Typewriter title sequence...

[racer-x]

Here are a couple variations on your BirdLand over an image, along with project files if your interested. I use a picture of my late dog for the background.

[poisondeathray]

RE: creating alpha channel clip in avisynth with subtitle()

This example uses subtitle(), but you should be able to use subtitleEX() as well. The idea is to make the same text, but using 100% white as a mask (both fill and stroke outline, which is called "halo_color" in avisynth).

BlankClip(pixel_type="RGB32")
Subtitle("test", size=40)
Mask(ShowAlpha().Subtitle("test", size=40, text_color=color_white, halo_color=color_white))
Trim(0,-1)

I use ffmpeg to encode a PNG with alpha channel

ffmpeg -i input.avs -pix_fmt rgba test.png

[pippas]

There's quite a lot to get my head round now.... brain's currently in overload mode!

Thanks again both for the tips and ideas....

[jagabo]

I know the OP has already moved on, but this struck me as an interesting little problem for AviSynth. There's no audio handling here (and I don't know how to implement it yet) but here's a start:

##########################################################################
#
# worker fuction for typing()
#
##########################################################################

function _typing(clip v, string msg, int xpos, int ypos, int pos)
{
	len = strlen(msg)
	msg = leftstr(msg,pos)
	Subtitle(v, msg, x=xpos, y=ypos, text_color=$000000, halo_color=$ffffffff) # black text, no outline
}


##########################################################################
#
# simulate typing on a typewriter
#
# start_frame = frame the typing should start at
# msg = message to be typed
# fpc = frames per character (typing speed)
# xpos = x position on screen of first character
# ypos = y position on screen of first character
#
##########################################################################

function typing(clip v, int start_frame, string msg, int fpc, int xpos, int ypos)
{
	end_frame = start_frame + fpc * strlen(msg)
	Animate(start_frame, end_frame, "_typing", v,msg,xpos,ypos,0,  v,msg,xpos,ypos,strlen(msg))
}


##########################################################################

BlankClip(length=300, fps=24, color=$ffffff)
typing(20, "This is a typing demo.", 4, 20, 20)
typing(150, "Another line being typed.", 3, 20, 40)
typing(240, "THE END.", 3, 20, 60)

That's typing at a constant speed. AviSynth has a Rand() function for generating random numbers. You might somehow be able to use that to randomize the typing speed per character.

[pippas]

I know the OP has already moved on....

No.... still here.... brain hurts a bit - but still here!

Most of the 'typing' samples I've seen so far all seem to use constant speed typing.... and racer-x showed in an earlier post that, in reality, you can get away with that in most cases...

If you do want to take things a stage further, and 'sync' the appearance of the characters to an audio track of 'real' typing, then I believe you need to start with the audio track, and sync the video to that.

PDR's idea of using a mask 'reveal' in Aviutl looked very interesting. I tried using a single 'movable' rectangle as the graphic, but ran into problems when used with my original scroll type background. As you move the 'mask', you need to re-size it as it approaches the edge of the background where the colour changes, and that causes problems - because of course you are dealing with a single graphic.... so any re-sizing can cause havoc with previous keyframes (sorry, I should say 'midpoints')

I did find a slightly easier way, using separate graphic 'masks' for each letter - and using copy and paste makes that quite a quick process.
I did a quick sample, taking care to sync the audio for each character as I did originally.
When (if?) I need to repeat the exercise, I'd probably do it that way again.... It's quicker using an Avisynth script in the way I did originally.... so that's been a useful discovery.

I liked racer-x's animation of the 'Birdland' characters.... so I swapped the background, pinched his pngs, and re-synced the audio as an experiment. Next stage is to try and make my own 'alpha channel' pngs... there's a lot to learn about animation in Aviutl !...

Both you and Maverick have now put forward the idea of starting with the video..... I have to say i think 'syncing' characters in an audio track to an existing video sequence would be quite difficult.

So far, I haven't found the audio editing features in Aviutl suitably detailed enough to finding the 'sync' points accurately... but that could just be my ignorance of Aviutl.....so I tend to make and edit the audio separately in my ancient version of Cool Edit, which i'm used to...

