Aviutl Tips, Tricks and Support thread

[tyee]

Thanks racer-x. I will play around. That is typically what I would want to do but regarding this question, after I hit the first split I do not want to delete any part, just put a transition at the split point. Is that possible?

[racer-x]

Thanks racer-x. I will play around. That is typically what I would want to do but regarding this question, after I hit the first split I do not want to delete any part, just put a transition at the split point. Is that possible?

Looks like you'll have to do it like this:

[tyee]

Ok, I will try it. There must be a setting to split audio at the same time as video? Have you run across it?

Trying to figure out how to render only a section of the timeline. Can't seem to select a portion of it, like just around this transition to render. Ok, my mistake, use the left right brackets.

[poisondeathray]

EDIT: sorry doesn't work properly . I was about to suggest a workaround to group the tracks, then right click split - but it seems buggered - you have to ungroup to physically move them

[poisondeathray]

EDIT: sorry doesn't work properly . I was about to suggest a workaround to group the tracks, then right click split - but it seems buggered - you have to ungroup to physically move them

Actually it does work, you have to enable it in the advanced editing preferences

You can set it split to where the playhead is too instead of the cursor

Right click the adv editng background window and select preferencs



Sorry more complete instructions:
To group tracks select them by CTRL+Clicking them, then right click => group
In the adv editing preferences, I enabled "make objects into separate groups when splitting"

For my own preferences, I enabled "grouping multiple objects when importing by drop and drag", and "midpoint addition and splitting always carry out at current frame"

So rightclick + s will split at the current time indicator now and the split sections are linked and movable



IMO, the grouping option for D &D should have been enabled by default configuration, it seems like the more common usage scenario to have A&V linked (or maybe it was already and I buggered my preferences up)

[tyee]

I set all the same preferences as you and now I can't split at all. Strange. If I click either the video or audio track they both highlight together but "S" does nothing.

Update - OK, I see now. The keystroke "S" does not work now, you have to Rclick and select split. Hmm...interesting. Why does S get disabled? But wait, I moved the latest split back up to tracks 1&2 and now they are not grouped anymore. My other split sections in tracks 1 & 2 are still grouped but not this latest. Why? So do I have to manually group every split when I move them to other tracks to do a transition?

Ok, looks like if I want to do a transition that overlaps video I have to have those videos in separate tracks. If I don't want overlap in the two videos, for example just splitting and deleting a section, I can use only one video track and add the transition below that track.

[poisondeathray]

But wait, I moved the latest split back up to tracks 1&2 and now they are not grouped anymore. My other split sections in tracks 1 & 2 are still grouped but not this latest.

They remain grouped for me if I move them to other tracks, and I can still edit them and both A&V are affected

[MaverickTse]

Is there a complete manual available? I see all the links on MaverickTse's site but one document with all the pages would be nice?

I'm trying to find out how to split one video track at one point and create a cross fade (or dissolve) to the part that was split. This is all on the same video track. I'm a little used to Vegas so when I tried the split, I can't move the split parts apart, they all move together. What am I doing wrong. I can't find this in any of the help pages.

I am already working on such a book but in Chinese.(may get published at the end of this year or no later than Feb next year.)
If I were to make a reference book on AV-editing basics + AviUtl in English, how much would you pay for it?

[MaverickTse]

Every source is different, but in my test source, MK-I gets better results than MK-II.

I get you point.
Mode 1 and 2 in MK-II is somewhat shaky, while DF=1 or CamShift in MK-I do stick tightly to the object IF it can acquire the target.

but in many situation, the MK-I will fail completely...(e.g. an object falling down crossing the sky-land boundary. Tracking will likely fail in MK-I when the obj image cross the boundary). You can also try tracking the bike driver.... a very challenging target to track... and should fail very soon if you use MK-I

If some researcher or university students working on Computer Vision can help FIXING the TRACKING MODULE in OpenCV3.0 (on github, Itzee), we can expect a lot from the TLD module (Method4, which is now broken)

[MaverickTse]

[QUOTE=poisondeathray;2356777]

EDIT: sorry doesn't work properly . I was about to suggest a workaround to group


IMO, the grouping option for D &D should have been enabled by default configuration, it seems like the more common usage scenario to have A&V linked (or maybe it was already and I buggered my preferences up)

Your voice is heard, and I'd like to know how others think about on this.

[tyee]

I agree! Grouping Enabled.

