VideoHelp Forum
+ Reply to Thread
Page 1 of 2
1 2 LastLast
Results 1 to 30 of 37
Thread
  1. This is a follow up from my earlier posts. Summary: I am my son's high school marching band's videographer. My job involves shooting my own video plus integrating random video from other parents. Therefore, I need a custom workflow that allows me to quickly convert, harmonize, ingest, edit, and encode for consumption in formats like YouTube, DVD, and perhaps BD. Many thanks to all on this forum who have helped me get this far. So I thought I would post my workflow and solicit feedback.

    1. Convert all video shot by myself or others to an MPEG2-Intra DI 1920x1080p30 m2v format using ffmpeg.
    2. For 30i assets, I will use QTGMC in avisynth with the fast preset within ffmpeg.
    3. For upsizing 1440x1080 video I will use Lancsoz4Resize in avisynth and ffmpeg.
    4. For 30p PsF assets I will use a simple weave in ffmpeg
    5. For audio I will simply demux it from the video into a wav using ffmpeg. All the audio should be 48/16 with the only complication mixing stereo and 5.1. But I can handle that in PP/AE.
    6. Bring the MPEG2-Intra DI into PP to build my timeline.
    7. Do minor color correction in AE if needed.
    8. Export final timeline from AE to an MPEG2-Intra DI
    9. Encode that DI to whatever using ffmpeg.

    I am posting in this sub-forum because so much of the workflow involves conversion, by far the most complicated steps in my workflow. So free to comment on what if anything looks good/bad. Thanks everyone.
    Quote Quote  
  2. Originally Posted by SameSelf View Post
    so much of the workflow involves conversion, by far the most complicated steps in my workflow.
    Your sources can all be read by Premiere, so these conversions are unnecessary. Make a 1080i/60 project. Master to Cineform or DNxHD 1080i/60. Use the master to create your Blu Ray (interlaced), DVD (interlaced) and web (progressive) conversions. Your YouTube-type conversions can be run through QTGMC if you insist, but yadif will be faster and ultimately just as effective.
    Quote Quote  
  3. Three reasons that I want to convert my video:

    1. I have a bias against interlaced material. IOW I want to have complete control over deinterlacing.
    2. I want to work with a DI, like others have suggested in this forum
    3. I do not want to edit AVCHD video, period
    Quote Quote  
  4. You must be billing by the hour.
    Quote Quote  
  5. Everyone has their own way of doing things, but I would do it like smrpix suggested, and remember to archive the originals

    The problem that workflow is unecessary processing . I wouldn't use QTGMC on all interlaced footage early on , simply because lots of it will probably be cut out anyway. (And you're not going to tell me that some "random parent's" footage won't have cameras pointing to the sky, ground, out of focus etc... you 're not going to include that I hope).
    Quote Quote  
  6. All good comments. I don't have a lot of experience doing this just yet, so the learning curve is quite steep. I am confident after a while I will be searching for more efficient methods. It's just right now my computer chokes on AVCHD footage so I gotta convert it. And if I assemble a timeline from various sources in PP/AE how do export that mishmash out for deinterlacing in QTGMC (just those portions that need it)? Unless you are telling me Adobe's deinterlacer is better?
    Quote Quote  
  7. Adobe's deinterlacer sucks

    Converting to something to edit is fine, but deinterlacing for days when you're not even going to use it doesn't make any sense (think of the trees, global warming ) . QTGMC on "fast" is definitly faster than "slower", but it still pretty slow on HD footage... At least do rough cuts beforehand if you decide to go that route. You don't even need a frame accurate/smart editor. Even GOP accurate cutter is fine, because they are only rough cuts

    You need to decide on what overriding sequence settings / comp settings to use. There are pros/cons to either choice because of the mixed footage. Probably how much mixed footage, or what makes up the majority of your footage strongly influences that decision. If you're going to use 1080p59.94, then you're going to have hardcoded duplicates on the 29.97 footage . 1080p59.94 will be accepted for YT shortly, so that might be one reason to go that way. It also give you the most options (you can interlace it easily for BD, scale it easily for SD DVD , or progressive Web distribution) .

