Why IrMap over Brute force for animation rendering?

[^Lele^]

Did not believe so I had to render something.

This made me happy: job done. ^^

[AC5L4T3R]

That's correct.

It worked with all this sequence, in 2.X, with a very very similar approach (minus a couple of new, and life-saving techs we didn't have back then: retrace and leak prevention. So my per-frame subdivs had to be a bit higher still.), where the camera path failed badly due to the combination of sequences' lengths and geometric complexity per frame, but showed errors only at the end of the (1200+ frames long for some) sequence, with quite devastating results on the farm.

https://www.youtube.com/watch?v=ciST26Kx2z8

Were you at Pixomondo in Stuttgart when you were working on Oblivion?

[^Lele^]

Were you at Pixomondo in Stuttgart when you were working on Oblivion?

Indeed, as LnR and Pipeline TD (LnR was local,pipeline global).

[AC5L4T3R]

Indeed, as LnR and Pipeline TD (LnR was local,pipeline global).

Haha awesome, so you know a lot of people I know then. When did you leave? I joined in March '14.

[^Lele^]

Tell Mariusz he can go "s*** a bag ** *****" for me, if you're around (although, i may just do so myself, now that i think of it.). ^^

I left around feb/march 2013, and in a hurry (the time there, besides some amazing colleague, had been documentedly poor, if you recall, for us all).
But with a farm loaded to the brim with sequences set up like this, ready to render for finals: a whole week of farm time, with always, unerringly, usable outputs (i claim this, but there were Sups and comp Leads and TDs on the receiving end, which hopefully may testify to its veracity. ^^).
Perhaps i ought to have milked the cow a bit longer, with hindsight... :P

[4754simon]

After spending a few hours torturing myself over the Chaos animation documentation -- documentation that can only be described as USELESS!!!, someone was kind enough to share this thread with me. Although my understanding of the animation process is very limited at this stage, I am still trying to absorb everything that has been shared on this thread. Thanks for sharing your knowledge.... to help me transition to animation.

A lot of "corona converts" rave about the resources available to them!!!
Hopefully Chaos wake up to this, and throw a bit of attention to their documentation and resources at some point. No point in adding new features if no documentation is available .... unless the 'promotion video' is considered documentation ?!?. Unfortunately when i think i have finally found a great video tutorial... it is .... not in english. AGHH

Anyway. That's my rant for today. Thanks again for sharing your wisdom.

[iancamarillo]

In short use the default settings if you have a flexible deadline. If you need the animation now, and smooth, set your timeline to the length of the camera path, check "use camera path" in irradiance and light cache and render 1 frame. Save the files and change to "from file" on IR and LC. render

[sharktacos]

Lele,

Could you explain your strategy for how to determine the best retrace threshold amount to use for a scene?

[^Lele^]

There is no on-size-fits all for retrace, merely because it's dependent on how good a job the LC did beforehand on the specific scene.
To some, 8 may be too high, and cost a *bit* of rendertime, to others, too low, and some incongruity in the GI will be obvious.
If your GI seems allright in a two-frame test (ideally of the worse two frames, eh. that's generally good science), without obvious areas of splotching, or bubbling, then that's retrace done right.
We give guidelines with a certain amount of "engineering" threshold, preferring a slightly *higher* rendertime, perhaps, but with a more universal quality to the defaults (hence the retrace at 2.0 and 8.0).
You can think of retrace as a safety net for when the scene complexity has defeated the LC, so it's really down to how safe you think the LC calculations will be in a given sequence.
Intuitively, if the LC will have a difficult task ahead, retrace will need to be higher.

[sharktacos]

Thanks. I'm wondering if you could elaborate a bit on how exactly the Retrace Threshold amount works. As a point of comparison, The AA threshold gets to be higher quality the smaller it is. The retrace theshold for LC seems to work the opposite where making it a larger number (8 instead of 2) makes it higher quality. So that has me a bit confused (that is, I get the AA threshold idea, but am confused why a larger LC retrace threshold results in better quality. Could you shed come light on that?

Perhaps the problem is that I am unclear what the values of retrace 2 means exactly. 2 what? For the AA threshold I understand it to mean that for every pixel, samples are taken at the edge of the pixel. If the value difference is above the AA threshold amount then Vray will subdivide that pixel one more time, and so on. Thus a smaller AA theshold number means finer detail/better quality. The "threshold" here refers to the difference in value around a pixel. If it's above that number it gets subdivs, if it's under it, then the render calls it good.

I'd like to similarly understand how the LC retrace threshold works, and just what exactly "2" and "8" are. So I'm hoping you can nerd-out on me about that

[NicoC]

I'd like to similarly understand how the LC retrace threshold works, and just what exactly "2" and "8" are. So I'm hoping you can nerd-out on me about that

I second that

[piotrus3333]

retrace value will tell LC how much of secondary bounces work put back to BF. I remember very old post by Vlado about how to see that - rendering with precalculated LC (of red tinted GI) a scene with blue tinted GI and changing retrace value. color of GI should show how far BF has to retrace LC.

[NicoC]

retrace value will tell LC how much of secondary bounces work put back to BF. I remember very old post by Vlado about how to see that - rendering with precalculated LC (of red tinted GI) a scene with blue tinted GI and changing retrace value. color of GI should show how far BF has to retrace LC.

Thanks piotrus3333I'd be curious to see this post !

[piotrus3333]

Thanks piotrus3333I'd be curious to see this post !

heh, start digging.

[NicoC]

heh, start digging.

Oh, please believe I've tried ... no luck so far.