For the book, it would depend on how many pages I guess. Once I can get the basic workflow techniques down I could do most basic editing. For a lot of what you and racer-x have been discussing (lots of different filters) a book would be better for reference in the future if I went away for months then came back to editing.

[MaverickTse]

I agree! Grouping Enabled.

For the book, it would depend on how many pages I guess. Once I can get the basic workflow techniques down I could do most basic editing. For a lot of what you and racer-x have been discussing (lots of different filters) a book would be better for reference in the future if I went away for months then came back to editing.

In my current book, B5 size with 250-300 pages, BW printing+DVD(with demo videos for the more difficult operations+sample project/materials), includes:
- Introduction of some selected freewares
- Basic terminologies, ideas and workflow in video editing (for common layman)
- Common tasks using AviUtl Base program:
> Trim, Paste, Append, Crop
> Deinterlace (Auto, AFS and NNEDI3)
> Color correction
> Resize (minor and huge amount) and clipping
> Denoise
> Hard-coded subtitle removal
> Logo removal
> Chapter addition
> Use of Profile

- Adv. Editing
> UI introduction and basic operation
> Basic animation
> Adv. animation
> FX highlights
- -Puzzle and Shatter
- -BezierCurveT
- -Weather simulation and derived effects
- -Mirror
- -Noise
- -Displacement map
- -Polar transform
- -Motion tracking
> A brief description of ALL internal FX
("internal" means the vanilla Adv. Editing, Not those FX in the ./Script folder)
> Compositing examples

- Using MikuMikuDance for Background and Accessory materials

- Uploading to YouTube and NicoNico

- Other common task
> Image stabilization with FFmpeg
> 30fps -> 60fps conversion with SVP
> Subtitle Import and Export

-Advanced AviUtl Settings

-Translating AviUtl FX

-Common Troubleshooting for AviUtl

-Free Resource list
> AviUtl
> MMD
> Image material
> Audio material

Expect to sell at around USD$30

[tyee]

Is that available now?

[MaverickTse]

Is that available now?

Nope.

It is being edited by the publisher and me.
AND
it is written in Traditional Chinese.

If you know any US/Europe publisher willing to publish a book like this, please refer them to me.
My bank account is nearly depleted for the publishing...and cannot afford further 100% self-funded publishing.

[poisondeathray]

For the motion tracking MK I or II - Is it possible to edit the trackpoints manually (e.g. some might be off, can user "guide" them back into place) ? ie. can you manipulate or access the trackpoints on a dope sheet or some timeline ? Or where is the tracking data written after analysis ? That feature would make the tracker much stronger and usable in more situations. Because is very rare to get a "perfect" track everytime on 1st try (even for commercial trackers), especially when there can be things like motion blur, compression artifacts, complex scenes, objects crossing over another, lens distortions, rolling shutter etc...

[racer-x]

Sure, the Motion Tracking Filter essentially creates a MidPoint at each frame. You can go to any MidPoint and correct as needed. Attached is the MK-II tracking.exo in .txt form along with a close-up of timeline where you can view midpoints (aka key-frames).

[poisondeathray]

Sure, the Motion Tracking Filter essentially creates a MidPoint at each frame. You can go to any MidPoint and correct as needed.

apologies , I'm just starting to play with it now and should have fiddled around with it more before asking

[MaverickTse]

Sure, the Motion Tracking Filter essentially creates a MidPoint at each frame. You can go to any MidPoint and correct as needed.

apologies , I'm just starting to play with it now and should have fiddled around with it more before asking

and If you find importing the EXO one-by-one troublesome, you can try the new EXO Stacker (on post #3 of this thread) to merge several EXOs and then import in one go.

BTW, I think it is better to make Method1 of MK-II as default instead of Method2...1 seems more stable...

Be warned that MK-II Method 4 is now severely broken, while Method 3 can only track really easy targets.

I am now attempting to tweak M3 with some success, but I give up on M4.

[racer-x]

MaverickTse, is there a way to turn off the motion tracers? You can see them in the above image, they show as white squares. They get in the way if you want to edit the midpoints and it gets hard to see object. I looked and tried everything I could find, but no success so far.

[MaverickTse]

MaverickTse, is there a way to turn off the motion tracers? You can see them in the above image, they show as white squares. They get in the way if you want to edit the midpoints and it gets hard to see object. I looked and tried everything I could find, but no success so far.

Just close the Object Setting Dialog/Double click on object bar on the timeline, then move the timeline slider.

