Return-Path: Received: from host6.canaca.com ([66.49.160.142] verified) by media-motion.tv (CommuniGate Pro SMTP 4.2.10) with ESMTP-TLS id 4873714 for AE-List@media-motion.tv; Tue, 16 Oct 2012 07:04:04 +0200 Received: from static184-169-61-119.r.rev.accesscomm.ca ([184.169.61.119] helo=[10.0.1.12]) by host6.canaca.com with esmtpsa (TLSv1:AES128-SHA:128) (Exim 4.69) (envelope-from ) id 1TNzMD-00072M-N6 for AE-List@media-motion.tv; Tue, 16 Oct 2012 01:06:22 -0400 Subject: Re: [AE] RED in CS6 References: From: Jack Tunnicliffe Content-Type: multipart/alternative; boundary=Apple-Mail-965B37B2-5FF7-4B25-B56B-CC907C4FF644 X-Mailer: iPhone Mail (10A405) In-Reply-To: Message-Id: <88FDB9DE-EDEF-4EAE-8A11-0E42DDFFB239@javapost.ca> Date: Mon, 15 Oct 2012 23:06:15 -0600 To: After Effects Mail List Content-Transfer-Encoding: 7bit Mime-Version: 1.0 (1.0) X-AntiAbuse: This header was added to track abuse, please include it with any abuse report X-AntiAbuse: Primary Hostname - host6.canaca.com X-AntiAbuse: Original Domain - media-motion.tv X-AntiAbuse: Originator/Caller UID/GID - [47 12] / [47 12] X-AntiAbuse: Sender Address Domain - javapost.ca X-Source: X-Source-Args: X-Source-Dir: --Apple-Mail-965B37B2-5FF7-4B25-B56B-CC907C4FF644 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable Well I do. I'm not trying to start a debate but just finished a movie GPU de= code all Epic 5k. Starting another one tomorrow. Have done several TV series= , many, many commercials, etc, this way.=20 I love the way Davinci handles Red, real time playback at 1/4 resolution, wh= ich still looks great, even projected. Renders at full rez premium. The only difference I can see is a slightly softer image with GPU compared t= o Red Rocket output. I actually painted something back from a Davinci decode= on top of a Red Rocket decode today, where the VizFX had left a line from a= poor garbage matte across someone's face. The match was perfect, perfect en= ough to paint someone's skin back from another file. Sent from my iPhone On 2012-10-15, at 9:43 PM, Rendernyc wrote: > While gpu speeds up resolve I'm pretty sure no one does r3d decoding in gp= u >=20 > Red has been pretty strict about this requiring either a red rocket or sof= tware decoding. >=20 > I've heard many engineers and third parties grumbling about this many time= s over the last few years. >=20 >=20 >=20 > On Oct 15, 2012, at 11:04 PM, Jack Tunnicliffe wrote: >=20 >> Well I was avoid saying this but because you bring it up, Davinci does al= lowsyou to render RED with GPU.=20 >>=20 >> We have a rocket card in a system for doing digital dailies but we can al= so do them using Davinci Resolve an a $500 Cuda card. We've had our Rocket c= ards burn out a couple of times. They work hard. >>=20 >> A person could install the Lite version of Resolve, buy a Cuda card which= would give you 3D acceleration in CS6, speed up GenArts plug ins like Sapph= ire and monster. You'd be restricted to HD resolution out with Lite but for $= 999 you'd have 5K out to almost any other format. >>=20 >> I haven't rendered out 5K from Davinci. That might exhaust the GPU, but I= 've rendered out many other large formats. >>=20 >> Another 2 cents of thought. >>=20 >> Sent from my iPhone >>=20 >> On 2012-10-15, at 8:07 PM, Rendernyc wrote: >>=20 >>> First thing I do if I get an r3d in AE is to transcode it or create a 16= bit "proxy" so I don't have to wait for debayering >>>=20 >>> Maybe on day red will let us debayer on the gpu but still doesn't seem t= o be happening anytime soon.