Animation render speed

[fixeighted]

I have a few quite long sequences to render and I naturally want to limit the frame time, as I have to do it all in-house to actually make money from the thing.

I recall that not having the vfb active during rendering saves time, but the question is, how much? I haven't time to test this.
Given that my sequences are max 1400 frames, with a frame time of 5 mins, is that the only saving I can make, if it even saves much at all?

Any thoughts are welcome.

[hermit.crab]

The VFB is usually disabled to save memory. I doubt that there'll be a great performance increase from it. You can save time by lowering samples/increasing noise threshold and denoising, or f.e. rendering a lower resolution and using the Nvidia AI Upscaler.

[fixeighted]

Yeah I figured as much, thanks anyway.

I think I've optimised as much as I can. It's a big outdoor scene with a ton of detail and HD frames at 5 mins is already pretty good I guess - with denoiser/.01 noise.

I think I did a quick test on a few frames using the upscaler but honestly was not that impressed, which is a shame; so I guess I'll be sitting here for a month while it all renders as it is

[luckett_jason@hotmail.com]

fixeighted I would transition to Vantage for animations, it has some limitations but for the majority of work it's a rocket ship compared to rendering in-house on a farm.

[fixeighted]

Sadly Vantage is not an option, as the scene is too large and my 2080ti not capable enough to process it all.
Also, given the fact that it contains Tyflows, vdbs and other unsupported stuff, it would be a complete nightmare to
convert and otherwise hack a solution with it.

[luckett_jason@hotmail.com]

Fair play, as far as I know Tyflow is supported but limited to usability as you need the full Tyflow license, yeah vdbs still not resolved, that 2080ti, forget GPU-ing on that.

Maybe the devs can add to this as I see plenty of requests for some of these things.

Good luck

[fixeighted]

Hehe, yes my gpu is now outdated and just for games really.

I bought it when it came out, specifically to work with Vray gpu but literally every time I tried to utilise it I ran into issues, so abandoned
using gpu entirely. The headaches, incompatibilities, workarounds etc are simply too time-wasting for me.

I do wish that Tyflow operated monthly licenses, as I only seem to need its full capacity at random times, though with this project I can deal with the relative slowdown
of the free version.

One thing that has saved me an enormous amount of time is Embergen. Initially I thought to use Phoenix but the choice was pretty obvious.
Embergen is 20 quid for a month's license and is an excellent solution.
In comparison Phoenix is 90 and I know would be so, so much more difficult to achieve, or actually impossible currently, what Embergen can do with just moments of tweaking.

So, all in all, some wins and some losses. I'm happy enough with my solutions for this project.

[Pixel_Domain]

Hehe, yes my gpu is now outdated and just for games really.

I bought it when it came out, specifically to work with Vray gpu but literally every time I tried to utilise it I ran into issues, so abandoned
using gpu entirely. The headaches, incompatibilities, workarounds etc are simply too time-wasting for me.

I do wish that Tyflow operated monthly licenses, as I only seem to need its full capacity at random times, though with this project I can deal with the relative slowdown
of the free version.

One thing that has saved me an enormous amount of time is Embergen. Initially I thought to use Phoenix but the choice was pretty obvious.
Embergen is 20 quid for a month's license and is an excellent solution.
In comparison Phoenix is 90 and I know would be so, so much more difficult to achieve, or actually impossible currently, what Embergen can do with just moments of tweaking.

So, all in all, some wins and some losses. I'm happy enough with my solutions for this project.

Why don't you render everything separately? What's possible on vantage and the rest locally. And then composite it in post. There are some options to make it . Everything that's not supported by vantage you can render locally, with the rest as a shadow catcher. That should work, don't you think so?

[fixeighted]

Unfortunately it's a little more complex than that, as in fact the project consists of a number of separate camera moves, having the same start/end point but differing
in their midway paths. All sorts of stuff happens with variously sliced, hidden/unhidden geometry/Tyflows/vdbs/matched Embergen renders etc.

The GI needs to be consistent between all the shots, as do all integrated moving parts, so it's relatively simple but then again not simple at all, in
all sorts of potentially disatrous ways

This all leads me to wanting the smallest possible margin of error as I output each sequence, with the least amount of compositing needed.

[Pixel_Domain]

Maybe this option can help ?

[fixeighted]

Sure, there are many options, though none that allow me to essentially allow me to have everything in one manageable scene.

The way I have set it up means I only need to turn various layers, or certain modifiers or objects, off/on and then render the relevant sequence.

Setting it up for layered compositing, using whatever methods seem appropriate for each situation, is a series of accidents waiting to happen.

I think when you see the finished result you may agree with my approach. Maybe I'll have missed something obvious and I'll be interested to see
whether I could have saved some meaningful chunks of time.

This is going to be onging for probably months more (already has been 4 months), as the way they feed back is chaotic and unorganised, with me having to wrangle everything whilst
maintaining consistency, which of course adds to its overall complexity