    The other option is to use an interlaced timeline (it's called 1080i29.97 or 1080i59.94, but those are the same thing, just different naming conventions), and use QTGMC if you need various progressive exports after exporting out of adobe. If you decide to go that route - QTGMC is supposed to be adaptive - ie. it's supposed to weave sections that don't need deinterlacing, but it's not perfect. You can combine it with a comb mask and motion mask (TDeint + TMM) to be even more selective
    Quote Quote  
  8. You export your interlaced 1080i master to Cineform or DNxHD and run that through QTGMC. The progressive material will remain progressive, the interlaced material will become progressive, and you are still processing significantly less material overall.

    Also, while QTGMC is great, Adobe's internal deinterlacer isn't bad either. (Duplicate your interlaced Sequence and change the settings to progressive, then export progressive for best results.)

    edit: a bit of cross-posting and some honest difference of subjective judgement.
    Quote Quote  
  9. A lot of great experience and suggestions. This stuff is incredibly complex, but having you guys to collaborate with gives me hope
    Quote Quote  
  10. Originally Posted by smrpix View Post
    You must be billing by the hour.
    LOL all my work is volunteer. Right now if it weren't for me, no one would be making video. Last year his band director brought in a professional media company to record the Christmas concert. I thought, this is awesome and snapped up a recording which was a little high priced imo. When I finally got it about a month later I was shocked at how much I spent for an SD DVD. It was at this point I said, not again. I mean, my smartphone records in HD, what is this? The nineties? This is what I have to remember those magical moments?

    Anyhow rant over. So I am back because I finished up a video last night. It wasn't painless and I wanted to get some feedback.

    So I was working primarily with 1440x1080i30 AVCHD shot by a Sony HDR-UX7. This is actually the band's camcorder. I used ffmpeg and QTGMC to deinterlace and upsize to 1920x1080p30 MPEG2-Intra only codec. I brought this into PP to edit and it went OK. Working with separate audio and video tracks wasn't intuitive at first but I worked out some tricks to speed things up. The real problem was sometimes PP would just choke on the timeline. I thought my system was powerful enough. But maybe not.

    I exported for DVD authoring in Encore because we will be showing the video tonight on a projector that I know is not HD. Good thing too. Because I tried to export a lossless AVI from AE. My 20 minute video was 200 GB! I haven't installed any other codecs like DNxHD. My deadline for tonight has kept me from experimenting too much.

    Also, I had some 1440x1080 30 Psf HDV footage from last season that I wanted to include in the video. However, I could not get avisynth to read the file to preform a Lancsoz4Resize. All I was able to do was get rid of the PsF using a simple weave in ffmpeg. Fortunately PP/AE didn't seem to care that I was mixing resolution. But I would like to know what the problem was.

    Again thanks for all the help guys. I couldn't have gotten this far without your help. And I am really excited to show the DVD tonight.
    Quote Quote  
  11. Originally Posted by SameSelf View Post
    So I was working primarily with 1440x1080i30 AVCHD shot by a Sony HDR-UX7. This is actually the band's camcorder. I used ffmpeg and QTGMC to deinterlace and upsize to 1920x1080p30 MPEG2-Intra only codec. I brought this into PP to edit and it went OK. Working with separate audio and video tracks wasn't intuitive at first but I worked out some tricks to speed things up. The real problem was sometimes PP would just choke on the timeline. I thought my system was powerful enough. But maybe not.
    Exactly what do you mean by "choke" ? Do you have red render bar on top of the timeline ?

    Although you can, I personally wouldn't work with elementary video , audio and try to link up. There is higher chance of desync and things going wrong. It's happened to me before and many others - but YMMV .