[poisondeathray]

Another way to do a linear reveal in aviutl is to keyframe a crop value on the top layer (higher layer number in aviutl)

You can do fancy graphics/fonts in photoshop / gimp (free) / image editor of your choice, with PNG + alpha channel and import those into aviutl as well

So far, I haven't found the audio editing features in Aviutl suitably detailed enough to finding the 'sync' points accurately

Look at the audio waveform to help guide you. For something like a typing clip there will be a distinct peak

PDR's idea of using a mask 'reveal' in Aviutl looked very interesting. I tried using a single 'movable' rectangle as the graphic, but ran into problems when used with my original scroll type background. As you move the 'mask', you need to re-size it as it approaches the edge of the background where the colour changes, and that causes problems - because of course you are dealing with a single graphic.... so any re-sizing can cause havoc with previous keyframes (sorry, I should say 'midpoints')

For shapes/colors that are adjacent to each other, or there are curves - a "square" won't work. It's not precise enough. That's where real masking with bezier curves and splines are typically used. Unfortunately aviutl doesn't have those features. But you could import a custom shape and animate that.

What kind of havoc with resizing and keyframes ? - When you set keyframes (or midpoints) . You usually go the the frame before, set one to hold the previous value, then the actual frame where the keyframe value changes. Remember there are different interpolation types between keyframes. "Rectilinear" is just linear interpolation. So If you have a value of 10 on frame zero and a value of 30 on frame 2, then frame 1 would be 20 automatically without a keyframe/midpoint set. If you have it set to "teleportation", it would be 10 on frame 1, because it's essentially a "hold keyframe" - takes the same value as the last keyframe. There are a bunch of other types of curves like acceleration , deceleration etc... play with the interpolation types with a simple animation (e.g an object moving horizontally from A to B) and get a feel for what they do

(This is where autokeyframing saves a lot of time. As soon as you move or change the value, a keyframe is set. I would make a feature request to Maverick)


jagabo - It's possible to do some audio detection / audio reactive scripts in avisynth, but there some limitations. For example you can detect a certain dB threshold (e.g. a typewriter keystrike, or a big beat for audio), and if it's above a certain value, return video A , otherwise B with conditionalfilter() or conditionalselect(). But that essentially just swaps between different versions of a video, like A,B,C etc... on the same frame number.

I'm wondering if it's possible to modify or use other functions like scriptclip() to return progressive values, like N+1, N+2, so essentially the "typing" or "reveal" is automatically timed to the audio. For example, if you had a monospaced font type prepared , it could do incremental shift with animate() in x-axis "x" pixels each beat or something to do the reveal ?

[poisondeathray]

I took a youtube video of a typewriter typing (with hammer strokes and ribbon), and "removed the paper", leaving an alpha channel, so you can replace the "paper" and "text" with... whatever. Just a rough job not too precise, and only 6 different key strokes. So you can edit and reuse the strokes for different letters in different orders to confuse the viewer. Unfortunately for pippas, it's 720p29.97 FPS , not 25 FPS. It's UT Video Codec, RGBA. In most NLE's you can just drag it to the timeline. In aviutl, you need to checkmark the "import alpha channel" box for the video

The other video is a special just for racer . Notice how he spells it "Tittle" all the time in his examples, tutorials, and gif's ?

http://www.merriam-webster.com/dictionary/title

[pippas]

I took a youtube video of a typewriter typing (with hammer strokes and ribbon), and "removed the paper", leaving an alpha channel, so you can replace the "paper" and "text" with... whatever. Just a rough job not too precise, and only 6 different key strokes. So you can edit and reuse the strokes for different letters in different orders to confuse the viewer. Unfortunately for pippas, it's 720p29.97 FPS , not 25 FPS. It's UT Video Codec, RGBA. In most NLE's you can just drag it to the timeline. In aviutl, you need to checkmark the "import alpha channel" box for the video

Now that is impressive!... what did you use to make the moving paper and text?.....

I've made a 25fps copy to have a play with, and converted it to Grass Valley HQX, which I tend to use for most things. It's an excellent intraframe intermediate codec with an alpha channel, IMHO. ....and it's now free from Grass Valley.
Smaller file sizes than lossless, but still excellent quality....
( More details here if it's of any interest: http://www.grassvalley.com/docs/WhitePapers/professional/GV-4097M_HQX_Whitepaper.pdf )

[racer-x]

Well nobody ever accused me of being super intelligent anyway......

I didn't spend a lot of time on mine, so it isn't as accurate as yours, but it's good enough to test out in Aviutl. I would have to add more midpoints to make the paper pause on the key strokes....

[poisondeathray]

what did you use to make the moving paper and text?.....

