How well does your editor create slo-mo ?

grahamlthompson

In memoriam
Following another thread that went off topic decided to create a thread discussing how various editors handle creation of slo-mo.

Initial tests used a sequence of 25 frames with each frame being 1920 x 1080 containing a sequence of number 00, 02 --- 48. Photoshop was used to create the images and the resulting 1 second video clip (25 fps), though any editing software could have been used.

The resulting clip when loaded into a video editor and single frame advanced generates a clear image for each frame despite mpeg compression.

The clip can be found here

ADrive | Online Storage, Online Backup, Cloud Storage

I then used Vegas Pro 12 to stretch the clip to the maximum possible in one go and the same using Magix Pro X6.

The results are interesting in that Vegas clearly uses frame interpolation to produce new intermediate frames and Magix merely repeats the first frame a number of times.

The images show the results of the intermediate frames generated between the original frame containing the number 12 and the next frame containing 14.

Next step I am going to add a simple 25 frame animation to the original frame sequence to see how that gets treated, unless someone has a better suggestion.
 

Attachments

  • Vegas25fpsStretch.jpg
    Vegas25fpsStretch.jpg
    41.1 KB · Views: 58
  • MagixX625fps.jpg
    MagixX625fps.jpg
    45.5 KB · Views: 76

PhilipL

Well-known Member
Hi

This is a pointless exercise, no editor or software is clever enough to create an extra number for interpolated frames in your test clip.

Also your sample doesn't contain any actual motion, it is just a series of changing numbers, how can you slow motion that?

Just search Google for "creating slow motion" no need to try and re-invent the wheel, loads of information out there.

Vegas will interpolate or add extra fields if you change the clips frame rate and turn off smart resampling, you choose what looks best for the type of footage.

Regards

Phil
 

grahamlthompson

In memoriam
It's anything but pointless.

I have two Sony HD camcorders, both of which declare some degree of slow motion capability. Analysis of footage from both cameras shows clearly they cheat. They do not record at an increased frame rate at all but in fact simply use the camera buffer to record the same frame more than once to the recording file.

The shutter speed they are using precludes recording at a higher framerate (you cannot for instance record 100fps if a single frame exposure takes longer than 1/100 second). Simply looking at the footage shows no increased noise or a reduction in the ability to handle low lighting conditions.

It's very easy to use the same tricks to precisely emulate what the camera does in an editor.

This has some advantages :

There is effectively no limit to the duration of the footage you can process.

You can retain and pitch correct the audio.

No one expects that it will be possible to recreate the missing frames, the whole point in using footage to test is that you know exactly what each frame will contain when shot normally. This is by far the easiest way to spot adjacent identical frames.
 

PhilipL

Well-known Member
Hi

I would try setting your Sony cameras to a higher shutter speed, if the shutter speed is lower than the frame-rate you would see duplicate frames, no surprises there.

Also it depends on what the camera is playing back as, if it is 100fps but slow motion is quarter speed then to conform to 50i/50p you will get two duplicated frames.

If they still are just duplicating frames take it up with Sony as clearly they are false advertising. Typically when a camera states it can record at a higher frame rate to produce good slow motion, than that is what it does, if it doesn't take it back for a refund.

Regards

Phil
 

rogs

Well-known Member
I have two Sony HD camcorders, both of which declare some degree of slow motion capability. Analysis of footage from both cameras shows clearly they cheat. They do not record at an increased frame rate at all but in fact simply use the camera buffer to record the same frame more than once to the recording file.
Just another version of the problem that comes with some camcorders that claim a 720/50p capability when recording in the 'AVCHD Lite' format. They don't exactly 'cheat' - they do tell you in the specs that the sensor output is 25p.
To make the footage into 50p, they simply then use 'frame doubling' - presumably to make the footage Blu-ray compatible?

Adding interpolated frames in software to help improve 'motion smoothness' can work OK with this kind of footage.. but it depends on the subject.

This tool is the best I've found so far: MVTools

Extending it's function by 4x or 8x, for slo-mo is a bit less predictable though...it can be very good. Whether any of the commercial editors can do better?......
 

Terfyn

Well-known Member
I use the option in VideoStudio by stretching or compressing the video on the timeline. Within limits it produces good slow motion.
I can also use the same technique on the audio track, this is interesting as the pitch does not change but the time slows or speeds up. Very useful for matching the audio to a video sequence

Whether any of the commercial editors can do better?......
Why don't you download the free trial programmes to test your theory.
 

grahamlthompson

In memoriam
Hi

I would try setting your Sony cameras to a higher shutter speed, if the shutter speed is lower than the frame-rate you would see duplicate frames, no surprises there.

