I have been searching for a step-by-step guide to performing 60i-to-60p-to-24p 40% slow-motion. Barry's DVX book and numerous other posts and articles talk about this, but don't really tell you how to do it. The www.100fps.com finally gave me the guidance I needed to figure the process out, but the page is a bit cluttered so I wanted to write my own step-by-step guide. Here we go!

Abstract:
This method will take 60fps interlaced footage and turn it into 60fps progressive footage, turning fields into frames in the process, for use in a 24fps timeline to achieve fluid 40% slow-motion.

Theory:
The DVX allows for recording video at 60 fields per second (called 60i or 60fps interlaced). What this means is that it actually captures 60 fields per second to tape, but each "field" consists of half a frame. Each field is missing every other horizontal line of resolution. We can however "fill-in" the missing information by taking "guesses" at to what should go in these lines by examining the lines of resolution that do exist, and this creates 60 full-frames per second. There are many methods of guessing how to fill in the missing lines of resolution - some better than others - but all will result in some loss of resolution from true 30fps progressive recording (which captures only at 30fps, but captures a full frame, not fields). So we lose a little image quality, but we get video that is 60 frames per second, which is full-frame double-speed video. We can then slow down this footage to get perfectly smooth 40% slow-motion in a 24fps video (40% of original speed - so this is slighly slower than half-speed!). No jerky, stroby slow-motion video - if done right it looks great!!

Tools needed:
Virtualdub - www.virtualdub.org
Avisynth - www.avisynth.org
DV Codec - many available - do a google search. Here is a link for the panasonic DV codec - http://www.free-codecs.com/Panasonic_DV_Codec_download.htm

Instructions

(note, there are some sub-bullets, like "5a", as intermediate steps to verify that you have done all the previous steps correctly)

1. Shoot your footage in 60i (interlaced) mode on your camera.

2. Capture the video to your computer using your normal method. Additionally virtualdub can capture video, but I haven't tested it (I used Adobe Premiere to capture). I named the captured file "60interlaced.avi"

3. Install all software - virtualdub, avisynth, DV codec. The DV codec can be usually installed by right-clicking on an ".inf" included with the codec and selecting "Install". Sometimes you have to double click on a ".reg" file, and then copy the DLL file to c:\windows\system32

4. In the directory that you saved your captured file, create a text file called "script.avs". In that file place the following lines:

AVISource("60interlaced.avi")
separatefields

5. Open virtualdub. Go to the File menu, then select "Open video file". Open the "script.avs" file created earlier. You should see your video in the virtualdub window. There are two video windows in virtualdub - the left is the source video, and the right is the output video.

5a. To verify that you have a 60fps file, go to the File menu and select "File Information". The framerate should be 60fps (or 59.xxx). And the number of frames should be double the number of frames if you just opened the AVI directly.

6. Go to the Video menu and select "Filters". In the dialog that appears, click "Add". Then click on "Field Bob", and OK. Then the "field bob" options dialog appears. You can experiment with these options, and depending on the video material, choose the options that appear the best for you - if you aren't sure, just use "Normal". Click OK to get back to the filters dialog.

7. Again click on "Add" to add another video filter. Select "Resize" and then click OK. Then the "resize" options dialog appears. In the new width field enter "720", and in the new height field enter "576". Then click OK twice to get back to the main virtualdub window.

7a. The video should now be properly sized, and in the output window of virtualdub (the video on the right) the image should have interlacing artifacts removed (no black lines)

8. Go to the File menu and select "Save as AVI". Virtualdub will now create a new AVI file that is 60fps progressive.

8a. Open the new file in a video player - it should now place twice as fast!

9. Drop this new 60p file into your NLE program on a 24fps timeline. I do this in Adobe Premiere. After dropping it in the timeline, right click the video clip and select "Speed/Duration". In the dialog, set "Speed" to 40%. I also had to right-click the clip and select "Field Options", and then uncheck "Frame Blend Speed Changes" for the final movie to appear smooth.

