VHS to digital - capture video is too fast

[manono]

Adding DoubleFPS2() doubles the number of frames, so playback speed is now one-fourth.

Nope, not if you're playing the encoded final output and not the script. The speed will be 3/4 the speed of the source (15/20=0.75).

And good going for getting that complex script working.

[jagabo]

In case it's not clear: your computer isn't fast enough to run that script in real time. But the final encoded video will play at the rate it's supposed to. So even if your computer can only run the script at 7 fps the final video will play at 30 fps.

[tns1]

I saved an avi based on this first script. The first playthrough of the avi on Vdub bogged down at one point, but subsequent playthroughs worked fine. There do appear to be some new blur issues during fast panning sections, so I will compare with a ChangeFPS version. Should I be creating a new script to work with this intermediate avi (I assume this is faster), or is there some advantage to having a single monolithic script working on the original (is loss-less really no loss)?

Same question for encoding - use original script or use intermediate avi?

I have loaded HCenc, and there are a lot of settings in the gui. Which ones do I care about?

[jagabo]

Motion interpolation doesn't work well when motions get too large. Partly because the picture has motion blur making it more difficult to detect what's moved where, and partly because it only searches a limited distance for motions. You may end up with just a blend of the two frames or distortions like those pictured in the link I gave earlier.

[tns1]

If I want to clean up the video, shouldn't I be attempting that before I introduce more frames with duplicate (or new) issues?

[Cornucopia]

Yes that is better, but you'll want to first make sure what you have so far moves the way you want it to.

Scott

[jagabo]

Yes, generally you want to decimate to the original film frames, then perform the cleaning, stabilization, etc., then motion interpolation.

[tns1]

As a test I used HCenc with default settings to produce an m2v. It was a relief to see 9G of data reduce back to 150M.

So starting with my 14.99fps progressive video:
ChangeFPS produces cleaner images, but with jerky motion,
DoubleFPS2 produces nice smooth motion, but less clean,
Speed of either one is fine.

Of course I want both clean images and smooth motion. Is it safe to assume I can always get the smooth motion and correct speed later, and take a step back for the cleanup? What would you do next?

[jagabo]

You can try svpflow instead of mvtools.
http://www.svp-team.com/wiki/Plugins:_SVPflow
http://www.svp-team.com/wiki/Plugins:_SVPflow#Base_script

Or instead of ChangeFPS() try ConvertFPS() which will create blended frames between each of the originals.

[Cornucopia]

Or spend a whole lot of time by doing all 3 and then averaging the frames.

Scott

[manono]

Is it safe to assume I can always get the smooth motion and correct speed later, and take a step back for the cleanup?

As already mentioned twice, decimate to 14.985fps, do any and all cleanup, and lastly make it for the final framerate. You might want to do these things in stages, particularly if using slow AviSynth filters along the way, saving out as a lossless AVI during each stage until ready to do the final encode in HCEnc.

Is it safe to assume I can always get the smooth motion and correct speed later...

It's a bit of a tradeoff. You can go for clean and slightly jerky playback, or you can go for sometimes artifact-ridden but smooth playback. With jerky and handheld video as seen in the sample, there are times when you won't get both clean and smooth playback no matter what you use to do it. I, for one, would refuse to use ConvertFPS on it. It'll leave a blended mess, something you spent the first part of this thread learning how to remove. Sometimes there are differences of opinion on how to do things. It depends on what looks best to you and your family and friends. For example, most people might not notice the artifacts created by motion interpolation during certain kinds of movement. For those of us who do a lot of this stuff and examine these things frame-by-frame, those artifacts sometimes stand out like a sore thumb.

[jagabo]

Do you still have the film? A film scanner can give you much better results than the projection method that was used to make your VHS tape. Here's post with a Youtube link by a guy who's playing with a super 8 film scanner now:

https://forum.videohelp.com/threads/366069-The-Reflecta-Super8-film-scanner-to-avi-conv...=1#post2336582

[jagabo]

By the way, if you save intermediate stages with VirtualDub be sure to keep the source colorspace -- that will be YV12 with what you're working on. The easiest way to do that is to use Video -> Fast Recompress mode in VirtuaDub.

[tns1]

There is a small chance the film still exists somewhere, but it isn't likely I can get it. That scanning process looks expensive anyway.

In Vdub, the default mode is set to full processing. If I select YV12 for input,output colordepth instead of rgb24, the output avi is about half the size without noticeable loss which makes sense. If I select fast recompress, colorspace is greyed out, and the resulting AVI is tiny and not playable. Bear in mind that I only just now figured out how to mark-in, mark-out to trim the video, so I think you are way ahead of me.

[manono]

If I select fast recompress, colorspace is greyed out, and the resulting AVI is tiny and not playable. Bear in mind that I only just now figured out how to mark-in, mark-out to trim the video, so I think you are way ahead of me.

With Fast Recompress you get back the same colorspace as you started with. I don't know why you're getting a small and unplayable AVI. Are you choosing a codec and configuring it? You should install a lossless codec. I use Lagarith but there are others. Make sure it's outputting in YV12, like your source.