    If your specs are listed correctly, it should not be choking on a single MPEG2 intra stream without any effects/filters. But a smoother/faster editing intermediate is cineform - but you cannot batch with ffmpeg, but you can queue them up in vdub

    ATI cards don't work as well with PP.

    If that was the primary camera - you're throwing away 1/2 the information by using 30p. Motion will be more jerky . I guess if they are all static shots it won't matter too much, but since this is a "marching" band , that implies they are marching. The difference will be visible on your DVD. I would argue the motion difference is more noticable to your average viewer than the difference between using a simple bob and QTGMC




    Also, I had some 1440x1080 30 Psf HDV footage from last season that I wanted to include in the video. However, I could not get avisynth to read the file to preform a Lancsoz4Resize. All I was able to do was get rid of the PsF using a simple weave in ffmpeg. Fortunately PP/AE didn't seem to care that I was mixing resolution. But I would like to know what the problem was.
    I would index with DGIndex , because it's the most reliable for MPEG2. If you setup directshow correctly , you should be able open with DirectShowSource() but it's less reliable. Other options are FFMS2 and L-Smash
    Quote Quote  
  12. Originally Posted by poisondeathray View Post
    Exactly what do you mean by "choke" ? Do you have red render bar on top of the timeline ?
    Poisondeathray you are a guru. Yes I had the red bar. I was working fast and just didn't have time to troubleshoot. But the strange thing is for one specific clip, I would try to watch it in the source monitor and the timecode would be moving normally but the video would be stuck on one frame. When I pulled it down into my timeline to watch in the preview monitor, same problem and just on that clip in the timeline. All the other video was fine. Very strange.

    I definitely hear you on the sync problems because I had some. I won't be doing this a second time if I can avoid it.

    I guess I need to sell my ATI card and get Nvidia

    Thanks for the DGIndex tip, will def take a look.

    I will start looking into the other codecs that have been suggested now that I made one deadline. I am thinking strongly about either ProRes or DNxHD as I am thinking about buying an Atomos Ninja for my Canon HV40.
    Quote Quote  
  13. QTGMC is a must for SD resolution but not really for HD resolution, try yadif, double frame rate, and you might not see any difference.

    your HV40 might become obsolete soon, I would re-consider that hdmi capture device purchase, mpeg2 is quite manageable by NLE, H.264 needs more power. ReStream changes video to true progressive (check frame type progressive and progressive sequence, uncheck top field first) or there are scripts to do that for Sony Vegas that interpret footage to 30p (not sure about Premiere).
    Quote Quote  
  14. I'll say this again then stop, but I really think you're shooting yourself in the foot on ingest. Premiere should handle HDV material natively like butter unless you get into very long timelines. Even then, building shorter "scenes" then assembling it all into a larger "master" sequence can easily get you past that.
    Quote Quote  
  15. Originally Posted by _Al_ View Post
    QTGMC is a must for SD resolution but not really for HD resolution, try yadif, double frame rate, and you might not see any difference.

    your HV40 might become obsolete soon, I would re-consider that hdmi capture device purchase, mpeg2 is quite manageable by NLE, H.264 needs more power. ReStream changes video to true progressive (check frame type progressive and progressive sequence, uncheck top field first) or there are scripts to do that for Sony Vegas that interpret footage to 30p (not sure about Premiere).
    I have done some comparisons of QTGMC vs yadif on HD footage and the difference is noticable, less blurring using QTGMC. See this thread: https://forum.videohelp.com/threads/366455-Deinterlacing. For me, the question is what setting? Slow, Fast, other niggly settings? I get nice results IMO using Fast and nothing more. I have this thread to thank for alerting me to QTGMC http://casparcg.com/forum/viewtopic.php?t=1113. Very infomative.