I cheated and used AE, because the audio timing can be used to drive animation and effects, linked through expressions . I'm trying to figure if something similar can be done in avisynth

But you can do the same thing manually in aviutl, just takes a bit longer, using the techniques discussed earlier. Conceptually , it's the same thing.

Racer mentioned this earlier - you can use "grouping" in aviutl, so layers animate together. So you would group the text layer and paper layer together. This is called "parenting" in other progams. When the "parent" moves, the "children" move with the "parent." ie. When the paper moves, the text moves with it. Another way in aviutl is with the "group control object"

Typing sounds are very distinct. If you look at the audio waveform, there is a clear peak when a strike occurs . So even if you didn't do the timing beforehand, it's not too difficult to figure out within aviutl

If you break down this animation (at 29.97), there is a keystroke (the "hammer" covers the letter, as the letter "appears" behind it), the paper shifts left, paper shifts left again. So you would use groups of 3 keyframes in aviutl. It might be slightly different at 25fps, and slighlty different with different audio (eg. faster typist, slower typist)

[transporterfan]

These are great examples guys. I had a similar project a few years back were I had to simulate a golfball printer.
It looked something like this: https://www.youtube.com/watch?v=VF1RiTyCeUQ
I eventually used Flash and ActionScript.

To the OP. An easier solution might be to use something like this (assuming you can type accurately and at speed):
http://www.nolad.com/vt/ and use screen capture software to create a video from your desktop.

[racer-x]

OK, this one is better, I added some midpoints (aka key-frames) and some texture to the paper. I also added a VH logo as a letter head.

It was done all in Aviutl, I can post the project file if anyone is interested.

[poisondeathray]

I've made a 25fps copy to have a play with, and converted it to Grass Valley HQX, which I tend to use for most things. It's an excellent intraframe intermediate codec with an alpha channel, IMHO. ....and it's now free from Grass Valley.
Smaller file sizes than lossless, but still excellent quality....
( More details here if it's of any interest: http://www.grassvalley.com/docs/WhitePapers/professional/GV-4097M_HQX_Whitepaper.pdf )

Maybe this is a good time to demonstrate or clarify "luma matte" or "mask", or how it's done in aviutl (or at least 1 way of doing it)

Whether or not if you want "lossless" or "lossy", there are some situations when you want a separate alpha channel instead of an embedded alpha channel. In this example, the filesize is tiny and uses lossy temporal compression. "base.mp4" is the same UT video, but without alpha channel, and compressed with AVC. "alphaYV12.mp4" is the separated alpha channel compressed with AVC as well - ie the white/black clip that shows what is transparent. 100% white is fully transparent. 100% black is fully opaque. "shades" in between have variable transparency. ("Transparent" in the sense that the layer in question will show 100% of those areas that 100% white. And 100% black areas will not be shown, thus become "transparent" themselves in the main composition layer stack order, so the Background layer (or the next adjacent layer in decreasing layer number) will show through those black areas)


In scene 1
Layer 1 - "alphaYV12.mp4"

In the root scene
Layer 1 - background clip of anything
Layer 2 - "base.mp4"

On "base.mp4", apply the mask effect, in the drop down menu => select from scene => scene 1. Click Match with original size

As you can see, it's practically the same thing, but tiny in filesize

This luma matte or mask is the same concept when used in avisynth with overlay() with mask , or mt_merge() with mask

There are times when you might want access to manipulate the alpha channel separately, for example animation and reveals. Or you might want to clean up a green screen key, etc...

[racer-x]

Or just use this one encoded with a blue screen that can be easily chroma-keyed out in most video editors. This one is 1080p @ 24 fps with audio track.

It's only 500 kb.

[pippas]

An easier solution might be to use something like this (assuming you can type accurately and at speed):

There's the problem.. I can't!... ... plus the fact that attempting to do it this way has shown me a lot more about the power of Aviutl than i ever would have discovered otherwise... and that's been really interesting...

... I can post the project file if anyone is interested.

Yes please...

Maybe this is a good time to demonstrate or clarify "luma matte" or "mask", or how it's done in aviutl

Thanks for those suggestions... I shall try and get my head round some of the concepts. At the moment, i can't even work out how to use alpha masks in Avisynth to make a top layer fully opaque, instead of 'mixed' with the background... so I have some way to go, even on basics, at present....

all interesting stuff though...

