Hi, I know this is potentially a redshift issue, but maybe worth asking here as well.

I'm trying to set up a multi-camera sequence in Solaris using redshift. I have a simple two camera sequence being fed into a switch. The switch is being driven by @shot variable set in the context options. It then goes through an edit context options node which sets a frame range for each @shot value.

It’s all working in the viewport, I can scrub through and the cameras will change via the switch node. The problem is when I render to disk with “render all frames with single process checked” It seems to render the entire sequence based on whatever camera was active on the first frame ignoring the switch. When I try it without that option ticked, I get the expected result, but at the expense of having to do scene extraction per frame which ends up being super slow. Is there a way to get around this?

Thanks

I asked the same question a while ago. I settled using the flipbook render, scaling the viewport to the exact resolution that I need. It is of course a bit of a hassle to set it up, but In my experience it is still by far the fastest scene extraction time. I hope the offline rendering will be more streamlined in the future.

Do you know if the flipbook method will respect rendervars/AOVs as well?

Galleryworm
Do you know if the flipbook method will respect rendervars/AOVs as well?

Unfortunately no.. Flipbooks do not include AOVs. In my last thread ( where I basically asked the same question ) somebody suggested that you might be able to render several views per render product. https://www.sidefx.com/forum/topic/76906/ [www.sidefx.com] But I doubt this is possible.

spinori

Galleryworm
Do you know if the flipbook method will respect rendervars/AOVs as well?

Unfortunately no.. Flipbooks do not include AOVs. In my last thread ( where I basically asked the same question ) somebody suggested that you might be able to render several views per render product. https://www.sidefx.com/forum/topic/76906/ [www.sidefx.com] But I doubt this is possible.

Sidefx confirmed that it's not yet possible.

Here's a quick update for anyone who may be having similar issues:

I was able to work around this by using Deadline and TOPS to cook separate usdrender rops for each camera in my sequence instead of having them all flowing into one single rop. Using the deadline scheduler in TOPS (for whatever reason) was the only way I could cook the rops while respecting the "render all frames with single process" avoiding the heavy scene extraction times.

*also kept my context options setup intact, except I linked the frame range of each rop to the time-based options, which allowed me to view it as a sequence live in the viewport.