Slow motion quality using MSP

[Patrick]

Ok, for all the MSP experts out there...

If I set the speed of a clip to play back at say 50%, it is really obvious during the preview on my TV monitor that the quality is less than it was when playing back at full speed. Yet, if I play back the same clip with the same settings in place using the Scratch Pad, the quality is every bit as good as the original clip. The clip is just as sharp and there is no slight "strobe" effect. It looks like perfect slow motion.

Is this normal? Can I not achieve the Scratch Pad level of quality with slow motion when the clip is instead being previewed (for cut-list editing) off the timeline?

I'm using a Mill2/RRS combination.

[jeff b]

I've done that a couple times, and it was kinda interesting on the final print back to video. It wasn't Sony Super Slo-Mo, of course, but it produced the desired effect.

I used this on a music video for my sister. I took the song she had recorded and made a double speed copy. Then I had her lip-sync to the double speed playback in front of a green screen. I captured that video and knocked it down to half speed, and spent a considerable amount of time syncing the song to it, but it came out quite nice. I only used a couple seconds of that, color keyed here and there in the final music video, but it really was a nice effect.

Another speed change effect I did once was to capture a clip with chroma backed out (black and white). I sped the clip up to double speed, and overlayed a brownish tinge to it, took out the audio track and replaced it with some 40's Glenn Miller music ('Little Brown Jug'). It looked like an old film, and went over nicely with the people I made it for.

[Patrick]

Jeff, have I missed something in your post? Did you touch on my questions at all?

[fluggo99]

If you've got the field options set right, MSP should do some nifty things with the fields to make slo-mo look wonderful. Done it, loved it. Check the following things, Patrick:

- You need to be calculating field stuff to see the effect. Make sure your clips are set to the appropriate fieldage ("polarity" I think is the word);
- that your preview shows all the scan lines (vert rez = 480 NTSC or 576? PAL);
- and that your preview is set to the right field polarity (option on the first tab, make sure it's not frame-based)
- It should look fine now but if it doesn't, try setting Preferences->Resample Quality->Best. I really don't think that's necessary for good quality preview if your preview is set to full rez.

There are a few clips I've done slo-mo on that just don't cooperate (maybe high-motion clips?), so if NONE of the above works try it on a few other clips. Leave speed at 50% - I've never tried it at anything but multiples of two. (50, 100, 200)

I think Scratch Pad looks great because it's working with the full-rez file.

[jeff b]

Sorry Patrick... it was around 5am...

Fluggo's more on the ball, though, since he actually answered your question. It's all a matter of setting up Preview options to match the properties of the clips you're using in the project. With RR-S, use Field order A. Select 'BEST Preview' method, check 'apply video filters'. Run 29.97fps, and select STANDARD frame size of 704x480 (if NTSC). Check 'interleave video and audio' and pack audio data for every 1 frame. Configure the Matrox MJPEG codec for Maximum compression quality, and optimize for quality. I also use 44.1kHz stereo on everything, whether I need it or not, just to be consistent across all my projects.

The one thing about doing this slowdown that can make it look not so good is having a lot of motion: the more and faster the motion, the worse this effect comes off.

[Patrick]

Fluggo and Jeff, thanks for all your tips. I double and triple checked everything and my settings are correct. However, having said all that, I think I've discovered something that is simply amazing. I can hardly wait to share this.

My complaint with the slow motion was that the image quality was poorer than with the original clip. Even when the preview rendering was going on, I could see that every other frame (field?) being rendered in the preview window was of poorer quality. Yet for some reason, if I played backed this slow motion clip using the Scratch Pad, the image quality was fine. I found this very frustrating. Out of desperation, I tried changing the field order and/or frame settings in several locations. Nothing worked until.....

With every other setting where it was supposed to be, I left-clicked the original (unaltered) clip, right-clicked to bring up the Properties, changed Field order A to Frame ( I did NOT select Flicker reduction), changed the Speed to 50%, then exited and clicked Preview. The first thing I noticed was that the preview window did NOT show any image and the preview rendering was VERY fast. Keep in mind that Preview Options were set for Field Order A. When the rendering was finished and I clicked Preview to see the slow motion clip, I couldn't believe my eyes. It was perfect!!! The image quality of the slow motion clip was just as good as the original clip. It's unbelievable!

It seems too good to be true. The rendering time for the slow motion clip was faster and the results were superior. I would really like to hear if anyone can duplicate these results.

[Biker]

Hi Chaps

Just had to comment. - Although enjoying Desktop editing very much with the G200 Marvel + MSpro packaged version.

The two things that really get me are -
1) Really nice titling program (but there is probably a plugin somewhere for that?) and
2) Really AWFULL Slo Motion! Most of my video editing is fast Motorcycles, that not only move forward fast but hit bumps and go vertical unexpectedly And MSP really misbehaves.

It is at this point that I have to go back to my Panasonic FS90 S-VHS edit deck. Which out puts PERFECT Slo Motion, BUT the matrox card will NOT accept it. the picture goes to pieces!!! But Linear tape catches it perfect, albeit 2nd generation.

Any suggestions please?

Biker

------------------

[Patrick]

Biker, by looking at the time that you posted, I would have to guess that you were probably unaware of my last post. Try what I did and see if it makes any difference for you. I'm anxious to hear if it does or not.

[Patrick]

When I got home tonight, I did some more experiments. I found that there can indeed be some problems when slowing down a frame based clip if the camera was panning during the shot (or if there's any kind of fast movement as both Fluggo and Jeff pointed out). However, I'm still delighted about some other findings which I'll pass on tomorrow after some sleep.