[poisondeathray]

Maybe this is a good time to demonstrate or clarify "luma matte" or "mask", or how it's done in aviutl

Thanks for those suggestions... I shall try and get my head round some of the concepts. At the moment, i can't even work out how to use alpha masks in Avisynth to make a top layer fully opaque, instead of 'mixed' with the background... so I have some way to go, even on basics, at present....

A top layer fully opaque wouldn't need anything special like mask or alpha channel, etc... You would see the top layer directly.

It's really the same thing in avisynth. The layer order is "normal" in avisynth. Things on the bottom go on the bottom. Layers on top "cover up" layers on the bottom. This is the way most programs do it, including photoshop, gimp, NLE's etc... Note the order is reversed in aviutl, so it might not be as intuitive

Here is a simple example using Overlay() . For these purposes, we are working in RGB (mt_merge only works in YUV), and still images with non animated masks

background=ImageSource("cancun.png")
foreground=ImageSource("mona.png")
mymask=ImageSource("mask.png")

Overlay(background, foreground, mask=mymask)

You can think of it like you're putting "mona" onto "cancun", and the "mask" has holes punched in that allow you to see through areas in white on the mask



cancun.png


mona.png


mask.png


Final Result

[racer-x]

Here are the project files. Inside the zip is a better quality blue-chromakey typewriter. You can see in the project how to key it out.

I went this route to minimize file size...

[pippas]

.... and still images with non animated masks

Thanks for those details PDR ... I think I have the concept in my head now ... it was an animated mask I was trying to create,,, without thinking the process through correctly....

Here are the project files. Inside the zip is a better quality blue-chromakey typewriter. You can see in the project how to key it out.

Thanks racer..... I've already started to pull the project file apart, to see how and what you did........probably the quickest way to learn, I think?


Thanks again both... plenty to keep me occupied for the moment I think!

[transporterfan]

I guess it all depends on how much information you want to display on screen at one time.
Reading the following would give me brain fatigue, but it works ok with the original concept (the scroll).
This was made using Flash, but I am sure racer can guide you through it using aviutl.

[poisondeathray]

.... and still images with non animated masks

Thanks for those details PDR ... I think I have the concept in my head now ... it was an animated mask I was trying to create,,, without thinking the process through correctly....

Yes, animated or static, it's the same concept.

That was the point of the earlier example, instead of animating assets, text etc... you can animate the masks. Recalll it can be difficult to animate the assets in a NLE. So you have your objects/text arranged and spaced beforehand in an image editor and reveal them with animated masks. It doesn't have to be a linear reveal - that opens up many more creative options. You can use multiple layers, multiple reveals - that's the power of having multiple tracks compare to something like wmm

[poisondeathray]

I redid the animation in aviutl. As you've seen already, there are slightly different ways to do this.
(I apologize if I mixup "keyframe" vs. "midpoint" - I used them interchangably here)

In this zip file, is the project file, background, and the luma matte for typewriter1.mp4. You also need to download typewriter1.mp4 from before. That is used for the audio, and keystrike animation.

In this version, I did the text type reveal animation by animating the crop value. (in aviutl crop called the "clipping" effect), with text generated in aviutl. The font is called "Travelling Typewriter" and is free.
www.dafont.com/traveling-typewriter.font

Note the text reveal is separate from the paper moving. So I did the text reveal first timed to the strikes, then parented it to the paper, and animated the paper by grouping them.

I typed the entire text out first. "AVIUTL was used for this typing video". I used the typewriter foreground as a guide for placement. Remember the lower opacity trick ("clearness" in aviutl) to see "through" to see what you are doing

When animating/ setting keyframes in aviutl, set the interpolation type for that parameter FIRST. "midpoints" aren't activated, until you hit the button to set the interpolation. So you would set it to "teleportation" for the text reveal, because each midpoint is discrete (you don't want interpolation between midpoints) . Move the playhead to the frame of interest, hit "P", then adjust the clipping value until the 1st letter is revealed. Move the playhead to the next keystrike, hit "P", adjust the value to reveal the next letter, repeat...So that's how individual letters are "revealed" one at a time with this method. The timing you have to do manually (ie. when a letter is revealed coinciding with a keystrike), but if you figured it out with an audio program , it should be quicker. I just re-used the same timing as the example before, "typewriter1.mp4"

Next you align the paper and text, to how you want the text to appear on the paper when fully revealed (centered, etc..). Parent them together. So from now on they move together. In this example I used the group control object (right click on the timeline => new media object => group control object). Set the Target ly# to 2 (number of target layers). This means 2 layers below the object. So far you have:

Layer 1 Group Contol Object
Layer 2 Paper BG
Layer 3 Text

(Possibly Layer 4+5), Video & Audio from "typewriter1.mp4" used as a "guide" . Remember you can deactivate a layer anytime by hitting the layer button.


