This discussion is locked
RicM
Level 1 (10 points)

Q: Local Edit Proxy VS Rendering on Server

My projects are shot on HDV, uploaded as productions to FCSvr with edit proxies set for ProRes 422 proxy. Many of these projects contain Photoshop graphics, usually "builds" of bullet points or logos. Will I see a difference in image quality, specifically the graphics, between rendering locally with checked out edit proxy files and outputting the project to compressor on the FCSvr, which is, as I understand it, working from the original HDV files and original Photoshop layered files? Does FCSvr modify Photoshop files that are downloaded for local editing in any way?

dual 2.8 Gig Quad Core Xenon, Mac OS X (10.6.4)

Posted on Dec 22, 2010 2:51 AM

by A. Richards,Solvedanswer
A. Richards A. Richards
Level 3 (625 points)
A:
If I understand, you are cutting in 1080 ProRes that originated as HDV. You finish at that resolution with the appropriate graphics and then export 640x360 for the web. Makes sense, and with the dowwnsampling they probably look very clean. But if your clients blow that same web file up full screen on an HD monitor it looks pretty crunchy, especially near text since that is where any method of scaling breaks down.

All you need to do is make three outputs for each project. Make your usual 640x360 and then also make a couple h.264 mov files at 1280x720 and 1920x1080. They will be compatible with almost any type of playback device or would at least be reasonable to transcode again for signage duty. And since they were encoded from an HD source to an HD resolution, they will look great displayed on any HD monitor.

Posted on Dec 27, 2010 1:43 PM

Close
RicM

Q: Local Edit Proxy VS Rendering on Server

  • All replies
  • Helpful answers

  • by A. Richards,

    A. Richards A. Richards Dec 22, 2010 6:25 AM in response to RicM
    Level 3 (625 points)
    Dec 22, 2010 6:25 AM in response to RicM
    Whether you export locally or submit the job to FCSvr, the source for the transcode will be the edit proxies FCP is currently working with. FCSvr does not relink to other media for transcoding jobs submitted from FCP. It only relinks FCP project files to the original media when they are checked back in. The only ways to export from the original media is to check out the project with the original media and export locally from FCP or open the project on a system that is set up for edit-in-place, in which case the original media remains linked to the checked-out project and will be the source for any exports.

  • by RicM,

    RicM RicM Dec 22, 2010 1:50 PM in response to A. Richards
    Level 1 (10 points)
    Dec 22, 2010 1:50 PM in response to A. Richards
    Thank you for your explanation. I think that is pretty much how I understood it. My question was more about the visual difference between the two outputs and specifically how that relates to composited graphics. There are three elements to my question:

    1) When a graphic element like a layered Photoshop file is part of a FCP project and goes roundtrip to FCSver and back, has it been altered in any way?

    2) HDV footage is about the same data rate as ProRes Proxy (which is what I have my edit proxies set for) and looks very similar in quality, but edits much faster (no conforming HDV) which is why I trancode at all. Is there a visual quality difference between the original HDV and the ProRes footage?

    3)When the graphics (#1 above) are composited with the ProRes or HDV (#2) should I see cleaner results from a local output render from FCP or a render from FCSver?

  • by RicM,

    RicM RicM Dec 22, 2010 5:31 PM in response to RicM
    Level 1 (10 points)
    Dec 22, 2010 5:31 PM in response to RicM
    OK I have read some more and I'm still working my own tortured way through a fuller understanding of this workflow. I have 3 iMac edit stations on a gigabit ethernet network and a tower running FCSvr that also has a FCP on it. If I want the best, closest to original image quality output, I should finish my edit on one of the stations and check it back in. Then, on the tower running the server and FCP, download as edit in place (everything is on the tower, 2 internal raid 1 volumes) and then "send to" Compressor with the appropriate settings & destinations for the particular project. Is that correct?

  • by RicM,

    RicM RicM Dec 25, 2010 10:38 AM in response to RicM
    Level 1 (10 points)
    Dec 25, 2010 10:38 AM in response to RicM
    After more reading I am thinking that with the setup described above it would make sense to edit with ProRes Proxy on the iMac clients, and check the project back in to FCSvr. Then open the project on the tower that runs both FCsvr & FCP except check out with original files (which are already on the FCSvr tower's internal storage) rather than proxies. I then export the finished file to my watcher folder that transcodes to each size & resolution through my automations. Does this sound like the path to best quality? Suggestions? Opinions?

  • by A. Richards,

    A. Richards A. Richards Dec 27, 2010 11:28 AM in response to RicM
    Level 3 (625 points)
    Dec 27, 2010 11:28 AM in response to RicM
    Sorry I haven't replied sooner. Holidays.