=20 >>>=20 >>> On Oct 15, 2012, at 9:42 PM, Phil Spitler wrote: >>>=20 >>>> Thanks Tim, the idea would be to get a Red Rocket for my workstation th= en dump the renders onto our farm but with these slow renders that is not fe= asible. >>>>=20 >>>> I really want to avoid having to transcode all the footage into anther f= ormat for a whole host of reasons... >>>>=20 >>>> Let's see how my CS6 test go tomorrow. >>>>=20 >>>> Thanks. >>>>=20 >>>> Phil >>>>=20 >>>>=20 >>>>=20 >>>> Phil Spitler | Creative Technologist | Bonfire Labs | t : 415.39= 4.8200 m : 415.571.3139 >>>>=20 >>>> Website | Facebook | LinkedIn >>>>=20 >>>>=20 >>>>=20 >>>>=20 >>>>=20 >>>>=20 >>>>=20 >>>>=20 >>>>=20 >>>>=20 >>>> On Oct 15, 2012, at 6:25 PM, Tim Clapham wrote: >>>>=20 >>>>> If you are working with a lot of R3D files then perhaps it is worth in= vesting in a Red Rocket card. The debayer process is realtime and can save y= ou hours of time compared with using software. Although if you are not deali= ng with many files it could be a tad expensive at just under $5K >>>>>=20 >>>>> ___=20 >>>>>=20 >>>>> Tim Clapham >>>>> Director >>>>> ___ >>>>>=20 >>>>> LUXX >>>>> 3D Animation >>>>> Motion Graphics >>>>> ___ >>>>>=20 >>>>> web : luxx.com.au >>>>> blog : helloluxx.com >>>>>=20 >>>>>=20 >>>>>=20 >>>>>=20 >>>>> On 16/10/2012, at 12:10 PM, Phil Spitler wrote: >>>>>=20 >>>>>> Hmmmm I'm gonna try CS6 tomorrow.=20 >>>>>>=20 >>>>>> Can't believe how slow this is in CS5.5 >>>>>>=20 >>>>>> Hoping it is ok in CS6.=20 >>>>>>=20 >>>>>> Cheers.=20 >>>>>>=20 >>>>>> Phil >>>>>>=20 >>>>>>=20 >>>>>>=20 >>>>>> Phil Spitler | Creative Technologist | Bonfire Labs | t : 415.3= 94.8200 | c : 415.571.3139 | Bonfirelabs.com >>>>>>=20 >>>>>> "Get Social" - http://www.facebook.com/bonfirelabs >>>>>>=20 >>>>>> On Oct 15, 2012, at 5:59 PM, Jack Tunnicliffe wrot= e: >>>>>>=20 >>>>>>> I was rendering out full 5K. I don't think that scaling would make m= uch difference. I'm using the latest build of CS6. >>>>>>>=20 >>>>>>> Sent from my iPhone >>>>>>>=20 >>>>>>> On 2012-10-15, at 6:31 PM, Phil Spitler wrote= : >>>>>>>=20 >>>>>>>> Weird, I am also on a really fast SAN and my files are 5k HD which a= re smaller than yours. >>>>>>>>=20 >>>>>>>> I have mine in a 1920x1080 comp so I am scaling too which could be p= art of the problem. >>>>>>>>=20 >>>>>>>> Were you rendering 5k or 1080? >>>>>>>>=20 >>>>>>>> Are you using CS5.5 or CS6 ? >>>>>>>>=20 >>>>>>>> I would LOVE to get to the bottom of this......=20 >>>>>>>>=20 >>>>>>>> Thanks Jack >>>>>>>>=20 >>>>>>>> Phil >>>>>>>>=20 >>>>>>>>=20 >>>>>>>> Phil Spitler | Creative Technologist | Bonfire Labs | t : 41= 5.394.8200 m : 415.571.3139 >>>>>>>>=20 >>>>>>>> Website | Facebook | LinkedIn >>>>>>>>=20 >>>>>>>> On Oct 15, 2012, at 5:24 PM, Jack Tunnicliffe wrote: >>>>>>>>=20 >>>>>>>>> Okay, just did a quick test with the same keys etc you are using b= ut my Epic 5K footage is 5120x2160 so 2.37 aspect. Set a key to render out a= 10 second clip and estimated time is 4 minutes on a 12 core Mac with 32 gig= s of ram. You have something that's really slowing down your system. I am ac= cessing files through fibre on a very fast SAN, but there is definitely some= thing going on. In fact it rendered out and finished before I could finish t= yping this email.=20 --Apple-Mail-965B37B2-5FF7-4B25-B56B-CC907C4FF644 Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: quoted-printable
Well I do. I'm not trying to start a d= ebate but just finished a movie GPU decode all Epic 5k. Starting another one= tomorrow. Have done several TV series, many, many commercials, etc, this wa= y. 

