi,
If you didn't resampled your screen captures, your preview render (733 x 412 px) is really smaller than your final render (1920 x 1080 px). So, i think it's not so absurd it takes more time to render in the picture viewer... Isn't it ? You should test the render speed in the Picture Viewer, not in the LV. The LV gives you the Ms/Sec but not the real render time if the pixel amount is different in final render.
To speed up your render, you could use Directlighting > GI_DIFFUSE with specular (0), glossydepth (3), diffuse (3) and alpha shadows unchecked.
Or if you want to keep pathtracing, 16 in Diffuse and glossy depth is really too much for this kind of scene. Respectively 3 and 3 should be high enough and alpha shadows should be unchecked (there is no alpha map in your scene and it takes ressources).

Actually your z-depth_max isn't adjusted in the Deep Channel. It will not give usefull results like this.

I hope this will help you.

hmm, right. i have tested on same res. Speed and render times are similar. I think that this friend couldn't think about different resolutions.

atome451 wrote:To speed up your render, you could use Directlighting > GI_DIFFUSE with specular (0), glossydepth (3), diffuse (3) and alpha shadows unchecked.
Or if you want to keep pathtracing, 16 in Diffuse and glossy depth is really too much for this kind of scene. Respectively 3 and 3 should be high enough and alpha shadows should be unchecked (there is no alpha map in your scene and it takes ressources).

My mistake... in this case glossy depth is better with a value of 4 or 5, not 3. Sorry.

LOL, i could notified until Atome says that, LV has different resolution. Uhmg, can't concentrate to topics sometimes.