Music Video
Moderators: Víctor Paredes, Belgarath, slowtiger
Music Video
I am in the process of creating a music video from the song "Sky Might Fall" -KidCudi..Heavily inspired be Mike Scotts videos.
I wanted to know if anyone knows how this process is done, as far as making a character walk through different scenes.
I watched is "making of videos," and I already have a storyboard and started on some of the backgrounds.
Im fairly new to photoshop and AS but I'm deteremined. Thks.
I wanted to know if anyone knows how this process is done, as far as making a character walk through different scenes.
I watched is "making of videos," and I already have a storyboard and started on some of the backgrounds.
Im fairly new to photoshop and AS but I'm deteremined. Thks.
- hayasidist
- Posts: 3521
- Joined: Wed Feb 16, 2011 8:12 pm
- Location: Kent, England
Hi,
I don't know Mike Scott's work - but if you want a character to walk across different backgrounds a basic technique is chromakeying. Film your character against a solid (evenly lit - and lit to avoid heavy shadows) green (usually, but blue or other solid colour) background which can be removed by a video editor (e.g. Premiere - not photoshop) leaving just your character against a transparent background (character must have NO green in clothing / make-up etc). If your character is animated (AS output) then make sure that the rendered output has an alpha channel and that you keep the background correctly empty. Then composite the character against different backgrounds in something like premiere.
Things to watch out for: mismatched camera movement in the bg and fg action; floating / sliding feet; scaling and depth errors; different lighting / shadow in fg and bg...
I don't know Mike Scott's work - but if you want a character to walk across different backgrounds a basic technique is chromakeying. Film your character against a solid (evenly lit - and lit to avoid heavy shadows) green (usually, but blue or other solid colour) background which can be removed by a video editor (e.g. Premiere - not photoshop) leaving just your character against a transparent background (character must have NO green in clothing / make-up etc). If your character is animated (AS output) then make sure that the rendered output has an alpha channel and that you keep the background correctly empty. Then composite the character against different backgrounds in something like premiere.
Things to watch out for: mismatched camera movement in the bg and fg action; floating / sliding feet; scaling and depth errors; different lighting / shadow in fg and bg...
Thks. hayasidist that was very helpful.
Heres Scotts video
http://www.youtube.com/watch?v=Cjwoit91 ... r_embedded
My video wouldnt be as complex as his, I think he had a team of people help.
I'll post a screenshot of what I have tommorrow.
Heres Scotts video
http://www.youtube.com/watch?v=Cjwoit91 ... r_embedded
My video wouldnt be as complex as his, I think he had a team of people help.
I'll post a screenshot of what I have tommorrow.
- funksmaname
- Posts: 3174
- Joined: Tue May 29, 2007 11:31 am
- Location: New Zealand
- hayasidist
- Posts: 3521
- Joined: Wed Feb 16, 2011 8:12 pm
- Location: Kent, England
Re: Music Video
Hi Funks,
I think the line
===
@datpiff90: the example from Scott is all animation - multiplane -- use the AS "z" dimension for depth / parallax; consider using "depth of field" (Project Settings) to bring different z planes into focus. Plan how to break up your work by time and by depth to keep AS file sizes (and outputs) manageable and render times low. If you need to break the timeline across files and preserve continuity of action use linear not smooth interpolation on camera / layer movements...
in short Plan Plan and Plan...
good luck! (and I'm looking forward to the screenshot.)
I think the line
is a bit of a question...Datpiff90 wrote:....
I wanted to know if anyone knows how this process is done, as far as making a character walk through different scenes.
....
===
@datpiff90: the example from Scott is all animation - multiplane -- use the AS "z" dimension for depth / parallax; consider using "depth of field" (Project Settings) to bring different z planes into focus. Plan how to break up your work by time and by depth to keep AS file sizes (and outputs) manageable and render times low. If you need to break the timeline across files and preserve continuity of action use linear not smooth interpolation on camera / layer movements...
in short Plan Plan and Plan...
good luck! (and I'm looking forward to the screenshot.)
- funksmaname
- Posts: 3174
- Joined: Tue May 29, 2007 11:31 am
- Location: New Zealand
well, this is bizarre reading this ha ha.
For 'get busy living' I did that one all by my lonesome, a lot of it whilst on a trip to the UK to see a friend get married, and partying like a mad-man.
Each scene is a separate file in Anime Studio. I planned the whole thing out, beat-by-beat, and put an animatic together.
From there, good luck. Its knuckling down and drawing, drawing, drawing.
In terms of the technicalities of it, its a switch layer flicking between frames of an animation, and then put the switch layer in a group, and you can drag it across the screen.