I love the way Davinci handles Red, real t= ime playback at 1/4 resolution, which still looks great, even projected. Ren= ders at full rez premium.

The only difference I can= see is a slightly softer image with GPU compared to Red Rocket output. I ac= tually painted something back from a Davinci decode on top of a Red Rocket d= ecode today, where the VizFX had left a line from a poor garbage matte acros= s someone's face. The match was perfect, perfect enough to paint someone's s= kin back from another file.

Sent from my iPhone

= On 2012-10-15, at 9:43 PM, Rendernyc <rendernyc@gmail.com> wrote:

While gpu speeds up resolve I'm pretty sure no one does r3d decodin= g in gpu

Red has been pretty strict about this requ= iring either a red rocket or software decoding.

I'v= e heard many engineers and third parties grumbling about this many times ove= r the last few years.



On Oct 15, 2012, at 1= 1:04 PM, Jack Tunnicliffe <jack@javap= ost.ca> wrote:

Well I w= as avoid saying this but because you bring it up, Davinci does allowsyou to r= ender RED with GPU. 

We have a rocket card in a= system for doing digital dailies but we can also do them using Davinci Reso= lve an a $500 Cuda card. We've had our Rocket cards burn out a couple of tim= es. They work hard.

A person could install the Lite= version of Resolve, buy a Cuda card which would give you 3D acceleration in= CS6, speed up GenArts plug ins like Sapphire and monster. You'd be restrict= ed to HD resolution out with Lite but for $999 you'd have 5K out to almost a= ny other format.

I haven't rendered out 5K from Dav= inci. That might exhaust the GPU, but I've rendered out many other large for= mats.

Another 2 cents of thought.

Sent from m= y iPhone

On 2012-10-15, at 8:07 PM, Rendernyc <rendernyc@gmail.com> wrote:

<= blockquote type=3D"cite">
First thing I do if I get an r3d in AE is to= transcode it or create a 16bit "proxy" so I don't have to wait for debayeri= ng

Maybe on day red will let us debayer on the gpu b= ut still doesn't seem to be happening anytime soon. 

On Oct 15, 2= 012, at 9:42 PM, Phil Spitler <ph= il@bonfirelabs.com> wrote:

Thanks Tim, the idea would be to get a Red Rocket for my workstation then d= ump the renders onto our farm but with these slow renders that is not feasib= le.

I really want to avoid having to transcode all the fo= otage into anther format for a whole host of reasons...

=
Let's see how my CS6 test go tomorrow.

Thanks.=

Phil


=
Phil Spitler=   |  Creative Technologist   |  Bonfire Labs=   |  t : 415.394.8200  m : 415.571.3139

Website | Facebook | LinkedIn


=


=


On Oct 15, 2012, at 6:25 PM, Tim Clapham wrote:

If you are working with a lot of R3D files then perhaps it is worth= investing in a Red Rocket card.  The debayer process is realtime and c= an save you hours of time compared with using software. Although if you are n= ot dealing with many files it could be a tad expensive at just under $5K
=

___ 

Tim Clapham
Director
___

<= span style=3D"font: normal normal normal 12px/normal Arial; color: rgb(35, 3= 5, 35); ">LUXX
3D Animation
Motion Graphics
___

web : = luxx.co= m.au
blog :helloluxx.com

On 16/10/2012, at 12:10 PM, Phil Spitler wrote:

Hmmmm I'm gonna try CS6 tomorrow. 

= Can't believe how slow this is in CS5.5

Hoping it i= s ok in CS6. 

Cheers. 

Phil



Phil Spitler | &= nbsp;Creative Technologist  |  Bonfire  Labs  |  t := 415.394.8200  |  c : 415.571.3139  |  Bonfirelabs.com


On Oct 15, 2012, at 5:59 PM, Jack Tunnicl= iffe <jack@javapost.ca> wrote:=

I was rendering= out full 5K. I don't think that scaling would make much difference. I'm usi= ng the latest build of CS6.

Sent from my iPhone

On 2012= -10-15, at 6:31 PM, Phil Spitler <phil@bonfirelabs.com> wrote:

=
Weird, I am also on a really fast SAN and my files are 5k HD which are s= maller than yours.

I have mine in a 1920x1080 comp so I a= m scaling too which could be part of the problem.

W= ere you rendering 5k or 1080?

Are you using CS5.5 o= r CS6 ?

I would LOVE to get to the bottom of this..= .... 

Thanks Jack

Phi= l


Phil Spitler&nb= sp; |  Creative Technologist   |  Bonfire Labs  |  t : 415.394.8200  m : 415.571.3139

Website | Facebook | LinkedIn

On Oct 15, 2012, at 5:24 PM, Jack Tunnicliffe wrote:

Okay, just did a quick test with the same keys etc you are usin= g but my Epic 5K footage is 5120x2160 so 2.37 aspect. Set a key to render ou= t a 10 second clip and estimated time is 4 minutes on a 12 core Mac with 32 g= igs of ram. You have something that's really slowing down your system. I am a= ccessing files through fibre on a very fast SAN, but there is definitely som= ething going on. In fact it rendered out and finished before I could finish t= yping this email. 
= --Apple-Mail-965B37B2-5FF7-4B25-B56B-CC907C4FF644--