10. Export your movie from the NLE, and run it in your media player. You should have SMOOTH slow motion - no jerky, stroby motion - but it will be at a lower resolution, but it may not be noticable unless it is side-by-side with non-interlaced footage.

10a. This can also be done in a 30fps timeline. Just change your speed to 50% instead of 40% in your NLE.


There are a TON of variations to this process. Probably a lot of them will produce slightly-better results. Experiment! See what you come up with. Post suggestions here! Please only on-topic to this method.

Comments? Are screenshots necessary? Something need better explanation?

'Just a couple of quick questions:

In step 7, why resize to 720x576? I would have thought 720x480. Some NLE's might stretch it to the proper size once its imported anyway, so it may not even be necessary to resize it at all (though the resizing algorithm used may vary, so it's likely that vdub's filter is better after all).

Step 9... rather than changing the speed to 40% in the NLE, what would happen if you changed it to 24fps in vdub (see step 5a)? I believe it allows you to simply change the rate without affecting the frames at all, so this might help with some NLE's.

Step 8... what compression codec do you usually use for the output from vdub? I believe it defaults to uncompressed. This is great for quality, but it can really eat up hard drive space. As an alternative, you might be able to output to DV (if you set vdub to 720x480x29.97fps) and avoid rendering in the NLE.

That said, great job on the tutorial! I've worked with AVISynth a couple times, and I know it's an awesome tool to use once you get comfortable with writing code. Mostly I'm just curious about these little details.

As an aside, the current beta version of AVISynth (v2.5.6 I think) has a new color adjustment command (example: separate gamma controls for R,G,B), but that's a topic for a new thread.

Jeremy

In step 7, why resize to 720x576? I would have thought 720x480. Some NLE's might stretch it to the proper size once its imported anyway, so it may not even be necessary to resize it at all (though the resizing algorithm used may vary, so it's likely that vdub's filter is better after all).

Your right. 720x480 should be the size for NTSC. I believe 720x576 is for PAL.

Step 9... rather than changing the speed to 40% in the NLE, what would happen if you changed it to 24fps in vdub (see step 5a)? I believe it allows you to simply change the rate without affecting the frames at all, so this might help with some NLE's.

After reviewing this, I probably should modify it so that this uses all free tools, and since VirtualDub can output at 24fps, that should basically be the final step.

Step 8... what compression codec do you usually use for the output from vdub? I believe it defaults to uncompressed. This is great for quality, but it can really eat up hard drive space. As an alternative, you might be able to output to DV (if you set vdub to 720x480x29.97fps) and avoid rendering in the NLE.

Good point. I'd assume that clip that you wanted to slow down would be relatively short (not a feature length film), so it should only be a few gigs, but if you are short on space you can go this route. I did it uncompressed, but other users can definitely output to the DV codec!


Thanks for the comments!

Wow Mark, this is really useful. I'm going to try this tomorrow. Thanks so much for the post.

I've been looking for a how-to like this for a while... Thanks!

But I'm having a little trouble I'm hoping someone knows how to fix. I'd never used VirtualDub before this and I'm getting the following error when trying to open the script file: "AVISource: couldn't locate a decompressor for fourcc dvsd"

Any ideas?

How to use a DV file in avisynth (http://forum.doom9.org/showthread.php?t=82242&highlight=fourcc+dvsd)

Justin,

You need to install a DV codec. I used Panasonic's which is available here:

http://www.free-codecs.com/Panasonic_DV_Codec_download.htm

or you could do a google search or a search on this forum for "DV codec". There are a few other implementations out there besides the panasonic, and everyone's got an opinion on their favorite! Good luck!

Ah, Ok. I thought as much... I'd downloaded the codec (the Panasonic one), but didn't know what to do with it (if I right-click the dll, there's no option to install). I was hoping it just had to be resident on the computer. Ok, so now I need to figure out how to install it.