    1) FCSvr doesn't process stills that are used in FCP projects, it only makes note of their link to the project so it can check them out if necessary. When FCP is addressing a still it will either be the original on shared storage or a copy of the original that was downloaded upon checkout of the FCP project from FCSvr.

    2) I doubt there is much of a quality difference between HDV and ProRes Proxy. All ProRes codecs are 10 bit and are designed to not degrade across multiple transcode passes. If anything, ProRes Proxy is capable of higher image fidelity than the 25Mbps MPEG-2 used by HDV. However, since you are starting with HDV, you don't gain anything in the upsample to ProRes. But you certainly don't lose anything either. Maybe some quantitative image analysis would show slight differences, but I bet you can't see any with just your eye.

    3) It should be a wash. I'm pretty sure FCP renders HDV as ProRes anyway (might be wrong, but there are definitely other long-GOP codecs FCP handles that way). Either way, the fact that ProRes won't degrade the image data of the HDV means that using either one as a source should yield functionally identical output quality. FCSvr doesn't do any rendering. All it does is catalog media and hand that media off to Compressor for any transcoding. Any rendering of a composite will take place via FCP based upon whatever media it is linking to.

    Honestly, if you don't see a difference in image quality looking at your HDV and the ProRes Proxy derived from it side by side, I see no reason why you shouldn't just do all your work and outputs in ProRes. Save the HDV as a source and a backup, but I don't think you are missing anything keeping your workflow in ProRes after ingest and transcoding.

  • by RicM,

    RicM RicM Dec 27, 2010 1:23 PM in response to A. Richards
    Level 1 (10 points)
    Dec 27, 2010 1:23 PM in response to A. Richards
    AR,
    Thanks for your reply and the insights within.

    This was a big question for me, and I did not really see it addressed in the offline (online for me) workflow documentation. Most of our work ends up on the web, Flash, at 640x360, so HDV footage is more than adequate for our needs. We do find customers asking to repurpose material to trade show screens, typically 42-50 inch, on occasion. The live footage, while not probably to many users of this forums' standards, looks OK, it's type in (photoshop) graphics, Illustrator eps logos, and and screen capture of software that really goes to **** quickly at these larger sizes. I now understand that if I want to go back to "original media" when I need to repurpose for the larger screen I should ingest graphics as is, but be sure to composit, and ingest other formats (screen capture and AE title sequences for example) at higher levels of ProRes.

  • by A. Richards,Solvedanswer

    A. Richards A. Richards Dec 27, 2010 1:43 PM in response to RicM
    Level 3 (625 points)
    Dec 27, 2010 1:43 PM in response to RicM
    If I understand, you are cutting in 1080 ProRes that originated as HDV. You finish at that resolution with the appropriate graphics and then export 640x360 for the web. Makes sense, and with the dowwnsampling they probably look very clean. But if your clients blow that same web file up full screen on an HD monitor it looks pretty crunchy, especially near text since that is where any method of scaling breaks down.

    All you need to do is make three outputs for each project. Make your usual 640x360 and then also make a couple h.264 mov files at 1280x720 and 1920x1080. They will be compatible with almost any type of playback device or would at least be reasonable to transcode again for signage duty. And since they were encoded from an HD source to an HD resolution, they will look great displayed on any HD monitor.

  • by RicM,

    RicM RicM Dec 30, 2010 11:03 AM in response to A. Richards
    Level 1 (10 points)
    Dec 30, 2010 11:03 AM in response to A. Richards
    Yes, I would never expect the 640x360 to play on the big screen application. Thanks so much for your insights.

  • by BenB,

    BenB BenB Jan 14, 2011 1:00 PM in response to A. Richards
    Level 6 (10,041 points)
    Audio
    Jan 14, 2011 1:00 PM in response to A. Richards
    1) No.

    2) HDV should be ingested as ProRes 422 anyway to start with, as it's a Intra-Frame codec, not a Long-GOP codec, and edits way easier, faster, and transcodes MUCH faster than HDV's Long-GOP format.

    3) FCSvr does not "render" anything, the rendering is done on the local machine. FCP will render codecs naively unless you go into the Settings and tell it to render everything as ProRes. HDV will take longer to render, be more of a PITA to edit with. You won't see any image degradation at all going from HDV to ProRes. You "may" see a difference with ProRes Proxy when editing with Proxies. But you won't with the original media. Again, HDV should be ingested into FCP as ProRes 422 to start with, in order to make life easier in the long run. DSLR footage should be ingested as ProRes LT.