9 Replies Latest reply on May 7, 2010 11:03 AM by Wil Renczes

    Is CS5's "DSLR Editing" truly Native?

    akira3d Level 1

      I finally got to try the "DSLR Editing" presets after registering my copy of Production Premium CS5...and am fairly impressed by how easily it handles the 5D Mark II MOV files.  I have composited up to five layers of 1080/24p clips and had realtime playback (as long as the monitor is set to 1/2 resolution...no CUDA hardware here), so I am genuinely pleased by how much better this is than CS4.  I have also been impressed that encoding is actually faster than playback.  Both wins in my book.

       

      However, I question whether or not the preset sequences will render truly "Native" when footage has not been altered.

       

      The docs state that a yellow bar over the timeline "indicates that a clip does not match the settings of the sequence, but can generally still
      be played back in real-time without rendering."  The original MOV files created by my 5D Mark II clips have this yellow bar when matched with their respective preset (1080p 24 fps ).  If I basically do a straight export of the original MOV file (using the "Match Sequence Settings"), I notice that the video encoder selected is MPEG-I with no adjustable settings or options.  The resulting file is in a .MPEG container and is considerably smaller than the source (lower bitrate too), another warning sign that the file has been recompressed.  Inspecting the file suggests that the video is now MPEG-PS...heck, even the audio has been compressed.

       

      This does not fit my definition of native.  To me, native means Premiere Pro should simply pass an unprocessed sequence without recompression (.h264  /uncompressed audio).  The container format can change...I have no issue with that.

       

      Bringing the resulting .MPEG file back on to a "DSLR Editing" timeline has no color bar over it...which suggests to me that MPEG-I is the native format for this sequence.

       

      Am I doing something wrong?  Or is this how "DSLR Editing" presets are supposed to work?

        • 1. Re: Is CS5's "DSLR Editing" truly Native?
          shooternz Level 6

          Editing a file in its Native format means exactly that.

           

          It edits the fle without need to transcode it to another format / CODEC

           

          Exprting a file is another matter. It is entirely your choice what you export it as. The Sequence preset does not determine the export setting.

           

          BTW: In your case ...your camera is producing a compressed file.

          • 2. Re: Is CS5's "DSLR Editing" truly Native?
            Colin Brougham Level 6

            This does not fit my definition of native.  To me, native means Premiere Pro should simply pass an unprocessed sequence without recompression (.h264  /uncompressed audio).  The container format can change...I have no issue with that.

            Then you'll have to redefine "native" for purposes of CS5. "Native" editing means that you don't need to transcode your source files to something else in order to import and edit them. What you're talking about is usually referred to as "smart encoding" or "smart rendering," which means that footage that is not altered/"effected"/edited is able to be passed from a source file, through the NLE's machinery, into an output file, unfazed and unscathed. Premiere Pro is not able to do this, at least not with MPEG-based assets. Since most folks are not going back to the same format from whence they came, and since most folks are also changing their footage while it's in the editor's timeline (e.g. color correction, effects, titles, etc.), this is SOP.

             

            If you have to go back to the same format/codec that you started with, then you'll have to accept Premiere's reprocessing of the footage. I know there is at least one plug-in (from MainConcept, I believe) that enables smart rendering of MPEG assets, but I'm not sure if it's available for CS5 or ever will be.

             

            Ultimately, a sequence preset/setting has no bearing on what your export format is or will be. For example, I edit DVCPROHD in a DV sequence all the time, with a perpetual red bar, but that footage is never rendered as DV. I export to MPEG-2 for DVD or H.264 for the web or something else, but the footage is never being rendered as DV. This is still native, because I'm able to work with the DVCPROHD files perfectly well, even if the program is flagging them as incompatible with the sequence I'm using. The very fact that you can bring your 5D files in, edit them, stack them five deep, and have an otherwise pleasant experience tells me that, yes, Virginia, this is a native workflow.

            • 3. Re: Is CS5's "DSLR Editing" truly Native?
              akira3d Level 1

              I guess I have mistakenly equated native editing with smart rendering...based on the assumption that my old DV workflow would eventually pertain to my DSLR.  Perhaps I should call this naive editing since I keep upgrading Premiere Pro hoping it will one day work as intelligently in HD as Premiere once did for standard def DV.

               

              Don't get me wrong, I am very happy with the performance of Premiere Pro CS5 with the 5D Mark II...something I couldn't say with CS4.  This is something of a relief considering that I do not have one of the supported GPUs...was afraid I'd have to swap out my old ATI card.  I am very glad I held off on purchasing an interim codec solution (or wasted money on larger capcity hard drives for my RAID0 just to handle uncompressed clips) because CS5's performance and Adobe Dynamic Link (my favorite feature from CS4) definitely eliminates the need for an intermediate render before bringing the sequence into After Effects and/or Encore.  I am very happy that the MPEG files generated when using "Match Sequence Settings" with the DSLR Editing mode render fast, look great, and playback smoothly on both my PC and PS3 without the need to fiddle with any settings...though I do plan ot target very particular .h264 settings on final output (especially for Blu-ray or the Web).

