Can any free software do this? Isolate dancing guy and put in new video

[ShakeIt]

This is what I need to do... I want to edit a guy that's dancing in a video. I want to isolate just him and for the length of time that he's dancing. I want to take him dancing and put it in another video. Is there any free software that I can use to do this?

[poisondeathray]

blender, but there is a learning curve

search for tutorials on blender "rotoscoping" and "compositing"

[MaverickTse]

I thought I've seen a similar topic on this forum not long ago...
Anyway, how easy can you isolate "the dancer" depends on the background.
For a uniform single-color background, a simple "chroma keying" may do(many NLE can do this). Otherwise, rotoscoping like what PDR suggested.

[ShakeIt]

Thanks for your suggestions. I have noticed that Adobe is giving away a free version of their After Effects software. I'm not sure which version it is. But, I suspect that it is old. I have their free version of Photoshop CS2 and it is old. Does anyone know if the free version of After Effects can do what I'm looking to do?

[poisondeathray]

Absolutely After effects can. Even old versions can. It's probably the tool of choice for many users for basic rotoscoping tasks. If you have some complex/difficult aspects in the footage, then there might be more appropriate tools for this task (but none of them free)

[poisondeathray]

I haven't seen a free version of AE, do you have a link to it somewhere ? All I see is "free trial"

The difference between newer versions of AE, is they have better motion tracking, better mask manipulation tools, faster ways of doing basic roto jobs like rotobrush

Rotoscoping is tedious, tedious work. It's one of the most hated tasks by people that do FX work. There are techniques and tips that people use to make the job easier, I suggest you have a look at some basic tutorials first


Another free one (GNU/ GPL) that I've heard about, but never really tried is claxa
https://code.google.com/p/claxa/

[boardbuilder]

CineGobs Keyer is free and it might suit your needs. It will do chroma-keying and has rotoscoping tools as well. As I recall, I used it years ago and it's easy to learn, but jobs like you describe usually are time consuming and tedious.

[ShakeIt]

I think this is the link to the free stuff from Adobe. You need to sign up and they will give you access to a lot of older software. https://www.adobe.com/cfusion/entitlement/index.cfm?loc=en&e=cs2_downloads

[ShakeIt]

CineGobs Keyer is free and it might suit your needs. It will do chroma-keying and has rotoscoping tools as well. As I recall, I used it years ago and it's easy to learn, but jobs like you describe usually are time consuming and tedious.

The time consuming part is the bad thing. Is there any paid software that will do this quickly?

[davexnet]

The whole of CS2 is available from Adobe. They turned off the activation server, and provided a download link and serial number.
They made the whole thing public (once you logon) but warn they didn't really intend for non-licensed users to acquire the software.
The whole thing is a little weird.
https://www.adobe.com/cfusion/entitlement/index.cfm?e=cs2_downloads#

[poisondeathray]

The time consuming part is the bad thing. Is there any paid software that will do this quickly?

It's not just the software - an experienced rotoscoper (yes , there are people that do this as their primary job), will do this 10-20x faster than someone new at this. And factor in the time for you to learn how to do this, learning curve

Quickly vs. quality - do you want bad chunky edges ? or a "good job" ?

Rotoscoping is only part of the job. The other part is the compositing - that's also tricky and probably more difficult. If you don't do it properly , it will look obviously out of place like you cut & pasted like some kindergarten kid stuck it on - unless that was the desired "look" or goal ?

It depends on what your expectations are, and what you want in the final product - you could also look at hiring someone to do it for you

[ShakeIt]

The time consuming part is the bad thing. Is there any paid software that will do this quickly?

It's not just the software - an experienced rotoscoper (yes , there are people that do this as their primary job), will do this 10-20x faster than someone new at this. And factor in the time for you to learn how to do this, learning curve

Quickly vs. quality - do you want bad chunky edges ? or a "good job" ?

Rotoscoping is only part of the job. The other part is the compositing - that's also tricky and probably more difficult. If you don't do it properly , it will look obviously out of place like you cut & pasted like some kindergarten kid stuck it on - unless that was the desired "look" or goal ?

It depends on what your expectations are, and what you want in the final product - you could also look at hiring someone to do it for you

How long would it take for an experienced person to do something like this? To be honest, it was just an idea that I had for something funny. The idea wasn't for someone dancing, but that was the best way I could describe it. Like a person doing something funny and putting it in another video.

[poisondeathray]

How long would it take for an experienced person to do something like this? To be honest, it was just an idea that I had for something funny. The idea wasn't for someone dancing, but that was the best way I could describe it. Like a person doing something funny and putting it in another video.

