How to make a moving item appear larger/smaller on video?

[mistersk1ttles]

There is a bloat tool when editing pictures which allows you to make something appear larger. I have noticed this same tool used in some videos. This is the only example I can find.. It is a bodybuilder with no shirt so NSFW I guess. https://www.youtube.com/watch?v=Jr9726p3doA I want to do something similar to this. Any software out there for this?

[poisondeathray]

It would normally be done in something like after effects, nukex, or hitfilm . Basically they are compositing FX type programs

The procedure analogous to doing it for a single photo. It has the bloat tool and various other warping and manipulations tools. You need to isolate the foreground as a separate layer with whatever technique - mask selections, keying etc.. then apply the effect . With a moving object, the difference is you would use motion tracking to simplify your work. The basic procedure is called "rotoscopoing" (it's basically masking per frame) and there are tricks and tips to make it easier so you don't have to do it manually each frame. If it's just very short segment or tiny movement, it might even be easier to do it by hand. You would have to be more descriptive on what exactly the situation and target is if you want more help

[mistersk1ttles]

It would normally be done in something like after effects, nukex, or hitfilm . Basically they are compositing FX type programs

The procedure analogous to doing it for a single photo. It has the bloat tool and various other warping and manipulations tools. You need to isolate the foreground as a separate layer with whatever technique - mask selections, keying etc.. then apply the effect . With a moving object, the difference is you would use motion tracking to simplify your work. The basic procedure is called "rotoscopoing" (it's basically masking per frame) and there are tricks and tips to make it easier so you don't have to do it manually each frame. If it's just very short segment or tiny movement, it might even be easier to do it by hand. You would have to be more descriptive on what exactly the situation and target is if you want more help

It would be a very short clip but the object would move across the video and back. I'll check out those softwares you linked.

[budwzr]

Hahaha, I think they just snapped a still, cut out the upper torso, did the enlargements, and overlayed it back onto the video.

[mistersk1ttles]

Hahaha, I think they just snapped a still, cut out the upper torso, did the enlargements, and overlayed it back onto the video.

Well if you look closely, the enlarged body is still moving like the original one was

[budwzr]

Hahaha, I think they just snapped a still, cut out the upper torso, did the enlargements, and overlayed it back onto the video.

Well if you look closely, the enlarged body is still moving like the original one was

Yeah, but I think they animated the cutout still to match the neck. Look closely at the neckline.

Or they could have created a "Torso Matte" that reveals the enhanced torso from behind. That would make it appear perfectly in sync.

I think you can do that pretty easily. Doing the distortions, like Bulge, or Fisheye, depends on what's available in your NLE. Remember too, everything needs to be keyframed to move together.

[mistersk1ttles]

Ok, I have installed a trial version of after effects. Never used this before, so I'll do my best to try to figure this out but any advice is appreciated.

[poisondeathray]

You need to isolate the area of interest. It's similar to "mask selections" and "layers" in photoshop or your image editor. Otherwise, if you applied the deformation such as a buldge, the background would buldge too. It doesn't matter that it moves, your mask moves with the object/ area of interest - think of it as an animated mask. In the example below, the areas in white on the mask layer represent the area of interest, and you can see it moves in tandem with the area of interest. (I guess I should have "super sized" the buldge effect more for dramatic effect, but I'm sure you get the idea)



There are different ways of isolating areas, different techniques and tools (e.g. chroma keying, luma keying, the rotobrush in AE) - but the slowest (but most accurate) is manually rotoscoping (clicking points with the pen tool). You don't have to do every frame. You usually, do the beginning, end, and middle, and it will interpolate the "inbetween" frames for. Then you can refine it on a few frames if they need refining

Making something larger is usually much easier than making something smaller. When you make something larger , you can "cover up" the background layer. But if you make something smaller, you might not have enough background textures (ie. there will be a "hole" in the background, unless you can cover up that hole with parts from other frames)