Justin, I believe the way to install the codec is this:

1. Place the pdvcodec.dll in your windows/system32 directory
2. Then right-click on the INF file and select "install"

That should do it. The "install" part only installs the registry keys I believe, you still have to manually copy the file into your system32 directory.

Eureka! Thank you very much!

This is great and everything, but can anyone translate this so it can be done on a mac? Forinstance, can anyone mimic this in say... after effects or anything like that? I trust that all this works, but i can't verywell try it if i cant download any of the software.

Any suggestions?

what other commands can you give in the text tile I bet there alot of possibilities

You know, I just get an AVI import error. Never could make this work.

BTW, don't the major NLEs (Premiere, Vegas, FCP) offer deinterlacing functions? You would expect..

Okay, if you use separatefields, that'll make the picture have it's vertical resolution cut if half right? From x480 to x240 (using NTSC) So if you upscale it back to 480, how will you keep the image from becoming pixilated (jagged lines), unless there's something around that, that I'm not thinking about or reading correctly?



Jake Zalutsky: Avisynth is capable of many things, a lot of which are usefull for pre and post processing. I have also seen entire videos edited in AVIsynth (the editor did it for the challange, I personally would not recommend it without heavy medication :) ), so it's possible, but very tedious.

Hey Marc, thanks for all the effort to simplify this process, do you happen to have any end result footage?

~Shawn

You know, I just get an AVI import error. Never could make this work.


DJ Kinney, you need to install the Panasonic DV codec... See the previous replies for (slightly more) detailed instructions...

Okay, if you use separatefields, that'll make the picture have it's vertical resolution cut if half right? From x480 to x240 (using NTSC) So if you upscale it back to 480, how will you keep the image from becoming pixilated (jagged lines), unless there's something around that, that I'm not thinking about or reading correctly?

You use deinterlacing techniques to "approximate" the missing pixels during step 6 (adding the "Field Bob" filter). Yes, there is definitely quality loss - in essence it's not a 50%, half quality, loss though. I believe it is most noticable when there are vertical lines and fast motion. Most likely the quality will be acceptable, but try it for yourself and see what you get!

Can someone post there footage results please I would like to see how well this works.

Thanks

This is great and everything, but can anyone translate this so it can be done on a mac? Forinstance, can anyone mimic this in say... after effects or anything like that?

In After Effects, select the shot you want to slow down in the project window and from the menu bar choose File>Interpret Footage>Main...

Under "Fields and Pulldown" change the "Separate Fields" option from "Off" to either "Lower Field First" for DV or "Upper Field First" for HDV - this separates the fields and treats the clip like a 60fps progressive source. Check the "Motion Detect - Best Quality Only" to do a "smart deinterlace" which will attempt to maintain vertical resolution when/where there isn't any motion.

Edit: almost forgot, under "Frame Rate" check off "Conform to frame rate:" and change from 29.97 to 30. (This lets the slowdown to 24 later be an even, easy percentage)

Click OK to apply the changes.

Now create a new composition with a frame rate of 24 fps. Add your clip to the composition, select it, and from the menu select Layer>Time Stretch...

Set the stretch factor to 250%. Render. Admire silky smooth slow motion goodness...

If you are doing this with SD source material you get some vertical softening from the deinterlacing. If you use 1080 60i HDV source material you'll end up with a full resolution SD final comp which looks much better.

Edit: I just posted a sample clip here:

http://www.divergentshadows.com/hdv/slow_motion.html

Not a great shot, just some footage of my dog running which I shot yesterday. 17 second Quicktime, H.264, 3mbps, ~6.3MBs

Nifty little camera trick!! it works like a charm and I dig it.

http://www.amfmstudios.com/mejumpingslowmo.wmv

This is the only copy I have, sorry it's not full resolution so you can get a comparison but I thought I'd share the result for all to see (and I hope I don't get sued for the music either ieee)