You animate paper and text, by keyframing the group control object. You already have the keyframes in position on the correct frames marked on the "Type" layer. So use that as a guide to save time. Dont forget to activate by pusing the "X" button in the group control, and setting "teleportation". Then jump right to the next midpoint position, push "P", adjust the X parameter of the group control object, move to the next keyframe, push P, adjust the X value, etc....

Next is the 2 frame stutter cycle. Recall I said after the strike, paper shifts left, paper shifts left? So instead of setting keyframe interpolation to teleportation, set it now to "rectilinear", for linear interpolation for the group control object for parameter X. So on the 1st strike midpoint, advance 2 frames, push P, adjust the value to the same as the next keyframe on the group control object for "X" . (you can see the value on the right column, where as the left column is the current value, so copy the value of the right to the left. Remember you can type values.) This way, you reduce the amount of work and "midpoints" set - because with linear interpolation, it will automatically animate between the original "midpoint" and the one you just set. Since the value is the same on that 2nd midpoint and the next midpoint, there is no movement (no interpolation) and that's what you want

( It's too bad you can just copy & paste keyframes, or there is no keyboard shortcut for "jump to next keyframe". The "<<" and ">>" in buttons in the GUI will jump to next keyframe , but not as fast as keyboard shortcuts IMO. Or better keyframe access like a dope sheet, or curve editor. Those would be useful features for aviutl as well as autokeyframing)

Continue by jumping to the next keyframe (push the ">>" button in the GUI), advance 2 frames, push P, copy the right hand value to the left, repeat...etc...

Then you animate the carriage return. You should know how to do that by now, by setting keyframes.

Next is overlaying the typewriter strikes. I just reused the same timing, and I included an alpha channel for this project. So just use typewriter1.mp4 from before, and the matching alpha that is include in this zip "timedtype_showalphaYV12.mp4" . There were instructions in an earlier post, about using "mask" and linking it to "scene1". When doing this for your project (different fps) or different audio, you take the original alpha channel clip (the ut video version) with 6 strikes and edit them creatively so you mix up the strikes. You basically mix & match multiple edited copies, and extending by placing duplicate frames where there is no activity. If this doesn't make sense how to do this just ask.

And the very last thing are embellishment things, like motion blur, color correction, adding other effects like the Letterhead in Racer's example. You could have added the typewriter roller pin (instead of "black") etc...I didn't do any of those. Aviutl has a motion blur filter, but I didn't like the look.

[racer-x]

Yes, more than one way to skin a cat, bravo....

I didn't even know about the clipping filter, will have to experiment with it.

Do you still have the original typewriter video? Could you post an image before you keyed out the paper, I'd like to use the background if possible.

[poisondeathray]

Do you still have the original typewriter video? Could you post an image before you keyed out the paper, I'd like to use the background if possible.

There is no background except "paper" . It's from a youtube video for selling an after effects template. I didn't want to buy the template, so I just used the preview to "rip" the typewriter

www.youtube.com/watch?v=Sq49-QjpM-k


What kind of background were you looking for ?

[poisondeathray]

.... and still images with non animated masks

Thanks for those details PDR ... I think I have the concept in my head now ... it was an animated mask I was trying to create,,, without thinking the process through correctly....

Here is an example of an animated mask reveal, generated in Aviutl, for a "writing" project
https://forum.videohelp.com/threads/366724-Aviutl-Tips-Tricks-and-Support-thread?p=2368...=1#post2368025

[pippas]

Looking through the posts here again - and at the tips and ideas suggested has made me realise just how little I do know about masks.. and especially animated masks.
I can see that Aviutl has some very sophisticated options which allow you to perform functions only otherwise found in some pretty expensive alternatives (I'm thinking After Effects for example....)
But it does have a pretty steep learning curve.. and if your knowledge of the basics of what you're trying do is not very good, then combining the two at the same time can be quite daunting

So I took a step back, and tried out a simple animated mask project using software I am already familiar with, so that all I'm trying to do is understand the masking concepts... not trying to learn the software at the same time
I also thought I'd move away from the typing idea for the moment, to have a 'clear' start....