Impossible to answer without details

Length of source footage, quality of footage, compression artifacts ; same with destination footage . Obviously 3minutes will take longer to roto than 10 sec clip. Specifics like how clear is the person, e.g. are any objects moving in front obscuring the talent, lighting changes, field of view /optics changes in the shot - all affect how and if you are able to composite to footage in effectively . How is the 1st footage set up compared to the 2nd footage. Things like differences in lighting, grading , camera views, depth of field, all affect if/how it can be done, and how long it would take

Expectations - how good of a job did you want done ? Hollywood studio level , as if the 1st footage really was in the 2nd footage / seamless ? Or like some kindergarten kid pasted it on ? That 2nd one version is what parody type funny skits are like , or funny web videos . A funny rough obviously pasted on might be very quickly done. A studio level production might take weeks with a team of a dozen rotoscopers and compositors

[boardbuilder]

If you haven't shot the funny video, which you want to overlay onto the other video, just shoot the funny stuff in front of a green screen and use chroma keying. Almost any video editing or effects software will do this and it shouldn't take that long. The quality of the result will be determined by many factors including subject lighting, green-screen lighting and quality, camera quality etc. There are many tutorials on this. Once you get the hang of it, you can do all kinds of creative things...

[MaverickTse]

@ShakeIt needs to post a representative screenshot showing the dancer and the background for concrete advice.
If you created that dancing video, perhaps retaking the shot in front of a green/blue screen will be faster

Just for the sake of "fun", you don't need professional result.
The AFV(America's Funniest Home Video) often do outright simple "head replacement" than can be funny enough

Or use some other FX to cover up...
1) do a rough color-keying or rotoscoping
2) Add border-blur and a thick colored borderline
3) Try with different Blend mode (multiply, overlay, screen...)

[ShakeIt]

This is what I thought about doing. I'm not looking for Hollywood quality by a long shot. I'm just looking to insert something into a video that doesn't belong. This is the process that I thought of so far. Take the video and make the clip I want in Virtual Dub. Convert it to a GIF and upload it to an online GIF editor. Isolate and cut out the person I want. Then figure some way to take this GIF and enter it into a video. Does anyone know how I could enter the GIF into a video? To me this seems like the route with less work and much less skill involved. I'm not looking for anything new professional quality. I'm just looking to have a laugh or two.

[poisondeathray]

You usually don't want to use GIF in your workflow (256 colors), unless that's the effect you're going for

But if you did for some reason, you can use a free multi track video editor like aviutl to overlay the gif

Did you mean something cheesy/funny effect like "This is Sparta" a few years ago ?
http://www.youtube.com/watch?v=w2JFMd73AzQ

[smrpix]

Take the video and make the clip I want in Virtual Dub. Convert it to a GIF and upload it to an online GIF editor. Isolate and cut out the person I want. Then figure some way to take this GIF and enter it into a video. Does anyone know how I could enter the GIF into a video? To me this seems like the route with less work and much less skill involved.

This is a terrible way to go about this. What makes you think this makes sense?

You have been given a lot of very good advice, but without knowing the specifics of your sources, as MaverickTse says, no one can give you specific next steps

[ShakeIt]

You usually don't want to use GIF in your workflow (256 colors), unless that's the effect you're going for

But if you did for some reason, you can use a free multi track video editor like aviutl to overlay the gif

Did you mean something cheesy/funny effect like "This is Sparta" a few years ago ?
http://www.youtube.com/watch?v=w2JFMd73AzQ

This maybe what I'm looking for. I'm sorry that my terminology might be off. I know extremely little about video editing. I'm just looking to put a person from one video into another. The quality of it is the least of my concern. I'm not doing it to make any money and I'm not even looking to do a good job at it.

[poisondeathray]

But you see the difference is that example is a essentially still image (or a few still images) of a head pasted / keyframed . The reason that is 1000x more easier is the source image is primarily a still image (ie there is a lot less rotowork), not someone moving or dancing

Conversely, someone dancing requires you to roto out each frame . Do you know how to do masks in photoshop, where you click the pen tool (dots around the area of interest) . Imagine doing that for 1000's of frames. That's why it's tedious work. There are tools that help the workflow, like motion tracking, keyframe interpolation (you don't have to do every frame), rotobrush, etc... but still it's time consuming . That's why things like green screen / chroma keying make this a lot easier if you plan ahead of time and are doing the shooting yourself

You've got to be more specific, provide more details . Do you already have a source video in mind? For both the source and destination ? If so, post a clip or do a better job describing what you want

[ShakeIt]

But you see the difference is that example is a essentially still image (or a few still images) of a head pasted / keyframed . The reason that is 1000x more easier is the source image is primarily a still image (ie there is a lot less rotowork), not someone moving or dancing

