twixtor slow mo? at DVinfo.net

Eugene Presley · March 19th, 2008, 01:04 AM

an somebody explain step by step how to use the Twixtor plugin to get smoot slow mo?

Thx!

David McGiffert · March 23rd, 2008, 12:41 PM

Eugene,

Good question...
That's the reason I don't use Twixor, it's ridiculously
complicated for the occasional user...

Have you tried using SmoothCam inside FCS 2?
It's pretty good.
Also in Motion there is a way to slow footage that works
beautifully.

Good luck whichever way you go,

David

Robert Lane · March 24th, 2008, 10:28 AM

This question caught my eye; I've seen output from Twixtor and was impressed but have never used it, and after seeing Eugene's puzzlement I decided to download the demo and see for myself how difficult it would be.

I must say, Twixtor is without a doubt far easier, faster and more intuitive than using SmoothCam and unlike the Apple's Optical Flow time remapping it doesn't change the length of the clip! (Apples time-mapper can also *not* affect the clip timing but it's a much more laborious process to get that to work properly). "...it's ridiculously complicated..."? Couldn't disagree more.

After 5 minutes of fiddling with Twixtor I'm impressed; the output is glass-smooth and super-easy to manipulate with just keyframes alone, not to mention the other controls that I haven't even tried yet.

I can't say the "pro" version of Twixtor would be of value since it's hard to justify one filter costing over $500 and even the "standard" pricing is high, but in spite of that I'm definitely purchasing a standard version and in fact, I'm replacing this plug-in in several sequences with ones previously built with the Apple time-remapper.

This filter combined with footage from either the HPX2000 or 3000 practically gives me the P2 Varicam that Panny hasn't come out with yet - for far less money! (laughs)

Eugene, if you're not understanding Twixtor I'd suggest you first understand how keyframes work - maybe that's why it's not working for you. (?)

Ryan Mueller · March 24th, 2008, 03:56 PM

To answer Eugene's original question: it really depends on what kind of footage you are using. Are you using progressive footage or interlaced? Slow-mo in Twixtor also works best with a shutter speed of about 1/100th. Let me know what kind of footage you're working with and I'll write you up a quick tutorial when I get a chance.

Ryan

Jason Bowers · March 24th, 2008, 06:00 PM

Ryan,
I would also like a write up. We are shooting 24f at 1/48 and 1/24. Basically from scratch would be great.

Thanks

Robert Lane · March 24th, 2008, 07:37 PM

Not to put a damper on Ryan's generous offer, but it's really not that difficult, nor is Twixtor that finicky with footage.

I threw 24p (180 deg shutter), 30p 60i and 1080p(i) at it and it handled it all perfectly well. The online instructions are quite clear and as I mentioned earlier, it's far easier to use than SmoothCam - and much faster at rendering the revised clip.

The question might be how well it handles any non I-frame codec such as HDV where the long-GOP structure might cause stuttering playback issues (that's been the case with many who've tried using SmoothCam with HDV/XDCAM.

The easiest way to get your head around how this filter works - and whether or not it's worth the asking price - is to download the free demo and give it a whirl. I'm sure most will enjoy and like it's output; I certainly do.

Ryan Mueller · March 24th, 2008, 09:27 PM

Hey guys,

Robert is right, it is not that difficult to use, but some people still have a hard time with it. Rather than give them a hard time about it, I'd rather help them out.

Yes it does make a huge difference with the footage that you pull into it! 1/48th footage can be done, but shouldn't be slowed down further than 50% or it will look horrible due to motion vector compensation and the lack of information between frames. 1/24th I probably would not try to slow-mo, but you can play around with it, Jason, and see if you can make it work. 1/100th works extremely well and can be successfully slowed down to 25% and sometimes slower.

So with that said, here is a quick run-through to get you guys started:
Pull the clip into your viewer,
apply the twixtor filter,
set display "twixtored output",
set input fields to "none" for progressive footage or
set input fields to "lower field first" for interlaced SD footage or
set input fields to "upper field first" for interlaced HD footage,
set motion vectors to "best"
set image prep to "contrast/edge enhance",
set your speed percentage to desired speed i.e. 50(%)
set frame interp to "blend",
optional: enable smart blend and adjust motion blur compensation to desired look (this will sharpen the subject and remove some motion blur)

Any of the settings that I have not mentioned, I usually just leave in the default position and some of the settings that I mentioned are actually the default option. Keep in mind that it will not "stretch" your clip to fit the edit, so you may need to adjust the clip after adjusting the speed. The instructions on their site give you a full explanation of what I mean.