I found this post on another forum: http://www.videoforums.co.uk/threads/48419-Making-an-intro?p=211169#post211169
and thought it might be an idea to have a go at that.....

I stole the background from Racer-x's 'Cloud' file (...hope that's OK Racer??...)
I took the logo and made a transparent png version with my (very!) old version of Photoshop -- (I probably should have used Gimp) --and a mask from that png in the same way.
I then imported those files into a rather old fashioned program called Canopus 'Imaginate' (no longer available sadly) and used that to animate both the png and it's mask. (Probably not the best choice, but it deals with 3D interpolation of still images pretty well... and I have some idea of how to use it!)
Exporting from that as Canopus HQX video files with an alpha channel.
The audio track was made separately, and then synced to the video (previously I'd always synced video to audio..... it was Maverick who suggested this way round)

The final simple sequence was put together in Avisynth and Virtualdub (copy attached to this post)

Now I imagine that, armed with the same transparent .png and the attendant mask, there is a more elegant way to animate it using Aviutl. How easy it is to adjust things like 3d 'camera' movement and 'spline' timings of the image motion etc I have no idea?....

I'm guessing that some of the details on the 'moving paper' typing projects in this thread would give some idea... i must have another look at the previous posts.

Problem is - especially when you get a bit older like me - there is always a tendency to use what you are familiar with, rather than venture into learning something new.

Bad habit!....

[racer-x]

Use whatever tools you want to get the job done. People are generally only interested in the outcome, not the journey. I don't mind anyone using that clip, that's why I posted it.

[poisondeathray]

^ Exactly what racer said. That was the main point. There are many ways to get the same result. Use all the tools at your disposal or whatever you're comfortable with

BTW, this Super Nerd HQ guy posted at videohep too

Now I imagine that, armed with the same transparent .png and the attendant mask, there is a more elegant way to animate it using Aviutl. How easy it is to adjust things like 3d 'camera' movement and 'spline' timings of the image motion etc I have no idea?....

For a transparent png, the mask is embedded in the png itself. The mask is the alpha channel. So if you animate the png, the mask moves with it.

As for animation - it depends on what you want to do. You can animate objects (layers) or cameras or both. But aviutl isn't a "true" 3d program. It has z-space, but only uses 2d geometry in layers. It's often called "2.5d" because if you rotate around a layer, it will look "flat" if you catch it on edge

I'm guessing that some of the details on the 'moving paper' typing projects in this thread would give some idea... i must have another look at the previous posts.

That's just simple 2d keyframing in the x-axis. You already know how to do this. The only difference was "parenting" using group control object , so 2 layers move together in tandem

[pippas]

Just a footnote to this thread......
My thanks - especially to PDR and racer-x - for all the tips and suggestions.

Following up on PDRs advice on masking, I made some masks from a few frames 'cloned' from the you tube typewriter sequence that he described in post #37.

Using that, and guided by the tips received here, I had a go at trying to create something similar to that video. I used Canopus Imaginate to animate the carriage and paper, and AviSynth for the text and to assemble the 'keystrike' clips. The masks I made in Gimp

I was trying to be a bit too ambitious, I think....I was keen to try and copy the actual 'mechanics' of the carriage, as the makers of the template do in (some) of their promo video, but that meant creating masks to include the bits you can 'see through'. And I wasn't really careful enough with the mask 'edges' next to the paper.

The other problem with doing it that way is the inability to 'group' the text with the paper - a technique which both PDR and racer-x had described when using AviUtl.

That part is more difficult in Avisynth....

I've attached a 15 second clip of my latest effort... and I do now have those scripts for use in my originally planned project... but I think I should now look again at trying to do this with AviUtl. I think the implementation using that is a better way of doing this...

Thanks again for all the help.....

[pippas]

I realise I've had a Homer Simpson moment....by creating masks that didn't need to be there ...D'oh!.....

When I created the 'key strike' frames in Gimp I added an alpha channel. So when I imported those pngs into Virtualdub to create the video clips, and exported them as Grass Valley HQX, those clips came with the Alpha channel intact. So I had no need to create the masks as well....silly mistake.

So I has one last go.. this time doing everything with Freeware (just Gimp, Vdub and Avisynth) and I think this time I'm finally getting something like I originally hoped for...

It's been an interesting learning curve..... thanks again for the tips..