Also it depends on what the camera is playing back as, if it is 100fps but slow motion is quarter speed then to conform to 50i/50p you will get two duplicated frames.

If they still are just duplicating frames take it up with Sony as clearly they are false advertising. Typically when a camera states it can record at a higher frame rate to produce good slow motion, than that is what it does, if it doesn't take it back for a refund.

Regards

Phil

You don't have any manual controls when using the slo-motion capabilities. Sony don't make any claims other than it provides slowed down action on recordings .

Smth Slow Records a 3 second clip which plays back in about 12 seconds. It's only by actually looking critically at the footage you can see what is actually going on. I posted the actual algorithm previously.

You can recreate this in most editors by using a variety of frame holds.

Spec for XR500E (which lacks 50p recording). Standard Shutter speed 1/50sec 11lux minimum illumination, 1/25 second 3 lux, 0 lux in nightshot mode (IR).



See Smth Slow Rec in the manual here

http://pdf.crse.com/manuals/4131473111.pdf
 

grahamlthompson

In memoriam
Hi

If Sony don't make any claims then what is the issue? Who cares?

Regards

Phil

Anyone who wants to create slow motion the same way using any camcorder, or is simply interested in how it works or wants to create similar footage over an extended time period.

If you aren't interested why bother posting here at all ?

Just found the US version of the Panasonic V750 records slow mo at 120fps and interpolates in camera to 240fps (probably by doing the same as Sony). No idea if the UK version uses 100fps interpolated to 200fps.

Panasonic Continues to Lead the Way With the Introduction of Five... - Panasonic UK
 

Terfyn

Well-known Member
No idea if the UK version uses 100fps interpolated to 200fps
Perhaps if you read my posts or even downloaded the manual. Research always helps to eliminate simple errors.
 

PhilipL

Well-known Member
Hi

Most people with a serious need for slow motion for sports action etc will go the Go Pro route.

Regards

Phil
 

grahamlthompson

In memoriam
No idea if the UK version uses 100fps interpolated to 200fps
Perhaps if you read my posts or even downloaded the manual. Research always helps to eliminate simple errors.

Don't you think I have tried that ? The manual doesn't have this information, the only reference is on Page 56 of the pdf. Amazingly there is no camera specifications in it either.

All it says is

The shutter speed is 1/120 or more.
≥ The number of pixels/frame rate of FULL HD Slow Motion Video recording are fixed to
1920k1080/60p.

How does this answer the question ?
 

Terfyn

Well-known Member
To quote from #12 of the previous post:-
The camera records at 100fps (PAL) and interpolates to 200fps
This is directly taken from the features section of the Panasonic 750 web site.
 

rogs

Well-known Member
:offtopic: With all due respect this started as a thread about the 'slo-mo' options available in various editors . It has now become a discussion on the capabilities of the 'slo-mo' function of a particular camcorder.
Perhaps it might it be better to continue that discussion in this thread?:
Panasonic V750 100 fps | AVForums
 

Terfyn

Well-known Member
:offtopic: With all due respect this started as a thread about the 'slo-mo' options available in various editors . It has now become a discussion on the capabilities of the 'slo-mo' function of a particular camcorder.
Perhaps it might it be better to continue that discussion in this thread?:
Panasonic V750 100 fps | AVForums
Well in that case you had better get a post for GoPro as well. As it happens I was replying to GLT, who commented regarding the recording speed in a PAL camera. I respectfully suggest you read the context before criticising.
 

12harry

Distinguished Member
This is glt's Thread I think, so if he can discuss SloMo in both Camcorders and Editors - why not?
I too support that as the two are not entirely separate... if I can use Movie Studio to create stunning SloMo then I wouldn't bother with the SloMo in the CX410 - but that camcorder is very good at SloMo and the detail is excellent, so until proven otherwise I think the Camcorder does use a higher shutter-speed and frame-rate, but as discussed on other threads... trying to find a suitable "Test" is very difficult.

It's interesting the ( Post#1), difference glt finds between Magix and VegasPro . . . but if Magix is only repeating one frame . . . how does that make the motion right? - Certainly it will take longer, but won't folks notice if the image remains the same? That would be like a Stop-frame effect ( I mean where the movement freezes, e.g. just when the play finishes, etc.)... and the result would be Stutter, surely - or have I mis-read yr conclusions, perhaps?