[jeff b]

Patrick: Now that I think about it, it makes sense that the duplicated frames in a stretched out clip might behave better if you switch over to frame based properties. I'm assuming that you originally captured 704x480 field order A based... But what really gets me laughing is that it's possible, just possible, that somewhere hidden in the dark of all this, there just might be a... That's right, you guessed it!

>>>>>>> an EVIL FRAME? <<<<<<<

Script segment for game show:

Announcer: 'Tell them what they won, Don Pardo!'

Don Pardo: 'An all expense paid trip to the EVIL FRAME HOTEL!!! That's right, the EVIL FRAME HOTEL, on the beautiful shores of Lake Nonlinear in the Canadian province of Ontario! There, you will be served by seemingly beautifully captured clips, imported from distant devices, and served up in SUMPTUOUS luxury in your beatifully appointed time line lodgings!'

Announcer: 'Okay! Well, Patrick, are you ready to select your next property change, here on the world's most popular non-linear editing game show, "What's My Project?"; Or do you want to take the prizes you have now and retire?'

<click!>

[Alexei]

Jeff,

after I've spent half of last night red-eyed hunting for that one EVIL FRAME in my 25-minute project,

IT AIN'T FUNNY ANYMORE!!!!

[This message has been edited by Alexei (edited 21 December 1999).]

[Patrick]

Jeff, I'm afraid you're losing it...

Interestingly enough, I did look into how the dreaded EVIL FRAMES would behave during these tests. It turns out that there were some instances where the switch to frame based clip properties would eliminate them. But before you get all excited Alexei , I also have to add that the EVIL FRAMES were NOT eliminated every time that I ran my tests. So I normally "culled" the EVIL FRAMES from the clip before I continued my experiments. I have it down to fine art now!

What I discovered last night that really pleased me was the HUGE time saving I could achieve when I wanted to change the speed of a clip, either faster or slower. I capture all clips at 704 x 480, field order A, lowest compression. If I was to take a 10 second clip, retain field order A, and have it play back at 50% speed, it would take 129 seconds for the preview to render. If I instead switched the clip properties to frame based, it took only 12 seconds to render! That is more than 10 times faster! Not only that, but the added bonus is that the image quality is better!

Now as great as this all sounds, and it is great, this will not work with any and all shots. If the camera was being quickly panned, or if the subject was moving quickly through the shot, there will be a real jerkiness to whatever is moving. But with a lot of stuff that has been shot with the use of a tripod, the results are superb. That's not all...

Slowing a clip down is one thing, there's also the option of speeding it up. The problem normally associated with increasing the speed of a longer clip is the rendering time involved. Let's say you wanted to see a ten minute clip played back at 1000% speed. The rendering time would be enormous. However, change the clip properties to frame based and then try it. The rendering absolutely zips along. And again, the image quality is superb. With a speeded up clip, potential problems with jerkiness of movement are almost nonexistent, especially with the very increased speeds.

My original questions remain unresolved. I'd still like to be able to achieve better image quality during slow motion in situations where I'm forced to retain the field order A clip property setting. These would be situations where either the camera or the subject was moving too quickly, and changing the clip properties to frame based would lead to excessive jerkiness of movement.

Any other suggestions out there?

[This message has been edited by Patrick (edited 22 December 1999).]

[Keith Wiebe]

I like the way field order A clips look on the computer (running the Canopus Raptor DV card) but when outputting to TV or tape I get the "evil frame" syndrum. I have found out if I set the clip to frame based and select flicker reduction and render back into field order A it looks great although there is more jerky movement because I'm assuming half the movement gets thrown away when going to a frame based clip. Try 33% slow with frame based and render it to field order A and see what happens! P.S. I've never gotten an "evil frame" with the Canopus yet but when doing slow mo it will get jerky or under certain sphere moving paths of still pictures I had trouble with the jerks when trying to make a still picture out of the sphere.

[Patrick]

Keith, after reading your post I decided to try every combination possible with the clip properties to see which one resulted in the best image quality when creating an avi playing back in slow motion. In this case, at a speed of 33%. The clips were all rendered back to field order A and the same 10 second original clip was used for every test. There was both movement and no movement in different parts of this clip. The results are listed in order with what I believe to be the best overall quality first. The avi's were being rated on how they appeared on a TV monitor, not a computer monitor.

Clip Properties______Image__Jerkiness_Rendering time

____Frame___________Great_____Yes_____18 seconds!!

_Field A/deinterlace__Little grainy__Yes_____228 seconds

____Field A_________More grainy__None___209 seconds

Frame/flickerReduction_Less sharp__Yes_____260 seconds

The rendering time of the best overall avi is NOT a misprint. It only took a fraction of the time that the others took. In parts of the clip where there was little or no movement, the number one rated avi looked really good.

I want to make it absolutely clear that the jerkiness described here and earlier in this thread by me has nothing to do with EVIL FRAMES and the "vibration" that I've complained about in other threads. The jerkiness in these slow motion avi's has been caused by movement in the shot, either by the camera or by the subject.

I hope this is of some interest to people other than myself. Please feel free to ask any questions or even to dispute my findings if you've experienced different results.

[jeff b]

This discovery of changing a clip's properties to FRAME, and getting not only great motion on time stretched clips, but also the very fast rendering time is really a cool discovery! It would be interesting to know if anyone else had ever discovered this little gem, prior to this thread?