    As for my HV40 becoming obsolete soon? Why? Please explain. Currently I shoot in 1440x1080 anamorphic native 30p in a 4:2:0 8-bit colorspace which I then resize to FHD into a ProRes HQ codec for editing. So far so good. PP/AE handle the codec swimmingly. That feels pretty modern to me. If I get an Atomos Ninja, then I can capture the native progressive footage directly into a ProRes HQ codec and avoid the in-cam compression that throws away information. I see that as prolonging the life of my HV40. Because what other camcorder can give me similar results? Perhaps a $1500 to $2000 brand new cam that uses a built in intermediate codec? But remember, I am a parent volunteer, not a paid media pro. But if you have some suggestions/advice I am interested.
    Quote Quote  
  16. Originally Posted by smrpix View Post
    I'll say this again then stop, but I really think you're shooting yourself in the foot on ingest. Premiere should handle HDV material natively like butter unless you get into very long timelines. Even then, building shorter "scenes" then assembling it all into a larger "master" sequence can easily get you past that.
    Agreed. I like editing with HDV. Never once had a hiccup. But I guess ignorance is bliss. Many on this forum pushed me in the direction of an intermediate codec which I think is the right strategy because I like to do color correcting in AE sometimes. Plus I like uprezing my HDV assets to FHD. I just didn't know what codec to use. I tried MPEG-2 Intra first. It didn't work well for me, some strange random hiccups. But the problems I had using MPEG-2 Intra went away when I went to ProRes HQ. I haven't tried DNxHD yet like some have suggested. Maybe soon. But I also like using ffmbc.

    One thing I have learned is there is definitely more than one way to skin this cat. But the tools and the techniques I have landed on so far are:

    1. QTGMC in avisynth
    2. ProRes HQ codec using ffmbc
    Quote Quote  
  17. Originally Posted by SameSelf View Post
    As for my HV40 becoming obsolete soon? Why? Please explain.
    -I had transferred hours and hours of HV30 30p footage, it takes time, I'm getting tired to do that all the time.
    -Then I need to automatically transfer that 30f to 30p (I cannot store 30f for the future to give myself or others headache).
    -Stabilization on HV is very bad to nowadays standards, picture is shaking, you have to really keep it steady.
    -You have to have wide angle at the end otherwise it is almost impossible to shoot (Always there is a need to step at least 20 steps back, and likely fall off the cliff or to step through the wall or be run over by car or something ). Wide angle tends to blur picture if you are not careful . Canon wide angle seems to be sharp all the way but does not have a thread at the end for ND filter.
    -You need ND filter to have on with 30p mode shooting outside, specially on sunny day to keep shutter speed longer so video does not have strobe pulse feel. If you do not have ND filter, footage is garbage. I got Raynox 0.66X wide angle , it is great, but it is not sharp at the end, so I cannot zoom in further than half focal length with that wide angle, I have to take it down, then I do not have ND filter for shot, but at least not that light is coming in with that focal length.
    -60p is is way to go. But, you are focused to deliver on YouTube or something, so you do not seem to mind 30p.
    -Low light , as it was superb 10 years ago, nowadays nothing special, colorful noise if you are not careful and you have to permanently take ND filter down or put it back on.

    +editing is great though, and smart render is very good bonus, this is only what keeps me using that HV30, also manual mode shooting capabilities ...
    Last edited by _Al_; 30th Aug 2014 at 16:17.
    Quote Quote  
  18. Well, the thing is, I am filming a marching band at night, under the lights of a stadium where I need precise footage, not some blurred 24p film aesthetic under a blazing sun. So I fear we are talking past one another.

    Personally, I feel like I am quite lucky to even have an HV40 with the Canon wide angle lens that shoots in HD. ND filters, what are those? I don't even know what you are talking about or how those would help me in my situation. I am a parent volunteer who enjoys filming his son's band, not an aspiring indie film producer.
    Quote Quote  
  19. ND filter (neutral density filter) is a filter that takes away light coming into lens, if you need to record longer shutter speed. You just screw it on, on the top of lens. Longer shutter speeds like 1/30 or 1/60 would bring so much light into lens that video would be over bright. This goes for manual mode shooting. If you use automatic mode, camcorder chooses appropriate shutter speed automatically that corresponds lighting conditions, but under bright light (even just daylight) the shutter speed would be so short that it causes that strobe pulse (not fluent) in video recording 30p (not 30i where shutter speed could be shorter). Or aperture is almost closed (F8 or so) that is not optimal also. You need to have it lower to record nice picture.