I recall that the CX410 LCD-image in SloMo is a tad darker, but somehow that gets brightened when in Playback.... also the frame is not 16x9 when I view the clip in a "Player" on my PC - but after importing into Movie Studio timeline, it's back to 16x9 -

glt - that .pdf for yr Sony camcorder is dated 2009, so I'm not sure if we can presume Sony uses the same technique in 2013 (when I bought the CX410)....but nevertheless it's useful to have an Operator's Manual on file, thanks.

Surely their "golf-shot" demonstrates that a Sony camcorder can use a high shutter-speed - otherwise the head of the club would be a blur? Of course the same technique may not apply to the full-frame SloMo, but it strikes me that there's far more detail than a stretching of normal footage, which we know is filmed at a slow shutter-speed if only to make the action look fluid on playback. I think glt's CX410 is slightly different to mine... as mine films only 2 Sec and makes it 11sec.
That's slightly inconvenient guessing when to "start", if you know the thing stops so soon.... a pity I can make the "take" longer...but it's a super effect when replaying semi-random action like water splash, or apples falling from a shaken tree.

If a camcorder was playing with frames, as some suggest; then you should be able to stretch (in camera), any pre-recorded events (but you can't) - and surely the creation would be a complete blur, ( ~5x as much! ), wouldn't it? - That's a very good reason for not doing it that way.

Stills cameras have a wide range of "Shutter speeds" -although one has to remember these aren't exactly like mechanical shutters ( e.g. Compur, or Focal-place which have their own oddities! ) - AND- most will record video.
So, I don't understand why folks think a modest camcorder can only film at normal shutter speeds? There is no equivalent to a Bolex Cine, which had a SloMo feature at 64fps - but cost you so much in film, you wanted to Stop as soon as you Started!.

Where this is made complicated is that Sony/Pana won't tell us what we are getting when we press Start in SloMo mode . . . and I suppose some of us just "like to know"...
 
Last edited:

rogs

Well-known Member
In response to Graham's original thread title -- 'How well does your editor create slo-mo', I have to say that, in the case of Edius Neo, the answer is -- not very well!

I tried various options using this small clip: www.tvs.jp137.com/btorig.mp4

It's about 7 seconds long, and deliberately chosen as it has lots of 'nasty' verticals in the background, to test out the effectiveness of different approaches to 'slo-mo'. I've 'stretched' it by a factor of 4 in each case....

The first one here:

www.tvs.jp137.com/btaddframes.mp4
simply adds extra frames to create a 28 second version. (3 extra copies of each frame of course). As Graham describes in his first post, this is what Magix do. The central'featured' boat looks fine of course, but the background is a bit jerky (obviously)

The second one:
www.tvs.jp137.com/btedius.mp4
uses frame interpolation added from within the Edius timeline, by simply adjusting the speed to 25%. As the footage was inputted to the Editor in an intraframe format (Canopus HQ) it is easy to see how Edius has introduced interim frames in an attempt to improve on simple 'frame copying'.
As you can see, it doesn't work very well!

The last version:
www.tvs.jp137.com/btmvtools.mp4
is from an AVIsynth script, which uses the MVtools filter to estimate motion interpolation. Much better results than either of the other two.


I understand Vegas and Corel both use frame interpolation .... if they produce results as good as the MVtools version, then that could produce some very usable results.

If they're like the Edius interpolation, they'll not be much better than simply adding frames, as Magix do!.....
 
Last edited:

grahamlthompson

In memoriam
This is glt's Thread I think, so if he can discuss SloMo in both Camcorders and Editors - why not?
I too support that as the two are not entirely separate... if I can use Movie Studio to create stunning SloMo then I wouldn't bother with the SloMo in the CX410 - but that camcorder is very good at SloMo and the detail is excellent, so until proven otherwise I think the Camcorder does use a higher shutter-speed and frame-rate, but as discussed on other threads... trying to find a suitable "Test" is very difficult.

It's interesting the ( Post#1), difference glt finds between Magix and VegasPro . . . but if Magix is only repeating one frame . . . how does that make the motion right? - Certainly it will take longer, but won't folks notice if the image remains the same? That would be like a Stop-frame effect ( I mean where the movement freezes, e.g. just when the play finishes, etc.)... and the result would be Stutter, surely - or have I mis-read yr conclusions, perhaps?

I recall that the CX410 LCD-image in SloMo is a tad darker, but somehow that gets brightened when in Playback.... also the frame is not 16x9 when I view the clip in a "Player" on my PC - but after importing into Movie Studio timeline, it's back to 16x9 -

glt - that .pdf for yr Sony camcorder is dated 2009, so I'm not sure if we can presume Sony uses the same technique in 2013 (when I bought the CX410)....but nevertheless it's useful to have an Operator's Manual on file, thanks.

