Can anyone advise on my Hurricane animation?

Started by ianbaxter1, September 22, 2009, 04:59:10 AM

Previous topic - Next topic

ianbaxter1

Hi All

I am fairly new to tg2 and am having a few problems rendering. Below is a 15frame QT of the start of my animation:
http://www.davidlawson5.co.cc/ian/shadow_test.MOV

and here is a link to how I made it:
http://www.davidlawson5.co.cc/ian/tg2_hurricane_layout.jpg

I am using an animated texture projection to create clouds from and all the animation is taken care of in the 2d texture. Can anybody advise me as me to why it flickrs so much even though all my quality settings are fairly high? Cloud samples in the region of 200, detail 0.7 and aa 7. The actual thickness of the cloud seems to change from frame to frame in places which appears to be causing it.

Many thanks on advance.
Ian

p.s I took all density fractals out as the cloud appeared to 'swim' through them.

FrankB

do you still have GI on?

Very likely the flicker is caused by GI not being detailed enough. As I see it you have 4 options:
- turn off GI completely (0/0 and disable the enviro light) .... this might be your solution for this scene.
- turn up GI detail to 3 or 4, but then accept much longer render times per frame
- increase GI blur to insane levels, like 500-800, but then have all shadw details washed out
- use a fill light setup instead. can be downloaded from the file sharing area, sticky. This will work, but the question is whether it gives realistic results in this orbital scene. On our earth, there is no visible light reflections on the night side.

Cheers,
Frank

Tangled-Universe

If I'm not mistaking it is more important to increase the GI sampling than the GI quality.
I'd try to increase the blur radius to Frank's levels and then use a GI of 1/4.

A slightly different remark: cloudsamples are non-informative, since the the amount of samples depends on the density and depth of the clouds, so it's best to mention the quality-setting rather than the samples. This gives a much better idea what to expect.
200 cloudsamples for 4000m thick clouds with a density of 0.5 would be very bad for example.
Perhaps that should be stickied too sometime :)

old_blaggard

I don't think that the really big issue here is GI because I'm not seeing much flickering beyond the noise in the clouds. I would go for a GI detail of 1, samples of 3 or 4, and probably a blur radius of 100. If that's not good enough, then go up to a higher GI blur or samples, although that will, as Frank said, wash out some detail or increase render times. Also, I would try a couple of frames with the cloud acceleration cache disabled. I've noticed that sometimes it causes clouds to appear and disappear and could be causing this flickering.
http://www.terragen.org - A great Terragen resource with models, contests, galleries, and forums.

Matt

#4
Hi Ian,

I think this is a limitation of the way clouds are sampled per micro-polygon of terrain. The only thing that might fix this is a much higher detail level in the render settings, but I don't know how high you would need to go to remove the flicker completely. At higher detail you would be able to reduce other settings such as cloud samples and GI settings, so it may not affect render times as much as you think, but it's not an ideal solution.

I think I need to treat this as a bug, because theoretically it should be possible to get smooth non-flickering clouds if there are enough samples (whereas right now that's not possible). I need to make some changes to the cloud rendering to make this happen.

I don't know if this is an important project on a tight deadline, but if it is, please email me privately (or contact support via our website) so that we can talk about possibly prioritising this bug.

Matt
Just because milk is white doesn't mean that clouds are made of milk.

ianbaxter1

Hi all

Thanks for the quick replies.

I have set the gi detail to 5 and samples also to 5. My blur radius is also set to 400. I think as matt said, maybe this is a bug. I have contacted him as this is on a tight deadline so will keep you all updated with our progress.

Tangled-Universe - thanks for the info on the cloud quality/samples. I had noticed the values seemed to vary a lot but hadn't had time to figure out what was going on.

Thanks again,
Ian