9 Replies Latest reply on Jun 8, 2009 1:57 PM by Kuro M

    Exporting workflow from PP CS4 for intermediate files

    Kuro M

      Hi there,

       

      I have native footage from Sony EX1 (mp4 or MXF container with mpeg 2 long GOP codec) which I cut and slightly filter/modify in Premiere Pro CS4.1.
      Most times I shoot in 1080/50i (yes, I need that format) and need both: FullHD for Blu-ray and PAL-SD for DVDs of the same footage.
      Most of you will confirm, that simply downscaling with AME gives terrible results. So I use eg. Dan Isaacs great workflow with VDub and AviSynth to deinterlace and downscale and then encode with TMPGEnc Xpress for DVDs. This gives somehow good results but I am unsure, if I am doing right.

      Here is the problem: CS4 does not allow debugmode frameserver anymore so I have to render an intermediate file for further use eg with AviSynth.

      But how do I get fully valid intermediate video files out from CS4? Uncompressed AVI is pure horror, its slow rendering, even on an i7-PC, and giant in file size. I would like to export with HuffYuv or Lagarith codec, for example. But everybody says, EX1-footage is, as all HD footage, TopFieldFirst. When exporting "Microsoft AVI" with those codecs one has to choose BottomFieldFirst - all other tries result in black video. I assume only BFF is valid in Microsoft AVI, somehow.

      I would like to use Cineform but converting footage with NeoScene is only possible from HDV/AVCHD-camcorders and when exporting with its codec as Microsoft AVI it has the same BFF-only-problem. And I cannot afford NeoHD which would do my job perfectly with EX1-footage.

      How the heck do you export HD footage (with TopFieldFirst) for further use from PP CS4, please?

      Any hint is appreciated, at the moment I am lost,
      thanks a lot in advance!

        • 1. Re: Exporting workflow from PP CS4 for intermediate files
          Jeff Bellune Adobe Community Professional

          The 4.1 update cures the Lagarith and HuffYUV black video problem.

           

          I use 4:2:2 I-frame-only MPEG2 files at a CBR of 100 Mbps for my intermediate exports.  It's fast and virtually lossless.  The only problem is disk space.  You'll need a lot of it.

           

          Create a .d2v index file for the intermediate MPEG2 file using DGIndex.  You'll be able to use the .d2v file in your AviSynth script by setting MPEG2Source() to point to it (instead of AviSource).  Then you use VirtualDub to export the downscaled .avi file.

           

          -Jeff

          • 2. Re: Exporting workflow from PP CS4 for intermediate files
            Kuro M Level 1

            Thank you so much for taking part!

            The idea of using "perfect" MPEG2 is nice but somehow I am a little hesitant to have another mpeg2-compression in the workflow. Is the MPEG2 the way you described nearly as good as "lossless"?

             

            Jeff Bellune wrote:

             

            The 4.1 update cures the Lagarith and HuffYUV black video problem.

             

             

            Yes, it seems something happened. I use 4.1 and just exported

             

            • Microsoft AVI
            • choosing Lagarith lossless codec 1.3.9 with YUY12, 1920x1080, 32bit, maximum depth, top field first

             

            The filesize is now very similar to bottom field first (it was much smaller before PP 4.1) and although it still plays black after importing into timelime of PP CS4.1 it looks fine in virtual dub. I will check if I can use the workflow from Dan Isaacs with AviSynth only (no VDub) and see if I get something out of TMPGenc with that...

             

            Thank you so much for the hint!

            • 3. Re: Exporting workflow from PP CS4 for intermediate files
              joshtownsend Level 2

              Kuro M wrote:

               

               

              Most of you will confirm, that simply downscaling with AME gives terrible results.

               

              If you click the on the flyout AME and set it to use 'c'. It downscales great. Takes longer to render but not nearly as long as using an intermediate file.

              From http://help.adobe.com/en_US/PremierePro/4.0/WS630B3868-F095-46fe-8690-A5BDBCFD6EC8.html

               

              Maximum Render Quality:
              Maintains sharp detail when scaling from large formats to smaller formats, or from high-definition to standard-definition formats. Maximum Render Quality maximizes the quality of motion in rendered clips and sequences. Selecting this option often renders moving assets more sharply.

              At maximum quality, rendering takes more time, and uses more RAM than at the default normal quality. Select this option only on systems with sufficient RAM. The Maximum Render Quality option is not recommended for systems with the minimum required RAM.

              • 4. Re: Exporting workflow from PP CS4 for intermediate files
                _dbk Level 1

                Are you seeing comparable results from "Render Using Maximum Quality" compared to the 3rd-party work-around method?