It will will probably take you a few hours to learn the basics. There are a bunch of beginner tutorials on youtube and similar sites. There should be some tutorials on making someone's head bigger for fun - that sort of thing. If you provide a better description, or sample clip I can probably point you in the right direction at which technique(s) will be a lot faster. Or if it's something really quick and easy, I can probably offer to do it for you in a few minutes - but I'm not going to spend a few hours on it

[mistersk1ttles]

You need to isolate the area of interest. It's similar to "mask selections" and "layers" in photoshop or your image editor. Otherwise, if you applied the deformation such as a buldge, the background would buldge too. It doesn't matter that it moves, your mask moves with the object/ area of interest - think of it as an animated mask. In the example below, the areas in white on the mask layer represent the area of interest, and you can see it moves in tandem with the area of interest. (I guess I should have "super sized" the buldge effect more for dramatic effect, but I'm sure you get the idea)


[Attachment 32170 - Click to enlarge]

There are different ways of isolating areas, different techniques and tools (e.g. chroma keying, luma keying, the rotobrush in AE) - but the slowest (but most accurate) is manually rotoscoping (clicking points with the pen tool). You don't have to do every frame. You usually, do the beginning, end, and middle, and it will interpolate the "inbetween" frames for. Then you can refine it on a few frames if they need refining

Making something larger is usually much easier than making something smaller. When you make something larger , you can "cover up" the background layer. But if you make something smaller, you might not have enough background textures (ie. there will be a "hole" in the background, unless you can cover up that hole with parts from other frames)

It will will probably take you a few hours to learn the basics. There are a bunch of beginner tutorials on youtube and similar sites. There should be some tutorials on making someone's head bigger for fun - that sort of thing. If you provide a better description, or sample clip I can probably point you in the right direction at which technique(s) will be a lot faster. Or if it's something really quick and easy, I can probably offer to do it for you in a few minutes - but I'm not going to spend a few hours on it

Awesome, thanks. Thanks for the offer to do it for me, but I'm trying to learn how to do it. I will follow your advice though and try it now. This is one of the random clips I'm trying it on: https://www.youtube.com/watch?v=RVKv3gLI9f8, but I'm just trying to learn the skill so there is no particular video I have to do it to yet. I am trying to start off by just making the guy's head bigger, then I'll try to mess around with the ball.

[poisondeathray]

That random one you picked is slightly more complicated as a "first try" because of the camera movement (jerky handheld) . Most people would stabilize the clip first, then do the mask isolation, apply the effect, the reapply the shake if the shake was still desired. Stabiliztion is often used hand-in-hand with rotoscoping/masking because it makes the roto work faster - there is less movement or adjustments to make per frame. But since it's a short clip, you might just try it as-is

You can do the job as accurate or "lazy" as you want. For example, in the bodybuilding clip, it was a "lazy" job, because you could see the errors in the mask around the guys' quads (parts of the background wall from previous frames were attached to his quads)

[mistersk1ttles]

That random one you picked is slightly more complicated as a "first try" because of the camera movement (jerky handheld) . Most people would stabilize the clip first, then do the mask isolation, apply the effect, the reapply the shake if the shake was still desired. Stabiliztion is often used hand-in-hand with rotoscoping/masking because it makes the roto work faster - there is less movement or adjustments to make per frame. But since it's a short clip, you might just try it as-is

You can do the job as accurate or "lazy" as you want. For example, in the bodybuilding clip, it was a "lazy" job, because you could see the errors in the mask around the guys' quads (parts of the background wall from previous frames were attached to his quads)

I see what you're saying. I'll work with a more stable clip then

[budwzr]

You might as well learn something about "Compositing". And that's a verb, not a noun. That's the "Final Yard".

[racer-x]

Gimp has a nice "I-warp" tool that can save as frames, which you can then Motion Track over video.....

[mistersk1ttles]

Gimp has a nice "I-warp" tool that can save as frames, which you can then Motion Track over video.....

