OTOY Forums

I'm aware the estimated cost is not to be expected as it's not accurate. I did a small animation test render and it was off by 578%. It worries me if the price will be this wild on a long animation or even if I could expect a similar ratio.
The test was 49 frames 1280x720 that takes 15 to 17 seconds to render each on a machine with OB Score of 225. All frames have pretty much the same 3 objects in view. Estimation told me it would take 26.67 OB/Hrs, it took 154,16 OB/Hrs.

Questions is: are estimations really that wild, or constantly underestimated, or maybe the estimation was so off because of small rendering time but file handling (several passes) or scene processing taking a greater portion of the time?

Edit: something must went wrong between OctaneVR and the cloud render. At this rate I could purchase a new 1080 in about 2 days worth of render. Should I be aware of any different setting that may be causing slow down on the cloud render that doesn't happen on OctaneVR/Standalone/Plugin?

Hi there,

In addition to the information that you provided, the estimation feature takes into account compilation time. This is likely where some of the confusion is coming from. I would also recommend double checking your computers OB score to make sure that all of your information is as accurate as possible. If there is any inaccuracies in the information given to the estimator, it can generate extremely inaccurate estimates.

Please feel free to let us know if you have any other questions or concerns.

ivankio wrote:I'm aware the estimated cost is not to be expected as it's not accurate. I did a small animation test render and it was off by 578%. It worries me if the price will be this wild on a long animation or even if I could expect a similar ratio.
The test was 49 frames 1280x720 that takes 15 to 17 seconds to render each on a machine with OB Score of 225. All frames have pretty much the same 3 objects in view. Estimation told me it would take 26.67 OB/Hrs, it took 154,16 OB/Hrs.

Questions is: are estimations really that wild, or constantly underestimated, or maybe the estimation was so off because of small rendering time but file handling (several passes) or scene processing taking a greater portion of the time?

Edit: something must went wrong between OctaneVR and the cloud render. At this rate I could purchase a new 1080 in about 2 days worth of render. Should I be aware of any different setting that may be causing slow down on the cloud render that doesn't happen on OctaneVR/Standalone/Plugin?

Scene processing and film buffer export time runs at the same rate as OB/Hr for pure rendering: basically credit usage is mapped from the point where a GPU render node (typically 500-1000 OB) is activated and begins loading the data from a frame load, and ends when the film buffer (multi-pass, OCR, etc.) is saved to disk and the node is free to render the next work item. If the majority of the work is rendering (as it should be on ORC), these costs are negligible, but if your render takes only 10 seconds to complete at 1000 OB, but the scene takes 5s to load and 5s to export the render, only 50% of that OB/Hr is being used for GPU rendering work.

This is just one of the many reasons I won't use ORC yet.. was going to opt for it on a project I'm on at the moment but i've decided to go 3rd party render farm and spend my money with them.

I don't trust the speed of customer service OTOY provide or the reliability of ORC. There's also a lot of confusion about the service and how it functions, It feels like a risk to use commercially unless you're fairly experienced with it.

Pretty much any correspondence I've had with OTOY has taken minimum the best part of a week to be responded to, not good if you're on a deadline. Need more credits? Have to email and request, no thanks.

I think generally there is a distrust of OTOY's commitment to stuff like this, and I think it needs to be improved - seems like as a customer you're left to your own devices, apart from the plugin developers who seem to be doing the job of an entire team on their own and who are clearly overworked.

We need video tutorials, commitments to CS feedback timeframes... it's archaic that you if you want more than 50$ worth of credit per month you have to request it.

There's no centralised place on this forum with all the meaningful information - it's scattered in posts everywhere. It seems like moderators waste hours of their time answering the same questions over and over again because they haven't taken the time to organise important information into one place.

Has ORC matured to get more reliable estimates and competitive scene processing?
From what I gathered in this forum I got the impression the situation didn't improve. Capricious estimations and a much overpriced farm service. Is the impression wrong? Is RNDR another thing somewhere or another name for ORC?

ivankio wrote:Is RNDR another thing somewhere or another name for ORC?

RNDR has been built from ground up. We have advised for costs being better predictable. Therefore the billed time has been reduced to actual render time on the RNDR network. This means more consistent outcomes not only for large renders of heavy scenes but also for smaller jobs with short render times as most overhead is not getting billed anymore.

With the launch of the RNDR network there will be multiple priority tiers, reducing the overall costs by 2 - 4x compared to ORC as described in this article:
https://medium.com/render-token/rndr-to ... 8d5dea1d29