                • 5. Re: Exporting workflow from PP CS4 for intermediate files
                  Jeff Bellune Adobe Community Professional

                  It downscales great. Takes longer to render but not nearly as long as using an intermediate file.

                  I agree with your assessment of downscaling using Maximum Render Quality.  I completely disagree with you about the length of time needed for an intermediate file.

                   

                  Some notes about MRQ:

                  1. You get better deinterlacing and scaling when you nest an HD sequence inside of an SD sequence and then export that, as opposed to just exporting the HD sequence to SD using the settings in the AME.
                  2. Nesting-then-exporting takes 4 times longer than exporting to SD settings, but for 1080i/p material it's worth the extra time.  There is less of a difference (if there is a difference at all) when using 720p material.

                   

                  I can export an HD 4:2:2 I-frame-only MPEG2 intermediate file at a CBR of 100 Mbps from Pr, index the exported file in DGIndex, convert the HD file to SD using VirtualDub, AviSynth and Dan's hd2sd() script, encode the resulting .avi file to MPEG2 DVD using Squeeze, and then author and burn a DVD in Encore in less than half the time it takes MRQ to just export the nested SD sequence to MPEG2 DVD from Pr.  And the results of hd2sd() are superior to anything I can get out of Pr.

                   

                  -Jeff

                  • 6. Re: Exporting workflow from PP CS4 for intermediate files
                    Ann Bens Adobe Community Professional & MVP

                    Jeff,

                    Which codec do you use for the intermediate files?

                    What's your source?

                    How long does it take to convert 1 minute with MRQ?

                    Hope you do not mind me asking?

                    • 7. Re: Exporting workflow from PP CS4 for intermediate files
                      Jeff Bellune Adobe Community Professional

                      Hi Ann,

                       

                      I use MPEG2 as the codec.  When you set M and N frames to 1 in the export settings dialog, you get an I-frame-only MPEG file.  That's essentially an intra-frame compression as opposed to MPEG2 DVD encoding, which is a temporal, inter-frame compression (and very lossy).  When I set the bit rate for the I-frame-only MPEG2 file to 100 Mbps, it's visually lossless.  I set the Profile to 4:2:2 to preserve the color data and the Level to High to allow the very high bit rate.

                       

                      Sources for my tests have been AVCHD footage from a Panasonic HMC-150 camera (not mine, unfortunately!) and some AVC and MPEG2 HD transport streams.  I've tested 1080i60, 1080p24, 1080p30, 720p60, 720p30 and 720p24.

                       

                      I don't use MRQ to export the intermediate because no scaling or deinterlacing needs to be done to the HD footage.  The intermediate file gets exported at the same frame size and field order as the source.  All of the scaling and deinterlacing (if necessary) is done by the hd2sd script in AviSynth.  So my intermediate exports run just a bit faster than real time on my 8-core system.

                       

                      (Big kudos again to Dan Isaacs not only for the hd2sd script, but for giving me the proper settings for the MPEG2 intermediate export.)

                       

                      -Jeff

                      • 8. Re: Exporting workflow from PP CS4 for intermediate files
                        Ann Bens Adobe Community Professional & MVP

                        Thanks for the explanation.

                        Yes at one time Dan (I love that guy) gave some settings for MPEG here on the forum; lost them

                        Would like to give this MPEG workflow of yours a spin though, at the present i am fiddling with Cineform.

                        • 9. Re: Exporting workflow from PP CS4 for intermediate files
                          Kuro M Level 1

                          Sorry, I was on route for some days and could do only few tests between. I am very happy now with exporting with Lagarith codec and TFF from PP 4.1. I will try again Cineform NeoScene 1.3 (hopefully the new version will allow another trial period) for the intermediate avi since it uses much less space. But its hard to believe it will be "as lossless as Lagarith". I will try your mpeg2 hint soon too just to compare.

                           

                          I could not get AME to output anything as good as I get with VDub and/or AviSynth out of 1080/50i so I stopped trying that.

                           

                          I now export my Sony EX1 1080/50i-footage with Lagarith 1.3.9 with YUY12 color space, 1920x1080, 32bit, maximum depth, top field first.

                          Since I encode with TMPGEnc XPress 4 I have to convert to RGB24 somewhen, I think.

                           

                          I am just confused because of getting slightly better results with VDub only doing the deinterlacing (smart deinterlace) and resizing with Lanczos 3 than with Dan's advanced script using hd2sd("1080i.avi", OutputColorSpace="RGB24", interlaced=false). Its very little sharper and richer in color and with little less artefacts/jagging.

                           

                          Mmmmhh, I have to try more, I did not understand the YUV-RGB-conversion I think...

                           

                          Thanks again!