Yes, some of the stuff was done on the Z-plane, but a lot of the time, just increasing scale of various planes and images to simulate depth.
Good luck man. Thanks for the props.
For 'get busy living' I did that one all by my lonesome, a lot of it whilst on a trip to the UK to see a friend get married, and partying like a mad-man.
Each scene is a separate file in Anime Studio. I planned the whole thing out, beat-by-beat, and put an animatic together.
From there, good luck. Its knuckling down and drawing, drawing, drawing.
In terms of the technicalities of it, its a switch layer flicking between frames of an animation, and then put the switch layer in a group, and you can drag it across the screen.
Yes, some of the stuff was done on the Z-plane, but a lot of the time, just increasing scale of various planes and images to simulate depth.
Good luck man. Thanks for the props.
- hayasidist
- Posts: 3521
- Joined: Wed Feb 16, 2011 8:12 pm
- Location: Kent, England
duh! sudden GBO 2+2=4 mikdog = mike!! duh!! Hi Mike!! Hey - glad I got the "how to" nearly right. .
anyhoo..
@datpiff90 - yeah - in AS you're not really constrained by drawing canvas size.. if you wanted to stroll along a street that was 10 "screens" wide you draw something 10 screens wide and can then track the camera across it. same idea if you wanted to show a character closeup then pull out to show the figure as a "tiny dot" in a huge landscape.... the pic you've posted is a "typical" aspect ratio, so you could scale it... but if you "just" wanted a (say) long left-to-right walk, it's better to draw it "correct" height but way wider... I've done bg images in PS up to about 10:1 aspect (widescreen = 16:9) for this reason.
you might also like to think about splitting into different layers say fg sand, mid distance rocks, bg hills and sky with different z (note the scale difference for the "far-far-away" ones...)
in short AS has an output render area - size set in Project Settings - and a drawing canvas with effectively unlimited size. You can move / scale layers to come into and out of the render area and/or move / zoom the camera to show different parts of the canvas - try ctrl-J: (View; Show Output only).
If you have one walking character you **might** find it easier to do a static walk cycle with the character where you want them "fixed" in the output window then move the scenery behind them...
but it's all back to planning, especially if you're going to composite multiple files. Good luck.
anyhoo..
@datpiff90 - yeah - in AS you're not really constrained by drawing canvas size.. if you wanted to stroll along a street that was 10 "screens" wide you draw something 10 screens wide and can then track the camera across it. same idea if you wanted to show a character closeup then pull out to show the figure as a "tiny dot" in a huge landscape.... the pic you've posted is a "typical" aspect ratio, so you could scale it... but if you "just" wanted a (say) long left-to-right walk, it's better to draw it "correct" height but way wider... I've done bg images in PS up to about 10:1 aspect (widescreen = 16:9) for this reason.
you might also like to think about splitting into different layers say fg sand, mid distance rocks, bg hills and sky with different z (note the scale difference for the "far-far-away" ones...)
in short AS has an output render area - size set in Project Settings - and a drawing canvas with effectively unlimited size. You can move / scale layers to come into and out of the render area and/or move / zoom the camera to show different parts of the canvas - try ctrl-J: (View; Show Output only).
If you have one walking character you **might** find it easier to do a static walk cycle with the character where you want them "fixed" in the output window then move the scenery behind them...
but it's all back to planning, especially if you're going to composite multiple files. Good luck.
There is an old school way of doing an infinite pan.
Imagine a car chase down a street, one which requires 100:1 width to height ratio. Whilst you could make one, it is easier to make two short backgrounds, where the left edge of background 1 physically matches the right side of Background 2. Similarly, the left edge of background 2 matches the right side of Background 1. The edge is usually made along a building edge or a drainpipe. This way, you don't have to worry about an exact one-2-one match: You don't get any image pop if you try the physically match the two ends of the background.
To shoot it, just pan along the background 1 and just before its edge comes into shot, add background 2 to the pan: When Background 1 exits shot, move it down to the end of Background 2 and start all over again.
This is sometimes called a cycle background. Essentially a very old under-camera technique which works just as well digitally in AS as it did with film.
The great attraction of this is you never see the join, even at very slow pan speeds... you never see the take-over point as its aways out -of-shot.
Rhoel
Imagine a car chase down a street, one which requires 100:1 width to height ratio. Whilst you could make one, it is easier to make two short backgrounds, where the left edge of background 1 physically matches the right side of Background 2. Similarly, the left edge of background 2 matches the right side of Background 1. The edge is usually made along a building edge or a drainpipe. This way, you don't have to worry about an exact one-2-one match: You don't get any image pop if you try the physically match the two ends of the background.