BTW, you can just type @MT instead of my full call-sign.

[MaverickTse]

Updates to MK-II:
1> Bugfixes:
--fix corrupted image when using with Resize plugin
--fixed drifted selection box after running analysis once (now you can just change the Method slider and click Analyze to re-run analysis with identical selection box)

2> Added "HSV Cvt"
-- Convert RGB data to HSV representation, then display the HSV image as RGB or grey-scale for individual channel


3> Added "Background Subtraction"
-- Using MOG2 or kNN method, attempts to isolate the moving object from the background.
-- can either output the mask alone, or the masked image


Original Sample:


(Note: The single AUF contains ALL 3 plugins!)

[MaverickTse]

A practical use of background remover: when the background interfere with tracking.

With either MK-I or MK-II, the blocks are likely lost when crossing the sky-ground boundary


Apply the kNN background subtraction.
Track this block-only image with your method-of-choice:
(I tracked the bottom block with MK-II mode 3, while the left one with MK-II mode 1)


Use EXO Stacker to merge the two EXO, and load into Adv.Editing for final composition.


I have attached the source video AVI so that you can try it yourself.

[racer-x]

Great job on the improvements to Motion Tracker MK-II. I noticed that mode-3 will auto resize while tracking, That makes quick work of face blurring etc.....

Yes I know it's plain wrong to blur such a pretty face, but I had to test it on something. Motion Tracker MK-II > mode-3 + quick blur:

[MaverickTse]

Great job on the improvements to Motion Tracker MK-II. I noticed that mode-3 will auto resize while tracking, That makes quick work of face blurring etc.....

Yes I know it's plain wrong to blur such a pretty face, but I had to test it on something. Motion Tracker MK-II > mode-3 + quick blur:

Can you try the "Easy Privacy" option of MK-II on the same video and see how well/poor it works?

[MaverickTse]

Announcing mtDeSub2,
a rewrite of mtDeSub.

better UI, tighter mask, a bit faster.

http://mavericktse.mooo.com/wordpress/archives/1965

[racer-x]

Great job on the improvements to Motion Tracker MK-II. I noticed that mode-3 will auto resize while tracking, That makes quick work of face blurring etc.....

Yes I know it's plain wrong to blur such a pretty face, but I had to test it on something. Motion Tracker MK-II > mode-3 + quick blur:

Can you try the "Easy Privacy" option of MK-II on the same video and see how well/poor it works?

I did try it out at the same time. "Easy Privacy" was significantly slower. It did look a little cleaner on the face, however there were many strange errors on intermittent frames in various locations. Almost like a mask was appearing and disappearing all around the frame.

[MaverickTse]

Great job on the improvements to Motion Tracker MK-II. I noticed that mode-3 will auto resize while tracking, That makes quick work of face blurring etc.....

Yes I know it's plain wrong to blur such a pretty face, but I had to test it on something. Motion Tracker MK-II > mode-3 + quick blur:

Can you try the "Easy Privacy" option of MK-II on the same video and see how well/poor it works?

I did try it out at the same time. "Easy Privacy" was significantly slower. It did look a little cleaner on the face, however there were many strange errors on intermittent frames in various locations. Almost like a mask was appearing and disappearing all around the frame.

This surprise me... I suppose it would find one face only.
Can you upload the original clip?

[racer-x]

Attached is the source clip trimmed on I-frames, so no re-compression. Just for fun, I gave her a hat to shield her from the sun's rays.

[PhotoCat]

Created a new project in 1920x1080 but the project became 1280x720!

On top of the window, it says, 1920x1080_30fps_44100Hz.exedit (1280,720) [1/1]*...

I think it worked b4. Anything I am doing wrong? Tks!

[PhotoCat]

High resolution jpg file in slideshow. How to zoom to 1:1 in high quality?

I have imported a jpg file of size 5616x3744 using Maverick's auto shrink plugin. However,
I have discovered that when I zoomed the image in aviutl to 100%, the quality was significantly
worse than my image viewer (faststone) in 100% view (ie less detail, less sharp). Is there anyway to preserve
the original image quality in aviutl when zoomed in?

I am guessing the auto shrink plugin will just shrink down the image file into a smaller sized image that
aviutl can handle, and therefore zooming with aviutl of the same image will not bring back the original
image quality. Not sure if this is the reason...

Thinking about a workaround... Should I just crop the original image in the face area rather than shrink it... ?

Anything I am missing? Pls advise! Thanks!