The best thing to do beyond what I have mentioned is to just play around with the program and its settings as Robert had mentioned, and to support what Robert said, I too think that the product is well worth the price.

Robert Lane · March 24th, 2008, 10:57 PM

I almost never work with interlaced footage, so it could be that interlaced slower than 1/48th footage might not look good when using any time-interpolation, but so far all my test clips - even DPX footage from tele-cine looks great even as slow as 20% of original speed - that's -80 reduction - and much better than what Apple's Optical Flow was able to pull off. And damned disappointing too, considering how much they pushed the capability on us at NAB last year. humph.

Just as a point of reference: Most of my Burbank/LA clients are shooting either film or, 24p video with either 1/24th or 180deg shutter, and it was their footage that I first saw Twixtor being used. Having said that, it was also within the confines of a After Effects/Flame system but so far the FCP6 clips are giving near-identical results with the same or similar clips.

There could be a distinct reason for our differing results: I'm not using any image prep, no "contrast/edge enhance" because like a sharpening filter than can create very harsh looking edges when applied improperly - and could also be easily misinterpreted and over-done by the time-remap interpolation process. It's also important to consider that if the filter really only worked well with certain shutter speeds/angles then it really wouldn't be of much use as a plug-in, to have such built-in limitations. As Ryan points out, experimentation and lots of trial-and-error will show you the weaknesses of the footage you attempt to interpolate and what settings work best.

So Ryan's and my results are conflicting, which isn't that important and actually expected since we're using differing production methods, but as he says, it's well worth the trial and investment.

March 19th, 2008, 01:04 AM	#1
Eugene Presley Major Player Join Date: Jan 2003 Location: Antwerpen (Belgium) Posts: 220	twixtor slow mo? an somebody explain step by step how to use the Twixtor plugin to get smoot slow mo? Thx!

March 23rd, 2008, 12:41 PM	#2
David McGiffert Regular Crew Join Date: Feb 2007 Location: Topanga, CA Posts: 139	Eugene, Good question... That's the reason I don't use Twixor, it's ridiculously complicated for the occasional user... Have you tried using SmoothCam inside FCS 2? It's pretty good. Also in Motion there is a way to slow footage that works beautifully. Good luck whichever way you go, David

March 24th, 2008, 10:28 AM	#3
Robert Lane Go Go Godzilla Join Date: Nov 2004 Location: Scottsdale, AZ USA Posts: 2,788 Images: 15	This question caught my eye; I've seen output from Twixtor and was impressed but have never used it, and after seeing Eugene's puzzlement I decided to download the demo and see for myself how difficult it would be. I must say, Twixtor is without a doubt far easier, faster and more intuitive than using SmoothCam and unlike the Apple's Optical Flow time remapping it doesn't change the length of the clip! (Apples time-mapper can also not affect the clip timing but it's a much more laborious process to get that to work properly). "...it's ridiculously complicated..."? Couldn't disagree more. After 5 minutes of fiddling with Twixtor I'm impressed; the output is glass-smooth and super-easy to manipulate with just keyframes alone, not to mention the other controls that I haven't even tried yet. I can't say the "pro" version of Twixtor would be of value since it's hard to justify one filter costing over $500 and even the "standard" pricing is high, but in spite of that I'm definitely purchasing a standard version and in fact, I'm replacing this plug-in in several sequences with ones previously built with the Apple time-remapper. This filter combined with footage from either the HPX2000 or 3000 practically gives me the P2 Varicam that Panny hasn't come out with yet - for far less money! (laughs) Eugene, if you're not understanding Twixtor I'd suggest you first understand how keyframes work - maybe that's why it's not working for you. (?)