Surely their "golf-shot" demonstrates that a Sony camcorder can use a high shutter-speed - otherwise the head of the club would be a blur? Of course the same technique may not apply to the full-frame SloMo, but it strikes me that there's far more detail than a stretching of normal footage, which we know is filmed at a slow shutter-speed if only to make the action look fluid on playback. I think glt's CX410 is slightly different to mine... as mine films only 2 Sec and makes it 11sec.
That's slightly inconvenient guessing when to "start", if you know the thing stops so soon.... a pity I can make the "take" longer...but it's a super effect when replaying semi-random action like water splash, or apples falling from a shaken tree.

If a camcorder was playing with frames, as some suggest; then you should be able to stretch (in camera), any pre-recorded events (but you can't) - and surely the creation would be a complete blur, ( ~5x as much! ), wouldn't it? - That's a very good reason for not doing it that way.

Stills cameras have a wide range of "Shutter speeds" -although one has to remember these aren't exactly like mechanical shutters ( e.g. Compur, or Focal-place which have their own oddities! ) - AND- most will record video.
So, I don't understand why folks think a modest camcorder can only film at normal shutter speeds? There is no equivalent to a Bolex Cine, which had a SloMo feature at 64fps - but cost you so much in film, you wanted to Stop as soon as you Started!.

Where this is made complicated is that Sony/Pana won't tell us what we are getting when we press Start in SloMo mode . . . and I suppose some of us just "like to know"...

At 25fps intermediate frames are a blur, any rapidly moving object is blurred (mpeg compression). If you look closely at footage from HD TV it's very easy to see this. You don't notice this when viewed at normal speeds.

As to shutter speed. it's blindingly obvious. If you try to shoot 100fps footage at 1/50 second by the time you have shot one frame the next frame will be over (you have a double exposure). That's precisely why some camcorders have to cheat.

As to posting what the CX410 does, I aready posted that, for some reason you didn't believe me.
 
Last edited:

grahamlthompson

In memoriam
In response to Graham's original thread title -- 'How well does your editor create slo-mo', I have to say that, in the case of Edius Neo, the answer is -- not very well!

I tried various options using this small clip: http://www.jp137.com/lvs/btorig.mp4

It's about 7 seconds long, and deliberately chosen as it has lots of 'nasty' verticals in the background, to test out the effectiveness of different approaches to 'slo-mo'. I've 'stretched' it by a factor of 4 in each case....

The first one here:

http://www.jp137.com/lvs/btaddframes.mp4
simply adds extra frames to create a 28 second version. (3 extra copies of each frame of course). As Graham describes in his first post, this is what Magix do. The central'featured' boat looks fine of course, but the background is a bit jerky (obviously)

The second one:
http://www.jp137.com/lvs/btedius.mp4
uses frame interpolation added from within the Edius timeline, by simply adjusting the speed to 25%. As the footage was inputted to the Editor in an intraframe format (Canopus HQ) it is easy to see how Edius has introduced interim frames in an attempt to improve on simple 'frame copying'.
As you can see, it doesn't work very well!

The last version:
http://www.jp137.com/lvs/btmvtools.mp4
is from an AVIsynth script, which uses the MVtools filter to estimate motion interpolation. Much better results than either of the other two.


I understand Vegas and Corel both use frame interpolation .... if they produce results as good as the MVtools version, then that could produce some very usable results.

If they're like the Edius interpolation, they'll not be much better than simply adding frames, as Magix do!.....

That's pretty convincing :)

Any chance of making the original clip available so we have a yardstick to work with ?
 

grahamlthompson

In memoriam
In response to Graham's original thread title -- 'How well does your editor create slo-mo', I have to say that, in the case of Edius Neo, the answer is -- not very well!

I tried various options using this small clip: http://www.jp137.com/lvs/btorig.mp4

It's about 7 seconds long, and deliberately chosen as it has lots of 'nasty' verticals in the background, to test out the effectiveness of different approaches to 'slo-mo'. I've 'stretched' it by a factor of 4 in each case....

The first one here:

http://www.jp137.com/lvs/btaddframes.mp4
simply adds extra frames to create a 28 second version. (3 extra copies of each frame of course). As Graham describes in his first post, this is what Magix do. The central'featured' boat looks fine of course, but the background is a bit jerky (obviously)

The second one:
http://www.jp137.com/lvs/btedius.mp4
uses frame interpolation added from within the Edius timeline, by simply adjusting the speed to 25%. As the footage was inputted to the Editor in an intraframe format (Canopus HQ) it is easy to see how Edius has introduced interim frames in an attempt to improve on simple 'frame copying'.
As you can see, it doesn't work very well!

