10 Replies Latest reply on Nov 12, 2010 3:02 PM by Dave Dugdale

    Suggestions Wanted On Best Way To Test Encoding for the Web

    Dave Dugdale

      There are many questions out there asking what are the best settings for encoding your videos for sites like Vimeo and YouTube, only to be answered with the "you need to experement around to see what works well for you".

       

      I have a 30 second Sequence ready to test encode that really stresses out the encoding process with fast moving objects, random objects that fill the frame like rapids and simple stuff like keyed talking-head filmed at 24fps on my DSLR.

       

      I plan on running exhaustive tests to find out what works best for me publishing to these sites.

       

      But there are soooo many parameters to test, I want to make sure that I test each one at a time in the correct order. I am mostly interested in H.264 format (MainConcept).

       

      Here is the order I am thinking of:

       

      1. Start with the Preset "Vimeo HD": NTSC, 1280x720, 23.97fps, Progressive, VBR, 1 Pass, Target 4.00, Max 5.00 Mbps
      2. Same as above, but just change to 2 Pass.
      3. Same as above, but just change Profile to Baseline.
      4. Same as above, but just change Profile to High.
      5. Go back to Profile = Main, then run through all level settings.
      6. Repeat above for Profile = Baseline.
      7. Repeat above for Profile = High.
      8. Go back to Profile = Main, and then run through many of the Target and Max bitrate options.
      9. Reset again and check the box "Use Maximum Render Quality"
      10. Reset and do CBR.
      11. Reset and check the Render at Max Depth.
      12. Repeat this entire processes above but for 'QuickTime'.
      13. Repeat this entire processes above but for 'Windows Media'.

       

      Does that sound like a good way to approach this.

        • 1. Re: Suggestions Wanted On Best Way To Test Encoding for the Web
          Stan Jones Adobe Community Professional & MVP

          I applaud your efforts.

           

          Work fast, before Youtube changes its methods once again.  (Tongue in cheek, but that is a factor.  Tests today may not hold for a few months from now.)

           

          I don't see enough variation in your datarates.  But that may be a good approach to learning what you can, then adding more variations for a smaller sample.

           

          You don't say, but I assume the completion of each one of the steps is to upload.

           

          Keep them all up: we look forward to viewing the tests.

          • 2. Re: Suggestions Wanted On Best Way To Test Encoding for the Web
            Dave Dugdale Level 1

            Stan, yes I plan to upload them all to Vimeo first and then perhaps less uploads to YouTube when I start getting an idea what is working and what isn't.

             

            I do plan on running tons of tests on the bitrates since that seems very important.

            • 3. Re: Suggestions Wanted On Best Way To Test Encoding for the Web
              Stan Jones Adobe Community Professional & MVP

              I believe that Vimeo and YouTube may be different enough in their encoding/format options, that I would test both equally until I have better baseline.

               

              Your best option for Vimeo may not be the best option for YouTube.

               

              I use Vimeo, but if I were testing this much, I'd want to know, for this point in time, what the best export is for each.  YouTube just reaches a lot more people.

              • 4. Re: Suggestions Wanted On Best Way To Test Encoding for the Web
                Ann Bens Adobe Community Professional & MVP

                If you look at the presets (HD) for Vimeo and YT they are exactly the same but for the framerate.

                I use the preset for YT and it works wonderfull. ( i set the framerate to 25 as i am on pal)

                Just uploading the same video to Vimeo (which i hardly use) and see how that looks.

                • 5. Re: Suggestions Wanted On Best Way To Test Encoding for the Web
                  Dave Dugdale Level 1

                  It will be my guess that Adobe has done a ton of testing already and these presets

                  are a good balance of quality vs. speed and preformance. But since I like to test everything a

                  t least once I will give it a go.

                   

                  Dave

                  LearningDSLRVideo.com

                  • 6. Re: Suggestions Wanted On Best Way To Test Encoding for the Web
                    Jim_Simon Level 8

                    I would leave out the Profile changes.  These shouldn't affect quality in the way you think.  Rather they affect the parameters you're allowed to use like frame size and bitrates.

                     

                    Also, QuickTime is more of a file format than a video format.  Many different kinds of video, including H.264, can be contained in a QuickTime file.  The compression used is more of an issue than the extension here.

                    • 7. Re: Suggestions Wanted On Best Way To Test Encoding for the Web
                      Dave Dugdale Level 1

                      Jim,

                       

                      Ah, interesting. Looks like I need to study up on what they are exactly.

                       

                      Thanks!

                      Dave

                      • 8. Re: Suggestions Wanted On Best Way To Test Encoding for the Web
                        Colin Brougham Level 6

                        I would leave out the Profile changes.  These shouldn't affect quality in the way you think.  Rather they affect the parameters you're allowed to use like frame size and bitrates.

                         

                        That's not quite correct. Changing the profile from Baseline to Main to High will (potentially) have a significant impact on the quality. Each progressive profile increase "turns on" new encoding features available in H.264, and subsequently requires more of the decoding hardware/application. With these increased features comes a greater demand on the decoder, and, generally speaking, smaller files with a higher level of visual quality.

                         

                        Level--which is what I believe you were referring to, Jim--is what specifies the more typical video parameters, like frame size, frame rate, and bitrate. These can be thought of as "constraints" or "limits" to those parameters, such that they will theoretically guarantee playback (usually on a hardware device more so than a software player) when the decoder is able to decode videos encoded to a specific level. Again, it's devices like cell phones, iPods, disc players, and so on that have a hardware decoder where level needs to be more stringently adhered to. Profile affects hardware too, of course; for example, iPod video needs to be encoded using the Baseline profile as the hardware can't handle the other encoding features provided in the Main or High profile of H.264. I've got some 1920x1080 video from a Canon 5D Mark II that is encoded as Baseline @ 5.0, as an example.

                         

                        For web playback, like with upload to a video sharing service, the level isn't going to matter so much; you just find the level that let's you set the frame size, framerate, and bitrate necessary to properly reflect your source video. The profile can be altered then to squeeze a little more encoding performance out of H.264 with the drawback of needing a progressively more platform to decode the video.

                        • 9. Re: Suggestions Wanted On Best Way To Test Encoding for the Web
                          Jim_Simon Level 8
                          (potentially) have a significant impact on the quality.

                           

                          I suppose I'm a bit cynical on this one.  For the average user, I think you'd notice far more difference changing the bitrate than the profile.  Still, I'd be curious to take a look if such testing proceeds...

                          • 10. Re: Suggestions Wanted On Best Way To Test Encoding for the Web
                            Dave Dugdale Level 1

                            I just like to test everything so I can experience it.

                             

                            I too have a feeling that bit rate will have the difference in quality.

                             

                            I just need to get all my clips together that will stress out the encoder.

                             

                            Here is my list:

                             

                            • “Low Key” clip of two people, one lite with a 100w bulb and the other with candle
                            • Green screen test – fast moving hand against gradient background
                            • Fast random motion (rapids or leaves on a tree in the wind)
                            • Text in book far away – (resolution test)
                            • Color tests (Girl standing next to flowers) also with Color bars
                            • “High Key” Back lit scene
                            • Fast moving computer graphics. Also still text with clean edges