March 24th, 2008, 03:56 PM	#4
Ryan Mueller Major Player Join Date: Mar 2007 Location: Highlands Ranch, CO Posts: 336	To answer Eugene's original question: it really depends on what kind of footage you are using. Are you using progressive footage or interlaced? Slow-mo in Twixtor also works best with a shutter speed of about 1/100th. Let me know what kind of footage you're working with and I'll write you up a quick tutorial when I get a chance. Ryan

March 24th, 2008, 06:00 PM	#5
Jason Bowers Major Player Join Date: Feb 2005 Location: Thunder Bay, ON. Canada Posts: 374	Ryan, I would also like a write up. We are shooting 24f at 1/48 and 1/24. Basically from scratch would be great. Thanks

March 24th, 2008, 07:37 PM	#6
Robert Lane Go Go Godzilla Join Date: Nov 2004 Location: Scottsdale, AZ USA Posts: 2,788 Images: 15	Not to put a damper on Ryan's generous offer, but it's really not that difficult, nor is Twixtor that finicky with footage. I threw 24p (180 deg shutter), 30p 60i and 1080p(i) at it and it handled it all perfectly well. The online instructions are quite clear and as I mentioned earlier, it's far easier to use than SmoothCam - and much faster at rendering the revised clip. The question might be how well it handles any non I-frame codec such as HDV where the long-GOP structure might cause stuttering playback issues (that's been the case with many who've tried using SmoothCam with HDV/XDCAM. The easiest way to get your head around how this filter works - and whether or not it's worth the asking price - is to download the free demo and give it a whirl. I'm sure most will enjoy and like it's output; I certainly do.

March 24th, 2008, 09:27 PM	#7
Ryan Mueller Major Player Join Date: Mar 2007 Location: Highlands Ranch, CO Posts: 336	Hey guys, Robert is right, it is not that difficult to use, but some people still have a hard time with it. Rather than give them a hard time about it, I'd rather help them out. Yes it does make a huge difference with the footage that you pull into it! 1/48th footage can be done, but shouldn't be slowed down further than 50% or it will look horrible due to motion vector compensation and the lack of information between frames. 1/24th I probably would not try to slow-mo, but you can play around with it, Jason, and see if you can make it work. 1/100th works extremely well and can be successfully slowed down to 25% and sometimes slower. So with that said, here is a quick run-through to get you guys started: Pull the clip into your viewer, apply the twixtor filter, set display "twixtored output", set input fields to "none" for progressive footage or set input fields to "lower field first" for interlaced SD footage or set input fields to "upper field first" for interlaced HD footage, set motion vectors to "best" set image prep to "contrast/edge enhance", set your speed percentage to desired speed i.e. 50(%) set frame interp to "blend", optional: enable smart blend and adjust motion blur compensation to desired look (this will sharpen the subject and remove some motion blur) Any of the settings that I have not mentioned, I usually just leave in the default position and some of the settings that I mentioned are actually the default option. Keep in mind that it will not "stretch" your clip to fit the edit, so you may need to adjust the clip after adjusting the speed. The instructions on their site give you a full explanation of what I mean. The best thing to do beyond what I have mentioned is to just play around with the program and its settings as Robert had mentioned, and to support what Robert said, I too think that the product is well worth the price.

March 24th, 2008, 10:57 PM	#8
Robert Lane Go Go Godzilla Join Date: Nov 2004 Location: Scottsdale, AZ USA Posts: 2,788 Images: 15	I almost never work with interlaced footage, so it could be that interlaced slower than 1/48th footage might not look good when using any time-interpolation, but so far all my test clips - even DPX footage from tele-cine looks great even as slow as 20% of original speed - that's -80 reduction - and much better than what Apple's Optical Flow was able to pull off. And damned disappointing too, considering how much they pushed the capability on us at NAB last year. humph. Just as a point of reference: Most of my Burbank/LA clients are shooting either film or, 24p video with either 1/24th or 180deg shutter, and it was their footage that I first saw Twixtor being used. Having said that, it was also within the confines of a After Effects/Flame system but so far the FCP6 clips are giving near-identical results with the same or similar clips. There could be a distinct reason for our differing results: I'm not using any image prep, no "contrast/edge enhance" because like a sharpening filter than can create very harsh looking edges when applied improperly - and could also be easily misinterpreted and over-done by the time-remap interpolation process. It's also important to consider that if the filter really only worked well with certain shutter speeds/angles then it really wouldn't be of much use as a plug-in, to have such built-in limitations. As Ryan points out, experimentation and lots of trial-and-error will show you the weaknesses of the footage you attempt to interpolate and what settings work best. So Ryan's and my results are conflicting, which isn't that important and actually expected since we're using differing production methods, but as he says, it's well worth the trial and investment.