Return-Path: Received: from mail-qc0-f169.google.com ([209.85.216.169] verified) by media-motion.tv (CommuniGate Pro SMTP 4.2.10) with ESMTP-TLS id 4908510 for AE-List@media-motion.tv; Sun, 25 Nov 2012 19:24:26 +0100 Received: by mail-qc0-f169.google.com with SMTP id t2so6857521qcq.28 for ; Sun, 25 Nov 2012 10:28:03 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type:x-gm-message-state; bh=8ySzA1YaV1SUMDytUkIHmlyrBsvXBiBgbeNrHjX463E=; b=oZK6Oa6s1GuF4lwVcOtWzMLqsp5BDTwVivMnsMejB3YDhT2DObqQjvkcM+wa6tHbfs tfwxCEvu+GLZXj7nTtGJMTntFZnRCeJVPxbjZ93rQq1vmQVuK0OlbeubaAE6RyJ08ISD gJHqXHv4iYLUNqlOWSuyD+OVyh6F3cmyFFd9bl3ND1yE697uiam7SGHmiTviJLvRTwg2 zIIRlI0WdRE/P9Dlzv2Md1u38hu/kVn/Z1C3SytC6LL14M9TeJn6O6rjfezYjv6fMfD+ jR1ibOgMnGsrJdWOOqSCuZWwu5PhfKsguakFPqYPBBcvjRfq+hSui2lKKoICY61sUlLI iQng== MIME-Version: 1.0 Received: by 10.224.180.205 with SMTP id bv13mr9837784qab.7.1353868082891; Sun, 25 Nov 2012 10:28:02 -0800 (PST) Received: by 10.49.128.65 with HTTP; Sun, 25 Nov 2012 10:28:02 -0800 (PST) In-Reply-To: References: Date: Sun, 25 Nov 2012 12:28:02 -0600 Message-ID: Subject: Re: [AE] Arri flat From: Brian Higgins To: After Effects Mail List Content-Type: multipart/alternative; boundary=20cf30363f8d274a9d04cf55fa31 X-Gm-Message-State: ALoCoQldULM5aFWo7yC9deozJHmR2ubHhdW2ZtQbPn0wUsI+IYyi0bQQM3X8oPbjY16eOkNdPUvM --20cf30363f8d274a9d04cf55fa31 Content-Type: text/plain; charset=ISO-8859-1 No no no...you most likely want to convert the LogC footage into video gamma color space. We used to call this "linear," ala Log->Lin, but linear these days generally means scene-referred linear, which is another ball of wax entirely. Go here, go to the post-production section, and make an AE LUT: http://www.arri.com/camera/digital_cameras/tools/lut_generator/lut_generator.html What are you doing to the footage? Adding some CG/VFX? Adding graphics? Color correcting? And what is your client expecting as a deliverable? If you need to deliver LogC files back, then you'll need to work in a color space with high enough bit depth to hold the ~14 stops of range in that source file. I suggest working in float either way. If they want LogC ProRes files, add a LUT at the end to squish your work back down. If you're going this route, test it and get it working before you do any other work on the project. If they're fine with getting video gamma images from you, you can deliver as DPX or DNXHD or whatever. $.02, Brian -- *brian higgins | creative director Sol Design 312.706.5500 higgins@soldesignfx.com soldesignfx.com * * * On Sun, Nov 25, 2012 at 12:12 PM, Teddy Gage wrote: > OK that helps. That's sort of what I was doing already- but I'd really > like to avoid having to work on the effects directly in c-log, it's gonna > be a huge headache with color space conversion and preview. I was thinking > maybe a round trip through dnxHD 10-bit may be the answer. The final > deliverables are going to be HD5 for broadcast as far as I can tell, and > maybe a DCP. Looks like I need to have another conversation with the > director and online editor to figure out this workflow and how they're > doing the color correct. I wish they'd just shot dnxhd in the first place... > -TG > > On Sun, Nov 25, 2012 at 12:58 PM, Jim Curtis wrote: > >> I touched on this already, but what I've done is to put a first instance >> of levels on the clips or an adjustment layer to get the levels on a WFM in >> the ballpark of 1-100. Then, I add a second instance of other plugs (I >> like Frishchluft FL Curves best) for fine tuning. >> >> What you render to depends on where you're going. I was doing broadcast >> spots, so I rendered to ProRes422 (in RGB color space) and made MPEG2 for >> distribution via DG and ExtremeReach from those. >> >> To be honest, I don't know if that's a great method, but the footage >> looks great, and I love the expanded dynamic range and the minimizing of >> clipping in the whites that I often see with lesser capture formats. >> >> Monitoring: I have a calibrated HP DreamColor display and a Sony >> Trinitron CRT hooked up to a Kona for accurate monitoring. I use an Apple >> Cinema Display for my computer monitor, and I can attest that the images >> are vastly different between the computer display and the broadcast >> monitors. What looks right on the broadcast monitors is super-saturated on >> the ACD (using the default Cinema HD Apple profile - not flat). >> >> You can check yourself by looking at the scopes in Ae bundled Color >> Finesse, if you don't have any other scopes available. Or, you can check >> your work by using the scopes in Pr, if you have that. I run Synthetic >> Apps Test Gear full-time in Ae. >> >> I'm sure you're aware that home HDTVs have picture settings that alter >> the contrast, saturation, etc. Whether your HDTV gives you an accurate >> look depends on many settings and the hardware feeding your HDMI signal. >> But, if that's all you have, then see if your output compares favorably to >> what you watch on TV. If it looks right, maybe it is right. >> >> >> >> >> On Nov 25, 2012, at 11:26 AM, Teddy Gage wrote: >> >> ok looks like it is 4444, 276 mbps. I wasn't aware the arri encoded >> directly to prores so that solves one problem, namely what I thought was a >> bad transcode. Looks like these clips are direct from camera. >> >> Now what's the best way for this footage to make the round trip, FX-wise, >> without destroying the integrity of the 10bit 4444? I'm thinking: >> >> - convert to log c color space / 32bpc to work in. Match all FX footage / >> roto / stock elements to the log c preview? >> - render back out to lossless QTs in 10bit YUV from AE >> - do I render in video color space or should I render out in the log c >> profile? >> >> I don't have an HD monitor, I have a high-end "graphics" monitor (the >> eizo CG222, extremely flat response, color accurate, 99%+ of adobe RGB) but >> to get an accurate final preview I may throw it onto my HDTV (not ideal, I >> know, but at least should give me an idea what I'm looking at, right?) >> >> Any other tips / gotchas would be greatly appreciated. >> -TG >> >> >> >> On Sun, Nov 25, 2012 at 10:53 AM, Jim Curtis wrote: >> >>> You can tell if it's 422 or 4444 by looking at the clip info in the Ae >>> Project pane, or by using Get Info in QT Player. >>> >>> The Alexa footage I've worked with (just three projects, so I'm no >>> expert), the camera was connected to a cinedeck via 2 BNC cables, and the >>> native footage was ProRes4444+, or IOW, no transcoding was done or >>> necessary. >>> >>> If it was recorded to ArriRaw and transcoded with settings "baked," you >>> have been robbed of the post-production benefits of recording in RAW. >>> >>> >>> On Nov 25, 2012, at 9:40 AM, Teddy Gage wrote: >>> >>> OK so j. Penzer said the arri is using log c color space. This is the >>> answer I was looking for. Here's my concern. I am ingesting prores >>> transcodes, not the original camera files. The histogram for this footage >>> in 32bpc is compressed on both sides and very clipped in the middle >>> a) I am concerned the editor didn't do the transcodes properly. I'm not >>> even sure whether they were converted to 8 bit 4:2:2 or 10bit 4:4:4. I am, >>> unfortunately, one of the more tech savvy people on production. Is there >>> any way to figure this out? >>> >>> B) how should I handle final outputs? Should I use color space >>> conversion back to log c on export so it's consistent in fcp? Does fcp do >>> this conversion automatically? >>> >>> Thanks! >>> >>> >>> >> >> >> -- >> Animator & Editor >> www.teddygage.com >> Brooklyn >> >> >> > > > -- > Animator & Editor > www.teddygage.com > Brooklyn > > --20cf30363f8d274a9d04cf55fa31 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable No no no...you most likely want to convert the LogC footage into video gamm= a color space. =A0We used to call this "linear," ala Log->Lin,= but linear these days generally means scene-referred linear, which is anot= her ball of wax entirely. =A0Go here, go to the post-production section, an= d make an AE LUT:

What are you doing t= o the footage? =A0Adding some CG/VFX? =A0Adding graphics? =A0Color correcti= ng? =A0 And what is your client expecting as a deliverable? =A0If you need = to deliver LogC files back, then you'll need to work in a color space w= ith high enough bit depth to hold the ~14 stops of range in that source fil= e. =A0I suggest working in float either way. =A0If they want LogC ProRes fi= les, add a LUT at the end to squish your work back down. If you're goin= g this route, test it and get it working before you do any other work on th= e project. =A0If they're fine with getting video gamma images from you,= you can deliver as DPX or DNXHD or whatever. =A0

$.02,
Brian

--=A0
= brian higgins | creative director
Sol =A0Design
312.706.5500
higgins@soldesignfx.com
soldesignfx.com


On Sun, Nov 25,= 2012 at 12:12 PM, Teddy Gage <teddygage@gmail.com> wrote:=
OK that helps. That's sort of what I was= doing already- but I'd really like to avoid having to work on the effe= cts directly in c-log, it's gonna be a huge headache with color space c= onversion and preview. I was thinking maybe a round trip through dnxHD 10-b= it may be the answer. The final deliverables are going to be HD5 for broadc= ast as far as I can tell, and maybe a DCP. Looks like I need to have anothe= r conversation with the director and online editor to figure out this workf= low and how they're doing the color correct. I wish they'd just sho= t dnxhd in the first place...
-TG

On Sun, Nov 25, 2012 at 12:58 PM, Jim= Curtis <jpcurtis@me.com> wrote:
I touched on this already, but what I&#= 39;ve done is to put a first instance of levels on the clips or an adjustme= nt layer to get the levels on a WFM in the ballpark of 1-100. =A0Then, I ad= d a second instance of other plugs (I like Frishchluft FL Curves best) for = fine tuning.

What you render to depends on where you're going. =A0I w= as doing broadcast spots, so I rendered to ProRes422 (in RGB color space) a= nd made MPEG2 for distribution via DG and ExtremeReach from those.

To be honest, I don't know if that's a great method,= but the footage looks great, and I love the expanded dynamic range and the= minimizing of clipping in the whites that I often see with lesser capture = formats. =A0

Monitoring: =A0I have a calibrated HP DreamColor displa= y and a Sony Trinitron CRT hooked up to a Kona for accurate monitoring. =A0= I use an Apple Cinema Display for my computer monitor, and I can attest tha= t the images are vastly different between the computer display and the broa= dcast monitors. =A0What looks right on the broadcast monitors is super-satu= rated on the ACD (using the default Cinema HD Apple profile - not flat). = =A0

You can check yourself by looking at the scopes in Ae b= undled Color Finesse, if you don't have any other scopes available. =A0= Or, you can check your work by using the scopes in Pr, if you have that. = =A0I run Synthetic Apps Test Gear full-time in Ae.

I'm sure you're aware that home HDTVs have pict= ure settings that alter the contrast, saturation, etc. =A0Whether your HDTV= gives you an accurate look depends on many settings and the hardware feedi= ng your HDMI signal. =A0But, if that's all you have, then see if your o= utput compares favorably to what you watch on TV. =A0If it looks right, may= be it is right.




O= n Nov 25, 2012, at 11:26 AM, Teddy Gage wrote:

ok looks like it is 4444, 276 mbps. I wasn't aware the arri enco= ded directly to prores so that solves one problem, namely what I thought wa= s a bad transcode. Looks like these clips are direct from camera.

Now what's the best way for this footage to make the round trip, FX= -wise, without destroying the integrity of the 10bit 4444? I'm thinking= :

- convert to log c color space / 32bpc to work in. Match all FX footage= / roto / stock elements to the log c preview?
- render back out to loss= less QTs in 10bit YUV from AE
- do I render in video color space or sho= uld I render out in the log c profile?

I don't have an HD monitor, I have a high-end "graphics" = monitor (the eizo CG222, extremely flat response, color accurate, 99%+ of a= dobe RGB) but to get an accurate final preview I may throw it onto my HDTV = (not ideal, I know, but at least should give me an idea what I'm lookin= g at, right?)

Any other tips / gotchas would be greatly appreciated.
-TG


On Sun, Nov 25, 2012 at 10:53 AM, Jim Curt= is <jpcurtis@me.com> wrote:
You can = tell if it's 422 or 4444 by looking at the clip info in the Ae Project = pane, or by using Get Info in QT Player.

The Alexa footage I've worked with (just three projects,= so I'm no expert), the camera was connected to a cinedeck via 2 BNC ca= bles, and the native footage was ProRes4444+, or IOW, no transcoding was do= ne or necessary. =A0

If it was recorded to ArriRaw and transcoded with setti= ngs "baked," you have been robbed of the post-production benefits= of recording in RAW.


On Nov 25, 2012, at 9:40 AM, Teddy Gage wrote:

OK so j. Penzer said the arri is using= log c color space. This is the answer I was looking for. Here's my con= cern. I am ingesting prores transcodes, not the original camera files. The = histogram for this footage in 32bpc is compressed on both sides and very cl= ipped in the middle
a) I am concerned the editor didn't do the transcodes properly. I'm= not even sure whether they were converted to 8 bit 4:2:2 or 10bit 4:4:4. I= am, unfortunately, one of the more tech savvy people on production. Is the= re any way to figure this out?

B) how should I handle final =A0outputs? Should I use color = space conversion back to log c on export so it's consistent in fcp? Doe= s fcp do this conversion automatically?=A0

Thanks!<= /span>





--
Animator & Editor
www.teddygage.com Brooklyn





= --
Animator & Editor
www.teddygage.com
Brooklyn







--20cf30363f8d274a9d04cf55fa31--