    You do not need ND filter for low light conditions, night, evening shots. In a bright daylight shooting 30p in auto mode on the contrary you might start to wonder why the video is strobing, why it is not fluent, so just heads up. Shooting 30i is always fluent, even using very short shutter speeds, so shooting 30i (or 60p but not with this camcorder) gets rid of this problem.
    Quote Quote  
  20. I personally can't stand QTGMC, it's not a deinterlacer per se it's more a deinterlacing script and does way too much processing for my tastes. I prefer TDeint.

    The workflow the OP posted doesn't make any sense to me, why upsize 1440x1080? You don't know for sure that you will be creating BluRays and even if you were 1440x1080 is a valid BD resolution.


    Why Lancsoz4Resize? I much prefer Point Resize or Nearest Neighbor.

    I would use the camera's native format, import that into PP, perform my edits and color correction or whatever else I wanted to do, then export the DVD compliant mpeg-2 sans any deinterlacing, since interlaced mpeg-2 is perfectly valid for DVD and I would probably also do the same for Blu Ray, export right from the NLE sans deinterlacing since again interlaced footage is perfectly fine for BD as is 1440x1080.

    You Tube I would not really care all that much about because they just reencode whatever you give them anyway and by the time they are done with your source it will probably look lousy anyway.

    To me the OP is making way to much work for himself and more importantly the quality of the final product will suffer with all that processing in the form of deinterlacing, then resizing and converting with ffmpeg, then then importing that to AE/PP, editing the footage and color correcting it, then exporting back out again back to I-frame mpeg-2, then taking that and encoding it to the "final" delivery format via ffmpeg.

    I count at least 3 generations removed from the acquisition format and that is definitely not a recipe for maximum quality.
    Quote Quote  
  21. Originally Posted by sophisticles View Post
    To me the OP is making way to much work for himself and more importantly the quality of the final product will suffer
    I agree. But he's very determined.
    Quote Quote  
  22. Thanks for the comments everyone. Let me try to respond to some of the questions:

    Q: Why upsize to 1920x1080?
    A: 1. It's fast and easy. 2. Youtube wants square pixels. So what else do you suggest I do with anamorphic 1440x1080 video? I am truly puzzled here.

    Q: Why Lanczos4Resize?
    A: Simply because the wiki suggests it is the best for upsizing: http://avisynth.nl/index.php/Resize. No one has suggested any credible alternative. PointResize/Nearest Neighbor? Well, the wiki says, No!

    Q: Why QTGMC?
    A: Well, I only have heard of yadif and QTGMC up until this point. Well, I lie. I have heard of TGMC but I assumed QTGMC was a more modern incarnation. Anyway, yadif and QTGMC are the only comparisons I have ran on the footage that I am working with. See here: https://forum.videohelp.com/threads/366455-Deinterlacing. I might try to compare my results to TDeint if I get some time. But, please, bear in mind, as some have already pointed out:
    Originally Posted by smrpix View Post
    Originally Posted by sophisticles View Post
    To me the OP is making way to much work for himself and more importantly the quality of the final product will suffer
    I agree. But he's very determined.
    So, would you mind elaborating why you prefer TDeint? I don't really understand what you mean by "[QTGMC is] more a deinterlacing script...". What exactly is the benefit of TDeint? Faster? Better? In my day job, they say you can only choose two of the three: Fast, Quality, Cheap. Never all three. And I have all already chosen one. Cheap! Or, should I say Free? My next choice is Quality. So I don't necessarily expect Fast.

    Q: Why do I de-interlace?
    A: Because I don't trust PP/AE to do it for me. I want all my footage in a consistent format before I start building timelines. Maybe I am being determined, but I am trying to avoid problems and develop "best practices".

