2 Replies Latest reply on Jan 22, 2013 3:41 AM by Dave Merchant

    Is this ProRes theory a myth or is it true?

    yash-lucid Level 1

      I shoot AVCHD footage from an FS700 and edit and heavily grade natively in Premiere Pro CS6.


      I was told that if I convert my footage to ProRes and THEN edit and grade, the colours and details will hold up better than doing so with native AVCHD and will ultimately result in better looking footage.


      Here is a quote from the forum to justify that. Please can someone confirm if this is true?


      AVCHD is a very compressed and lossy format. Original footage shot in AVCHD will mostly look good, but the trouble begins when you start recompressing the footage. Recompression happens whenever you convert the footage to another format, or whenever you use the footage in an editing environment and you have to render your video.

      When you edit AVCHD natively every time you add an effect, title or whatsoever you will lose quality and this will get worse the more generations you make from your original footage. That's why effects, titles etc. are always rendered to ProRes 422 (or higher) in FCP10. Because ProRes withstands multiple re-encoding much better than AVCHD.

      When you first convert your AVCHD to ProRes you will get a first generation loss. But the advantage of ProRes is that it is a transparent format, i.e. once your video is in the ProRes format it will not further degrade even after multiple re-encoding. ProRes LT runs at 82 Mbps but that does not mean anything. The LT codec is much more lossy than ProRes 422 and 422 HQ, which means once again that it will not withstand multiple generations. That's why the best workflow is to convert any lossy format to a transparent format such as ProRes 422 and up. You will always lose a tiny bit of sharpness during the first conversion, but from there on you will be safe even when you use the ProRes footage in complex projects with multiple filter stacks.

        • 1. Re: Is this ProRes theory a myth or is it true?
          Dave Merchant MVP & Adobe Community Professional

          Not true for your workflow.


          In some other (non-Adobe) applications, footage is repeatedly changed and re-saved as it runs through a series of different apps, so it will deteriorate if a lossy codec is used. In Premiere Pro the timeline can have any number of things applied, and it's only rendered once at the export stage. The only exception is if you pass footage off to SpeedGrade, when a lossless image sequence is used.


          The only reason some Premiere users transcode to ProRes or AVI is to take some of the pressure off the CPU when scrubbing about the timeline. Camera footage in H.264 or AVCHD requires a lot of CPU cycles to decompress, so on a low-spec machine it can be stuttery. ProRes/AVI is easier to read  but at the cost of a lot more disk space, so if your disk throughput is slow the bottleneck is simply moved.


          Transcoding has no effect whatsoever on the quality of the output from Premiere Pro, as Premiere always upsamples everything on every timeline to 32-bit floating point. It may have a slight effect on how fast the render process runs, but nothing to get excited about.

          2 people found this helpful
          • 2. Re: Is this ProRes theory a myth or is it true?
            Harm Millaard Level 7

            A myth.


            When editing with FCP, there was a natural tendency to convert a lot of source material to ProRes 422, simply because you had no choice. Now you have the choice after moving to PR.


            The choice is quite simple, either edit in the native format your material was shot in, or convert to ProRes 422. The advise is to

            Not convert to ProRes.

            Using native format has the following advantages: No transcoding, no rendering, no data change, no recompression, no degradation and no damage. But converting to ProRes 422 has singnificant disadvantages, it takes twice as long, it uses nearly three times the disk space and shows no visible improvement. For example, ingesting a 16 GB card with MXF material on a certain machine takes around 6 minutes in native format but takes around 12 minutes when transcoded to ProRes 422, the storage requirements go from 16 GB to around 43 GB without any visible difference.


            Additionally, ProRes can not be exported from a PC and has problems with gamma shifts.


            The post you quoted has a serious flaw in it. It assumes multiple exports and imports, creating generation losses. That is a very strange workflow. One would normally edit and once finished, export once. That gives the same generation loss as converting to ProRes to start with. Only with ProRes you will suffer another generation loss when exporting from that.