The last version:
http://www.jp137.com/lvs/btmvtools.mp4
is from an AVIsynth script, which uses the MVtools filter to estimate motion interpolation. Much better results than either of the other two.


I understand Vegas and Corel both use frame interpolation .... if they produce results as good as the MVtools version, then that could produce some very usable results.

If they're like the Edius interpolation, they'll not be much better than simply adding frames, as Magix do!.....

That's pretty convincing :)

Any chance of making the original clip available so we have a yardstick to work with ?
 

grahamlthompson

In memoriam
Without any significant analysis SmthSlowRec on the CX410 outputs 1080i 1440 x 1080i at 25fps.

It produced a clip in the above format of 11.92 seconds duration (11:23). The on screen timer started recording at 03:117 seconds and completes at 06:091 seconds on the timer. A clip length of 3.74 seconds thus producing slow-mo at about 1/3 of the original speed (increases duration by a factor of about 3.18 times).

Video
ID : 4113 (0x1011)
Menu ID : 1 (0x1)
Format : AVC
Format/Info : Advanced Video Codec
Format profile : [email protected]
Format settings, CABAC : Yes
Format settings, ReFrames : 2 frames
Format settings, GOP : M=2, N=13
Codec ID : 27
Duration : 11s 880ms
Bit rate mode : Variable
Bit rate : 8 768 Kbps
Maximum bit rate : 16.0 Mbps
Width : 1 440 pixels
Height : 1 080 pixels
Display aspect ratio : 16:9
Frame rate : 25.000 fps
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Interlaced
Scan order : Top Field First
Bits/(Pixel*Frame) : 0.226
Stream size : 12.4 MiB (88%)

Advancing through the clip I frame at a time till I found the clear image of the timer at 03:264.

This frame is repeated 9 times, followed by 4 interpolated frames after which a clear timer image of 03:399.

So it looks like a combination of Magix repeating frames and some interpolation as per Vegas 12.

So a poor relation compared to the Panny, which on paper manages Full 1080i (1920 x 1080) 25fps output at 8 times clip duration.
 

grahamlthompson

In memoriam

12harry

Distinguished Member
glt Post19, I presume the CX410 info you meant was in Post 1,3 etc. but I haven't disbelieved it, rather I'm not aware that you tested it in a scientific manner.
I find it difficult to believe that Sony (the only Mfr I bought) is misleading folks by duplicating frames - they are far too sharp to be normal shutter speeds.....but therein lies another issue - we talk of Shutter-speed, but it's meaningless in an electronic sensor, which is creaming off data as it pleases and we (the Customer) are unaware of how this manifest itself.
We see "faults" as in helicopter blades, but cannot be sure how this might affect SloMo. We might guess that it will create distortion, but if we add-in Golf-shot, perhaps GS is designed to avoid the "bending" since this would be very obvious - Sony could bend the frame to counteract this.
However, not being a Golfer, is GS to show the player's stance, or the movement of the club as it hits the ball....? I don't use GS as the frames are far too narrow for me and SloMo is just fine, although I'd like it sharper still, who wouldn't?

I guess we may never know - and until we have agreed a decent/replicable Test, it seems to me, we're bandying words - since we can't yet test our theories.
PS
Tests using a stopwatch/seconds-counter...ie 7-segment displays, not monitors.
....I refer to the lag that LCDs exhibit when changing state, which we see as light-dark. This is not instantaneous and I suspect any hundredths of seconds 7-segment character would blur if filmed with a true SloMo camera costing many-times our gear. When the clock stops the display will show a true time, but during the count it's anyone's guess and it's not unknown for the displays to be limited so as to preserve life...this was certainly the case in filament displays, otherwise the LSD "00-0X" would fail 1000 times more quickly than the MSD"X0-00"
...The LCD display is further covered in Post48 on the earlier Thread.....
 
Last edited:

The latest video from AVForums

Star Wars Andor, Woman King, more Star Trek 4K, Rings of Power & the latest TV, movies & 4K releases
Subscribe to our YouTube channel

Latest News

Starzplay streaming service rebrands as Lionsgate+
  • By Andy Bassett
  • Published
JVC adds Filmmaker Mode to latest D-ILA projector firmware
  • By Andy Bassett
  • Published
Bowers & Wilkins launches Px8 headphone
  • By Andy Bassett
  • Published
Sky set to launch its plug-and-play Sky Stream solution
  • By Ian Collen
  • Published
Movies Podcast: 26th September 2022
  • By Phil Hinton
  • Published

Full fat HDMI teeshirts

Support AVForums with Patreon

Top Bottom