    Q: Why don't I use the camera's native format?
    A: I can't. For some reason, my computer chokes on AVCHD footage. And I have zero energy trying to get my computer/software to the point where it edits AVCHD footage like a hot knife through butter. Now, my initial thought was to convert all my AVCHD footage to HDV. But this thread that I started pointed me in a different direction: https://forum.videohelp.com/threads/366357-Converting-AVCHD-to-HDV. So, if I seem a little determined, I only have this forum to thank

    FWIW, I am already on my third project since I originally posted my workflow. So my determination arises simply because this has become a part time job already. I have already made some modifications. So I better post an update:

    1. Convert all video to 1920x1080p30 ProRes codec standard profile using ffmbc (I tried HQ and that seemed to offer no benefit)
    2. For 30i assets, use QTGMC in avisynth with the fast preset within ffmbc
    3. For upsizing 1440x1080 video, use Lancsoz4Resize in avisynth and ffmbc
    4. For audio demux it from the video into a wav using ffmpeg. All the audio should be 48/16 with the only complication mixing stereo and 5.1 surround. But I can handle that in PP/AE.
    5. Bring the video and audio into PP to build my timeline and add titles/rolling credits
    6. Do color correction in AE if needed
    7. Export final timeline from PP to a lossless format
    8. Encode as needed. Using Handbrake right now for Youtube encodings. No one has asked for Blu-rays or DVDs.

    So you can see, the big changes are 1) moved away from the MPEG-2 Intra format to ProRes standard profile. My reasons documented above. 2) Not using ffmpeg on the back end.
    Quote Quote  
  23. Member Cornucopia's Avatar
    Join Date
    Oct 2001
    Location
    Deep in the Heart of Texas
    Search PM
    Ask yourself: would the pros do that? and if so why? and if not, why not?

    1. Combine ALL footage to consistent Highest Denominator (60p), not Lowest (30p). Do NOT deinterlace at all until you HAVE TO.
    2. Use Lossless or near lossless for ALL intermediate work. Prores is NOT always the best choice on a PC, though workable.
    3. Lancsoz algorithms are great, but they also sometimes add ringing/coring/halos, which is a bitch for compression. Sometimes it's better to be soft-but-clean. Use appropriately on a per clip basis.
    4. You seem OK for audio. But those apps aren't GREAT for audio. If doing multi-track audio mixing/processing, use a proper DAW.
    5. Rest of your workflow is pretty good.
    6. Keep separate steps (Logging+Ingest, Edit, Composite/Grading/Mastering, Processing + distribution Compression), so you can easily step back to a previous intermediate, etc. if something doesn't work quite right.

    Scott
    Quote Quote  
  24. Questions:

    1. How do I convert 30p/30i footage to 60p? Is there something I am missing? That is something I haven't even attempted yet. Also, I don't have any cams that shoot native footage in 60p.
    2. Regarding intermediate codecs, I tried MPEG-2 Intra and my PC choked on some video. So I moved on to ProRes and it works flawlessly thus far. I am open to trying DNxHD but I don't know how to encode my AVCHD/HDV vid to DNxHD. Any suggestions? ffmpeg? ffmbc?
    3. Hmm, not sure if I have noticed any ringing/coring/halos yet. But I will keep my eye out. But bottomline, what should I do with my anamorphic video? Youtube wants square pixels. Is there a "best practice" that someone could recommend other than Lancsoz4Resize to FHD?
    4. Regarding audio, thanks for the tip. I use Cubase, shoot the audio separate from the video, (audio from a press box stinks), and mix it with some compression, denoising, etc. before bringing into PP. Audio is the least of my worries. In fact, the audio consumes much more of my time than the video. I don't even shoot the video, I post my younger son in the press box to push the record button while I am down close to the field monitoring the audio levels closely with my audio recorder.
    5. Thanks!
    6. Still learning!
    Quote Quote  
  25. Formerly 'vaporeon800' Brad's Avatar
    Join Date
    Apr 2001
    Location
    Vancouver, Canada
    Search PM
    QTGMC is a bob deinterlacer that outputs 60p if your source is 30i. The SelectEven (or SelectOdd) filter is what drops it to 30p. If you have native 30p footage, it's usually converted to 60p simply by showing every frame twice. Any NLE should handle that the same way, unless you tell it to invent in-between frames.

