Page 1 of 1
OFX efficiency?

Posted:
Wed May 16, 2018 7:16 pm
by Greg Huson
Generally, do OFX plug-ins work better with more GPU power? Or is that manufacturer dependent?
Re: OFX efficiency?

Posted:
Wed May 16, 2018 8:09 pm
by Dermot Shane
Mfg dependant, some like Paul Dore's run on cuda, others like BCC are locked to a single thread of a single proc.... ouch...
Re: OFX efficiency?

Posted:
Wed May 16, 2018 8:59 pm
by Greg Huson
That explains why my BCC OFX are so damn slow.
Re: OFX efficiency?

Posted:
Wed May 16, 2018 9:50 pm
by Dermot Shane
node cache is our friend....
Re: OFX efficiency?

Posted:
Wed May 16, 2018 9:53 pm
by Greg Huson
Dermot Shane wrote:node cache is our friend....
Smart Cache is actually the greatest invention since the wheel. Who doesn't all media software have this feature?
Re: OFX efficiency?

Posted:
Thu May 17, 2018 12:53 am
by Dermot Shane
Nucoda has vastly better cacheing, as did DS a decade ago... there's room for improvement,
imagine caches written as DPX
and imagine allowing the user to set where the caches are sent to
now imagine true background cacheing that is rocketship fast
then imagine setting caches to output cache
and imagine setting caches to a folder called "color timed master"
about 10 seconds after you grade your last shot the export of the show to DPX is complete, done, finished....
that's Nucoda.... In Resolve you have hours and hours of rendering ahead tieing up your machine
Re: OFX efficiency?

Posted:
Thu May 17, 2018 8:05 am
by Sam Steti
Dermot Shane wrote:Nucoda has vastly better cacheing, as did DS a decade ago... there's room for improvement,
imagine caches written as DPX
I think you can in Resolve
and imagine allowing the user to set where the caches are sent to
Like in Resolve
now imagine true background cacheing that is rocketship fast
like Resolve's smart cache ?
then imagine setting caches to output cache
as if you tick it in Resolve ?
and imagine setting caches to a folder called "color timed master"
like when you created this folder and told Resolve to cache there ?
about 10 seconds after you grade your last shot the export of the show to DPX is complete, done, finished....
like when you manually tell Resolve to do it or user cache...
Some of my points are to be checked but Resolve's not so remote imho
Re: OFX efficiency?

Posted:
Thu May 17, 2018 3:33 pm
by Dermot Shane
unfortuantly Sam, the caches are not readable by external software, ie; a clean DPX seq in a single folder
Resolve's caches are written seemingly randomly into a maze of subfolders, and the sub folders are not user selectable, only the root of the folders tree is selectable, not the sub folders where the cached frames themselves live
the other part of nucoda's great cacheing system is true bg rendering, i set the entire featute off to cache in the CTM folder, start working , and by mid-day the entire show is cached and i never waited for a frame ever
from that point on anything and everything i do is cached always, the CTM is always ready to fly
works great when you have a DS that also imports the same timeline and links to the cache folder, one can do gfx and set deliverabels useing it's parent-child timeline relationship, and be running a rack of deliverables with in seconds of someone saying "done"
far far far cry from where Resolve is as of v15
Re: OFX efficiency?

Posted:
Thu May 17, 2018 3:43 pm
by Igor Riđanović
I agree with Dermot. I use Resolve caching all the time, but there's much room for improvement. The principal problem is that Resolve caches are invalidated for seemingly innocuous user actions like muting a track on and off, sliding a clip and undoing the change, etc.
Re: OFX efficiency?

Posted:
Thu May 17, 2018 4:23 pm
by Sam Steti
True, the bunch of folders and structure itself prevents from reading from another 3d party piece of software. I admit too that it's hugely perfectible.
But you can approach the bg rendering you were referring to in Resolve too right now (of course, readable by Resolve only ruins the stuff, I agree), this is specifically what i wanted to point to...
Re: OFX efficiency?

Posted:
Thu May 17, 2018 4:49 pm
by Sam Steti
Igor Riđanović wrote:I agree with Dermot. I use Resolve caching all the time, but there's much room for improvement. The principal problem is that Resolve caches are invalidated for seemingly innocuous user actions like muting a track on and off, sliding a clip and undoing the change, etc.
I use Resolve's cache too, and also agree with you Igor.
The fact is I deliberately underestimate what you write about because I was on FCP7 years ago, and not only "caches were invalidated for seemingly innocuous user actions like muting a track on and off", but also going back to - say - "mute on" didn't always relink to caches; so if you made this - which therefore was considered as a mistake - you were forced to re-cache the whole stuff

At least Resolve relinks to caches in our example...
Let's say I'm not demanding enough now, just because it's already better than what I was used to.
Re: OFX efficiency?

Posted:
Thu May 17, 2018 6:37 pm
by Dermot Shane
Sam Steti wrote:But you can approach the bg rendering you were referring to in Resolve too right now (of course,.
with Nucoda, i set bg cacheing and keep working, no stopping to wait for caches to render, and the bg render is totaly transparent, you would have to monitor task manager to even know its running
in Resolve 15 you get either a working tool, or a rendering box, but never both at the same time, in Nucoda you get both