Conversely, someone dancing requires you to roto out each frame . Do you know how to do masks in photoshop, where you click the pen tool (dots around the area of interest) . Imagine doing that for 1000's of frames. That's why it's tedious work. There are tools that help the workflow, like motion tracking, keyframe interpolation (you don't have to do every frame), rotobrush, etc... but still it's time consuming . That's why things like green screen / chroma keying make this a lot easier if you plan ahead of time and are doing the shooting yourself

You've got to be more specific, provide more details . Do you already have a source video in mind? For both the source and destination ? If so, post a clip or do a better job describing what you want

I have no particular videos in mind. What I was thinking was taking one person and putting them into historical situations. Like a guy playing guitar for example while someone is giving a speech. No particular video in mind or speech in mind. I understand what you're saying. I guess what I'm saying is I don't have any particular source videos in mind. Just an idea to stick one guy from one video into another video.

[poisondeathray]

I have no particular videos in mind. What I was thinking was taking one person and putting them into historical situations. Like a guy playing guitar for example while someone is giving a speech. No particular video in mind or speech in mind. I understand what you're saying. I guess what I'm saying is I don't have any particular source videos in mind. Just an idea to stick one guy from one video into another video.

Got it. But you ask a general question , you get a general answer... The answer is still the same, blender for free . After Effects (and if the free CS2 version is still available , that would work too)

But there are a lot of difficulties and complexities that might not be apparent to you right now. Take that example of guitar guy and speech guy. If they are stationary in the same spots, easier to do. But what happens if guitar guy is dancing around the stage ? What if he goes "behind" the speech guy - well in that 2nd case, you'd have to roto out the 2nd video of the speech guy as well, plus any other objects that are supposed to be in front of the guitar guy. This layer concept of foreground and background layers is analogous to layers in photoshop . That's only in "2d" . What happens if the camera is moving as well ? rotating around the guitar guy and/or speech guy ? That gets a LOT more complicated because you now have "3d" or different camera angles and views with perspective, and possibly different camera movements like tracking, zooming, panning , etc... . You'd have to "matchmove" or 3d motion track to match the 3d characteristics of the shots now. By using a lot of careful planning, you might choose a straight on shot , with no camera movement, with roughly the same angles in the source & destination videos, with similar camera angles and scene lighting to minimize the trouble you're going to face later - at least to start with until you get used to the basics

[ShakeIt]

I have no particular videos in mind. What I was thinking was taking one person and putting them into historical situations. Like a guy playing guitar for example while someone is giving a speech. No particular video in mind or speech in mind. I understand what you're saying. I guess what I'm saying is I don't have any particular source videos in mind. Just an idea to stick one guy from one video into another video.

Got it. But you ask a general question , you get a general answer... The answer is still the same, blender for free . After Effects (and if the free CS2 version is still available , that would work too)

But there are a lot of difficulties and complexities that might not be apparent to you right now. Take that example of guitar guy and speech guy. If they are stationary in the same spots, easier to do. But what happens if guitar guy is dancing around the stage ? What if he goes "behind" the speech guy - well in that 2nd case, you'd have to roto out the 2nd video of the speech guy as well, plus any other objects that are supposed to be in front of the guitar guy. This layer concept of foreground and background layers is analogous to layers in photoshop . That's only in "2d" . What happens if the camera is moving as well ? rotating around the guitar guy and/or speech guy ? That gets a LOT more complicated because you now have "3d" or different camera angles and views with perspective, and possibly different camera movements like tracking, zooming, panning , etc... . You'd have to "matchmove" or 3d motion track to match the 3d characteristics of the shots now. By using a lot of careful planning, you might choose a straight on shot , with no camera movement, with roughly the same angles in the source & destination videos, with similar camera angles and scene lighting to minimize the trouble you're going to face later - at least to start with until you get used to the basics

I understand what you're saying. You are coming from this as someone who sounds very professional. Maybe even someone that does this all for a living. I'm just a guy that was kind of thinking out loud wondering if such a thing was possible. A guy who has no idea what he's doing at all. I never thought about those things that you brought up. I now see how complicated all of this stuff really can be. You've given me a whole new understanding of things. The next time I see something like this on TV, I'll have a little understanding of how difficult it is.

[poisondeathray]

I don't want to discourage you; you should try it out on some "simple" shots first. Pick your battles. Plan your composite where the singing/dancing guy doesn't go behind any objects (ie. always stays in front), and it will make life easier

When they do stuff on TV or movies - they shoot on green screen, with mechanized camera rigs and clean plates. It's prepared and planned well in advanced to make things much easier later for VFX work . They try not to do what you're doing (or at least minimize it as much as possible)