    Does upsizing the video from 1440 yourself actually look better than uploading it to YouTube as-is?
    Quote Quote  
  26. Originally Posted by SameSelf View Post

    1. How do I convert 30p/30i footage to 60p? Is there something I am missing? That is something I haven't even attempted yet. Also, I don't have any cams that shoot native footage in 60p.


    When you deinterlace with QTGMC, it's "bob" deinterlacing or "double rate deinterlacing" . Essentially each field becomes a frame . That's why motion is so smooth for that "live look". Yadif in mode=1 does the same thing (in terms of frame rate) . Yadif in mode=0 does single rate deinterlacing. That's why I was making such a big deal out of throwing out 1/2 the data, choppy motion

    You don't use deinterlacer on 30p footage


    2. Regarding intermediate codecs, I tried MPEG-2 Intra and my PC choked on some video. So I moved on to ProRes and it works flawlessly thus far. I am open to trying DNxHD but I don't know how to encode my AVCHD/HDV vid to DNxHD. Any suggestions? ffmpeg? ffmbc?

    I wouldn't bother with DNxHD, but you use ffmbc as well. There are 8bit and 10bit varities

    The 8bit variety is called with -pix_fmt yuv422p , 10bit called with -pix_fmt yuv422p10le

    Code:
    [dnxhd @ 022151e0] available bitrates in Mb/s for 10bits:
    [dnxhd @ 022151e0] 1080p24: 175
    [dnxhd @ 022151e0] 1080p25: 185
    [dnxhd @ 022151e0] 1080p30: 220
    [dnxhd @ 022151e0] 1080p50: 365
    [dnxhd @ 022151e0] 1080p60: 440
    [dnxhd @ 022151e0] 1080i25: 185
    [dnxhd @ 022151e0] 1080i30: 220
    [dnxhd @ 022151e0] 720p24: 90
    [dnxhd @ 022151e0] 720p25: 90
    [dnxhd @ 022151e0] 720p30: 110
    [dnxhd @ 022151e0] 720p50: 185
    [dnxhd @ 022151e0] 720p60: 220
    
    [dnxhd @ 022151e0] available bitrates in Mb/s for 8bits:
    [dnxhd @ 022151e0] 1080p24: 36, 115, 175
    [dnxhd @ 022151e0] 1080p25: 36, 120, 185
    [dnxhd @ 022151e0] 1080p30: 45, 145, 220
    [dnxhd @ 022151e0] 1080p50: 75, 240, 365
    [dnxhd @ 022151e0] 1080p60: 90, 290, 440
    [dnxhd @ 022151e0] 1080i25: 120, 185
    [dnxhd @ 022151e0] 1080i30: 145, 220
    [dnxhd @ 022151e0] 720p24: 60, 90
    [dnxhd @ 022151e0] 720p25: 60, 90
    [dnxhd @ 022151e0] 720p30: 75, 110
    [dnxhd @ 022151e0] 720p50: 120, 185
    [dnxhd @ 022151e0] 720p60: 145, 220
    To make the selection, you use -vcodec dnxhd and for bitrate, -b and the number in M. eg. 1080p60, 10bit would be -bitrate 440M . Notice the 8bit variety has "tiers" of bitrate selections, or selectable quality levels



    3. Hmm, not sure if I have noticed any ringing/coring/halos yet. But I will keep my eye out. But bottomline, what should I do with my anamorphic video? Youtube wants square pixels. Is there a "best practice" that someone could recommend other than Lancsoz4Resize to FHD?
    There is no "best practice", it depends on situation, condition of the footage. That's why there are different algorithms available you choose what works for your goal in that situation