I'll check this out as well

[poisondeathray]

Yes, compositing

In order to make things look more realistic - you often have to use combine multiple layers. It's really about "tricking" the eye.

For example, in that ball roll clip, if you made the ball larger, it should be casting a larger shadow - so you would need to adjust the shadow proprotionately larger. This usually can't be done all at once on the same layer with the same mask (because the shadow will "grow" in a different dimension compared to the ball)

If we take the earlier example and make it bigger, notice her breasts obscure the shadow line of the bikini on the abdomen (middle pic) - it looks less authentic especially near the end because the shadows don't match the angle of the sun.

If you extend the shadows down with another layer, it looks more real because the shadow rotates as the body rotates in relation to the sun. It's the same with reflections and other interactive optics. If something gets larger or smaller, the reflection/ shadows should adjust appropriately too



I wouldn't worry about the cosmetic stuff yet, just start with a basic mask isolation, 2 layers

[budwzr]

I see why she needed an enhancement. You left the old MetaData in the image, so I rolled it back.

[racer-x]

I see why she needed an enhancement. You left the old MetaData in the image, so I rolled it back.

I'll bet she is pissed that you found the original (natural) version......

[mistersk1ttles]

Yes, compositing

In order to make things look more realistic - you often have to use combine multiple layers. It's really about "tricking" the eye.

For example, in that ball roll clip, if you made the ball larger, it should be casting a larger shadow - so you would need to adjust the shadow proprotionately larger. This usually can't be done all at once on the same layer with the same mask (because the shadow will "grow" in a different dimension compared to the ball)

If we take the earlier example and make it bigger, notice her breasts obscure the shadow line of the bikini on the abdomen (middle pic) - it looks less authentic especially near the end because the shadows don't match the angle of the sun.

If you extend the shadows down with another layer, it looks more real because the shadow rotates as the body rotates in relation to the sun. It's the same with reflections and other interactive optics. If something gets larger or smaller, the reflection/ shadows should adjust appropriately too


[Attachment 32171 - Click to enlarge]

I wouldn't worry about the cosmetic stuff yet, just start with a basic mask isolation, 2 layers

I see what you're saying. I used the rotobrush to separate the part from the video that I want to enhance. Then, I added some bulge to that layer where I wanted it. Now, however all I see is the layer I separated and can't look at the whole video.

[poisondeathray]

I used the rotobrush to separate the part from the video that I want to enhance. Then, I added some bulge to that layer where I wanted it. Now, however all I see is the layer I separated and can't look at the whole video.

What do you mean by "can't look at the whole video" ? Was the buldge too large that it "covers up" too much of the background ? Or did you not put in a background layer ? Put a copy of the original video on the bottom layer, without the mask (just a normal copy)

[mistersk1ttles]

I used the rotobrush to separate the part from the video that I want to enhance. Then, I added some bulge to that layer where I wanted it. Now, however all I see is the layer I separated and can't look at the whole video.

What do you mean by "can't look at the whole video" ? Was the buldge too large that it "covers up" too much of the background ? Or did you not put in a background layer ? Put a copy of the original video on the bottom layer, without the mask (just a normal copy)

I didn't have a background layer, fixed it, thanks.

[poisondeathray]

I see why she needed an enhancement. You left the old MetaData in the image, so I rolled it back.

10min in the penalty box for budwzr vandalizing a national treasure. She's all natural. This is where the clip was from
https://www.youtube.com/watch?v=SIJmqmKiVD4

[poisondeathray]

Speaking of making someone's head "larger"...bahahahaha

[budwzr]

Hey, isn't he Canadien too?

[poisondeathray]

Hey, isn't he Canadien too?

You guys can have him . We'll trade you for the girl, throw in some maple syrup

[budwzr]

We'll give you that oil pipeline if you take Bieber back.

[racer-x]

Nice head transformation, I think he looks more real in the second shot...........