               

              I guess the only reasons I am bummed pertain to having to keep my source clips around until I am perfectly satisfied I have transcoded my footage into all of the delivery formats I ever intend to deliver.  That and I'll have to keep QuickTime Pro around a little longer so I can trim excess from my original MOV files without recompressing them.

               

              That all said, if the next version does not support smart rendering with h264 clips, I likely will not upgrade.  I think this is my fourth purchase of a suite containing Premiere Pro, and my fifth version of Premiere (I think 6.0 was my first).  This version performs well enough to keep me happy for the forseeable future.

              • 4. Re: Is CS5's "DSLR Editing" truly Native?
                Curt Wrigley Level 4

                Actually Pr does work very smoothly with HD.  You just happen to be using an extremely compressed version of HD, that is rather hard to decode for editing.

                • 5. Re: Is CS5's "DSLR Editing" truly Native?
                  shooternz Level 6
                  I guess the only reasons I am bummed pertain to having to keep my source clips around until I am perfectly satisfied I have transcoded my footage into all of the delivery formats I ever intend to deliver.

                   

                  Whats the issue with that?.  I keep all my source footage.

                   

                  BTW - You maybe talking about exporting not transcoding in your quoted sentence.  Two different things entirely.

                  • 6. Re: Is CS5's "DSLR Editing" truly Native?
                    akira3d Level 1
                    Whats the issue with that?.  I keep all my source footage.

                     

                    I usually keep my source footage, but trim away stuff I absolutely know for a fact that I will never use to save space.  I would use Premiere for this if I could export without recompressing, but will instead continue to use QuickTime Pro.  I do not want to waste space exporting uncompressed if I can avoid it...and refuse to compress more than two generations (i.e. source camera material is compressed, final output will be compressed...no compression in between). This is why I like Adobe Dynamic Link so much as I do most of my video processing in After Effects.

                     

                    Back when I used to shoot more DV format stuff on miniDV, I often recorded long events...and would record far more footage than I ever planned to use.  As a first pass, I would assemble loose cuts of footage on a timeline (no transitions, no effects), to get an idea of what looked usable and a general feel for how I wanted to sequence the material...without trimming potentially useful ends from the shots (so I could use them in transitions, for inserts, whatever).  Once I was happy that the footage was indeed all I planned to use, I would export a DV clip without recompressing.  Knowing that no data was lost, I felt comfortable ditching the source clips.  I could treat the new sequence as my source and move forward...a good way to get a few tapes worth of footage down to a single tape...and I wouldn't mind keeping a single tape as a master (10GB was a large chunk to leave on the hard drive in those days).  Which reminds me...I still have a lot of video that only exists on miniDV and microMV tapes.

                     

                    I don't plan to shoot video in a similar manner with my 5D Mark II, so keeping source footage isn't nearly the issue it once was.  Plus, I have a 2TB RAID for storing my source assets...and can easily burn DVDs or BDs when I need to free up space.

                     

                    For clarification, I am not complaining about the performance of Premiere Pro's playback of 5D Mark II .h264 video.  I just though native editing meant smart rendering...a feature I would still like to have.

                    • 7. Re: Is CS5's "DSLR Editing" truly Native?
                      joshtownsend Level 2

                      Mini dv and...say DVCprHD are both record every single frame making it easier to trim native footage without losing a generation. The DSLR's shoot with compression that is based on other frames...is it even possible to make trims with re-encoding?

                      • 8. Re: Is CS5's "DSLR Editing" truly Native?
                        Colin Brougham Level 6

                        Mini dv and...say DVCprHD are both record every single frame making it easier to trim native footage without losing a generation. The DSLR's shoot with compression that is based on other frames...is it even possible to make trims with re-encoding?

                        Josh,

                         

                        It actually is possible to edit MPEG-based footage (and probably other forms of interframe-compressed material) without re-encoding. Basically, edits just need to be made between GOPs in the datastream; at those points "whole" I frames exist, which serve as references for the B (bidirectional) and P (predicted) frames. If you preserve the I frames, nothing changes in the B and P frames between them, so no re-encoding is necessary. The two parts are just stuck together and copied into a new destination file.

                         

                        Some software, like Womble DVD Editor or TMPGEnc MPEG Editor make this relatively easy, though as far as I'm aware, they only work with MPEG-1 and MPEG-2 flavors. There may be others out there that offer MPEG-4 editing, but I'm not immediately aware of them. Additionally, both of these programs offer "smart rendering" where they will only re-encode frames as necessary if you happen to make an edit that is not on a GOP boundary. I've used TMPGEnc MPEG Editor for years, and it's always done what I needed it to do.

                        • 9. Re: Is CS5's "DSLR Editing" truly Native?
                          Wil Renczes Adobe Employee

                          I'll point out though - if you're applying *any* effects or using AE via DL, those segments can't be smart rendered.  Smart rendering only helps when you have naked clips on the timeline, and you're mastering back to the same (long GOP - as others have pointed out, with I frame formats, this is also moot) format as the source clips.  The minute you do anything to those clips other than trim them, smart rendering can't help.  The only user segment that I hear from where this truly speeds things up is from broadcasters/news editors, where the work flow is pretty much cuts only with the odd dissolve thrown in.