    Recall I said QTGMC uses strong sharpening and that I almost always decrease the value? Combining default with lanczos4 will almost always cause halos, unless your footage is soft to begin with
    Quote Quote  
  27. Ah, so deleting the SelectEven() just turns each field into a frame and 30i material becomes 60p. But is it true that using SelectEven() on 30i video is the equivalent of throwing out data? I thought the resulting 30p frame was an interpolation of sorts between two time separated 30i fields. Maybe it's why I haven't seen any problems in my deinterlacing, because I am loathe to "create" information by turning my 30i fields into 60p frames.

    As for 30p video, I get it that it doesn't need to be deinterlaced. It just sounded like someone was suggesting I turn it into 60p footage. Which I guess would be done by doubling the frame rate? No additional information, just a format specification.

    Also, I have yet to notice the halo effect. But my footage must be very soft as you stated, or just of such a nature that I magically avoid it. Trust me, I wouldn't be doing any of this if the results were bad.

    poisondeathray: where did you get that code snippet for ffmbc? You seem to have access to the magical decoder ring of the murky black box that is ffmbc. As always, thanks for your guruness!
    Quote Quote  
  28. I don't see the need to upsize because YouTube will just re-encode anyway and in doing so will up and down size for you.

    I don't see the need to de-interlace because of your 3 delivery formats, DVD, BluRay and Youtube, the first two support interlaced content just fine and the last re-encodes everything anyway and will de-interlace it for you.

    QTGMC does significant processing on the source, it sharpens, it denoises, it puts your source through the ringer. Some people may like what it does, I don't, I prefer to use a straight de-interlacer and handle any sharpening and de-noising myself.
    Quote Quote  
  29. Originally Posted by sophisticles View Post
    I don't see the need to de-interlace because of ....... DVD
    But it looks like, to go from HD interlace to SD interlace one must double frame rate de-interlace before resizing to get footage to be up to par with some standards (not flickering on screen). Just leave it to NLE does not give expected result somehow.
    Quote Quote  
  30. Originally Posted by SameSelf View Post
    Ah, so deleting the SelectEven() just turns each field into a frame and 30i material becomes 60p. But is it true that using SelectEven() on 30i video is the equivalent of throwing out data? I thought the resulting 30p frame was an interpolation of sorts between two time separated 30i fields. Maybe it's why I haven't seen any problems in my deinterlacing, because I am loathe to "create" information by turning my 30i fields into 60p frames.

    You're not creating anything, except interpolating some spatial information (filling in the missing lines) per field, when using a smart deinterlacer

    "30i" or really "29.97i" (also called 59.94i , same thing) is 59.94 FIELDS per second. Each fields represents a moment in time, distinct and unqiue . When you watch sports on TV, an interlaced DVD on a flat panel - it gets bob deinterlaced to display 59.94 moments in time per second. That's why interlaced provides the "live" look , motion is smooth. When you use SelectEven or SelectOdd after bob deinterlacing, it becomes 29.97p, instead of 59.94p. So yes you throw away 1/2 the information

    As for 30p video, I get it that it doesn't need to be deinterlaced. It just sounded like someone was suggesting I turn it into 60p footage. Which I guess would be done by doubling the frame rate? No additional information, just a format specification.
    You duplicate frames, so it fits in a 59.94 sequence . (Unless you produce VFR video, everything has to have the same frame rate)

    There are interpolation techniques to generate new frames with optical flow, but that's not what we are talking about




    poisondeathray: where did you get that code snippet for ffmbc? You seem to have access to the magical decoder ring of the murky black box that is ffmbc. As always, thanks for your guruness!
    It was posted on a message board (I think the ffmbc board), I have it copy/pasted somewhere
    Quote Quote  
Visit our sponsor! Try DVDFab and backup Blu-rays!