To shoot it, just pan along the background 1 and just before its edge comes into shot, add background 2 to the pan: When Background 1 exits shot, move it down to the end of Background 2 and start all over again.
This is sometimes called a cycle background. Essentially a very old under-camera technique which works just as well digitally in AS as it did with film.
The great attraction of this is you never see the join, even at very slow pan speeds... you never see the take-over point as its aways out -of-shot.
Rhoel
- hayasidist
- Posts: 3521
- Joined: Wed Feb 16, 2011 8:12 pm
- Location: Kent, England
completely agree with you Rhoel where a cyclic bg meets the need!
you can get even smarter where you have two cyclic layers - for example one for a basic set of buildings that that repeats every cycle and another (such as lampposts / mailboxes / person standing ...) that repeats every (say) 1.14 cycles - that way a duplicate pairing only comes up every 7 cycles ... and so on ...
you can get even smarter where you have two cyclic layers - for example one for a basic set of buildings that that repeats every cycle and another (such as lampposts / mailboxes / person standing ...) that repeats every (say) 1.14 cycles - that way a duplicate pairing only comes up every 7 cycles ... and so on ...
- funksmaname
- Posts: 3174
- Joined: Tue May 29, 2007 11:31 am
- Location: New Zealand
he means; get two bits of paper next to eachother
[1][2]
Now draw something that goes over the ][ part
then swap them;
[2][1]
and draw something over the ][ part again
now the joins between [1][2][1][2][1][2][1][2][1][2][1][2]
all join seamlessly.
What he describes is creating a cycle where you start with [1] on screen and pan it left till [2] is on screen, then offscreen move [1] to the other side of [2] and keep moving at the same speed until [1] is back at the beginning, and cycle.
What hayasidist is suggesting is to break up your depth into several layers on top of eachother that cycle at slightly different lengths so they don't perfectly match, creating less identical results on every cycle;
[...1...][...2...][...1...][...2...][...1...][...2...] - layer 1
[..3..][..4..][..3..][..4..][..3..][..4..][..3..][..4..] - layer 2
[.5.][.6.][.5.][.6.][.5.][.6.][.5.][.6.][.5.][.6.] - etc.
as you can see this means they don't perfectly match until a certain number of repeated cycles. It will also create depth with paralaxing as the foreground will move faster than the background.
say layer 2 is craters, you could have gaps in the cycle too or throw in a different layer for one screen depth every 5 cycles or whatever... Just experiment.
--
In other advice, if you need a lot of very specific help about things like this you should probably be experimenting instead of trying to do a full song. You have to have a good fundamental understanding of the process before you try anything so complicated! I can understand how watching Mikdogs videos can get your creative juices flowing though
[1][2]
Now draw something that goes over the ][ part
then swap them;
[2][1]
and draw something over the ][ part again
now the joins between [1][2][1][2][1][2][1][2][1][2][1][2]
all join seamlessly.
What he describes is creating a cycle where you start with [1] on screen and pan it left till [2] is on screen, then offscreen move [1] to the other side of [2] and keep moving at the same speed until [1] is back at the beginning, and cycle.
What hayasidist is suggesting is to break up your depth into several layers on top of eachother that cycle at slightly different lengths so they don't perfectly match, creating less identical results on every cycle;
[...1...][...2...][...1...][...2...][...1...][...2...] - layer 1
[..3..][..4..][..3..][..4..][..3..][..4..][..3..][..4..] - layer 2
[.5.][.6.][.5.][.6.][.5.][.6.][.5.][.6.][.5.][.6.] - etc.
as you can see this means they don't perfectly match until a certain number of repeated cycles. It will also create depth with paralaxing as the foreground will move faster than the background.
say layer 2 is craters, you could have gaps in the cycle too or throw in a different layer for one screen depth every 5 cycles or whatever... Just experiment.
--
In other advice, if you need a lot of very specific help about things like this you should probably be experimenting instead of trying to do a full song. You have to have a good fundamental understanding of the process before you try anything so complicated! I can understand how watching Mikdogs videos can get your creative juices flowing though
Here is one way I do sometimes. I made a group and put an image in it then duplicated the image 2x and flipped one. Line up the end so it matches the beginning then just cycle back. You can see the seams where the images line up I'm not sure whats up with that, I probably needed to overlap the images a tad more. I didn't spend a lot of time on this...
Last edited by sbtamu on Thu Sep 22, 2011 1:32 am, edited 1 time in total.