Workflow for 24p after completing project in 60i timeline

[Kwyjibo]

Hi,

After coming back from a trip to Australia I found I had too much footage to convert to an uncompressed or low-compressed (Cineform) intermediate to store on my harddrive. I then came up with the following (free) workflow, based upon all the hard work done by others on this forum. The goal was to get the pulldown removal benefits, without all of the intermediate work and files.

1. Edit the project in Vegas on a 60i timeline, without doing any prior conversion to the m2t clips.
2. Render the movie using Debugmode Frameserver to a placeholder avi (www.debugmode.com)
3. Open the frameserved clip in avisynth, and removing the pulldown using:
DirectShowSource ("serve.avi")
TFM()
tdecimate()
Now you can use any encoder to create the final product. I use MeGui or NicWMEnc.

I tested this workflow with a short segment with 4 scene changes that also change the interlacing cadence, and the pulldown removal seems to work properly. I couldn't find any info on how TFM decides which frames need to be deinterlaced, and how it deals with cadence changes. Before I use this workflow to edit my entire vacation footage, I would like to run it by this forum to see if I missed anything important that would cause problems later on (audio timing perhaps?).

Thanks, Mike

[lordtangent]

I would be concerned about the cadence changes. Regardless of how TFM works, there is a 2 in 5 chance that your edit frames (first frame of a clip or last frame of a clip) consists of a video frame that is only one field of a "24p frame". In that case the software would have no choice but to interpolate the single field into a frame. Those "confused frames" are going to look interpolated. Since they are right at an edit some peoples eyes might not see them, but the fact remains there would be a little glitch at the first or last frame of that cut. And what if you want to do a fade, cross dessolve or title? Your aporach would only work for cuts only editing and nothing else.

For cuts ONLY editing, one option would be to just be very careful to never put an edit on one of those frames. At worst it would only force you to move you edit by one frame, which really isn't enough to mess up the creative aspect of an editing choice. It's a little bit of a hassle, but it would totally allow for the workflow you propose. You have to be able to look at your output 1:1 magnification to see a "bad" edit though.

Another option, which allows anything: dissolves, fades, wipes, titles, etc... is to frame serve the stuff INTO Premiere with the 3:2 pulldown being removed in AVISynth and simply work on a 24p timeline. I've tested it and it works. The problem is it's not quite reatime. As a result I worked out a way to do something of a hybrid approach with proxies and frame serving. The Proxies allow for realtime editing and feedback and for final render you switch to the HD res frame served stuff. It allows you to work on a 24p timeline and not worry about transcoding everything to huge intermediate files. My tests worked well, but I haven't optimized it into a systen that could be used on wholesale volumes of footage. (certain parts aren't automated yet) I still need to write a script or something to make the final AVISynth scripts for frame serving the proxies. (I considered asking SSzudzik to roll it into his tool but I still haven't decided if it's a good match. It might be better to leave it as a separate tool.) One annoyance with making the proxies is you still need about the same amount of disk space free as the HDV files take up. Even using a compact I-frame only codec like MJPEG2000 or MJPEG, the SD res video files are about the same size as the HD video in HDV format. HDV is a very efficient codec.

If enough people are interested in this frame-served workflow, I can do a tutorial. It saves tons of disk space and removes potential generation loss from the transcoding step as the final render is run straight from the the origional HDV files. It's a Premiere only workflow. ( You need an open source AVISynth frame server loader plug-in that's Premiere only)

[vincehax]

If enough people are interested in this frame-served workflow, I can do a tutorial. It saves tons of disk space and removes potential generation loss from the transcoding step as the final render is run straight from the the origional HDV files. It's a Premiere only workflow. ( You need an open source AVISynth frame server loader plug-in that's Premiere only)

I would be extremely interested in seeing a tutorial, as this seems like something I would use all the time, being an HD Premiere editor.

[Kwyjibo]

Thanks for the answer lordtangent. I should have thought of this myself. Now that you mention the transitions problem, I remember I even tested this the day I got the camera. I noticed that a black curtain wipe from a 60i timeline resulted in stuttering when converted to 24p.

I saw your post on the non-transcoding workflow for premiere, thanks for creating that thread. Vegas doesn't seem to like avisynth scripts though. What I'm currently doing is edit the project on 24p timeline, then I save the project using the Vegas option to trim each clip to only the bits that are used. Then I convert those clips, and change the source files in Vegas. It's a little tedious because you have to replace each clip manually, but it works. Scripting might automate this task, but I haven't started creating scripts yet.

[lordtangent]

You should cehck out SSzudziks Vegas plug-in. It does something like what you are doing only it does it right on the Vegas timeline by calling AVISynth externally and having it only re-encode stuff on the timeline.

[Kwyjibo]

Thanks for all your help and suggestions lordtangent. After my wife remarked that I was spending more time getting this to work than actually filming our baby, I grudgingly spent the $250 on Cineform. It's expensive, but it works well.

[zcream]

Is there anything that can be used for pulldown of MJPEG ?
I guess somehow I need to either frameserve from avisynth to Premiere before starting the project or from Premiere to avisynth after finishing the project...
Avisynth to Premiere would be better - but I'm not sure how to adapt your scripts/tutorial for MJPEG.
Thanks!

[lordtangent]

Thanks for all your help and suggestions lordtangent. After my wife remarked that I was spending more time getting this to work than actually filming our baby, I grudgingly spent the $250 on Cineform. It's expensive, but it works well.

Would you mind doing a review?

[Kwyjibo]

There already is one by Eugenia Loli-Queru at http://www.freshdv.com/2007/10/review-of-cineform-neohd-for-hv20-pulldown-removal.html
I largely agree with her conclusions, so I won't copy what she has already covered. I'd be happy to write a longer post on my experience using it with Sony Vegas, and I could do a quick test on Adobe Premiere too. I did use Morgan MJpeg too so I can contrast it with that. The thing that took me the longest to figure out is how to get a proper color workflow going. Depending on how you created the Cineform file and which application you open Cineform files with, blacks below 16 and whites above 235 may get clipped. I think it would be useful to write up my findings. I'll see if I can get something done over the long weekend.

[zcream]

My workflow is actually MJPEG to pulldown to MJPEG while the one described by Eugenia is HDV to pulldown to MJPEG. I am recording using the Blackmagic Intensity card to MJPEG and this file is what needs pulldown.
Thanks!

[zcream]

Sorry. That was meant to be a question.
I am currently saving HDMI from HV20 via BM Intensity MJPEG to a MJPEG encapsulated within avi. 24P within 60i

How do I extract the 24P stream WITHOUT using cineform? The tutorials are only for HDV capture - I haven't seen what people do for MJPEG capture..